Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Human-building integration: Investigation of human eye pupil sizes as a measure of visual sensation in the workstation environment
(USC Thesis Other)
Human-building integration: Investigation of human eye pupil sizes as a measure of visual sensation in the workstation environment
PDF
Download
Share
Open document
Flip pages
Copy asset link
Request this asset
Request accessible transcript
Transcript (if available)
Content
HUMAN-BUILDING INTEGRATION: INVESTIGATION OF HUMAN EYE
PUPIL SIZES AS A MEASURE OF VISUAL SENSATION IN THE
WORKSTATION ENVIRONMENT
By
Xiaoxin Lin
Presented to the
FACULTY OF THE
SCHOOL OF ARCHITECTURE
UNIVERSITY OF SOUTHERN CALIFORNIA
In partial fulfillment of the
Requirements of degree
MASTER OF BUILDING SCIENCE
August 2018
2
COMMITTEE
Joon-Ho Choi
Assistant Professor
USC School of Architecture
joonhoch@usc.edu
(213)740-4576
Marc Schiler
Professor
USC School of Architecture
marcs@usc.edu
(213)740-4591
Shrikanth (Shri) S. Narayanan
Professor
USC Viterbi School of Engineering
shri@sipi.usc.edu
(213)-740-6432
3
ABSTRACT
Lighting is the most crucial factor impacting an occupant’s visual comfort in a
building environment. However, most prevailing current lighting guidelines deriving from
empirical values are designed primarily for paper-based tasks, rather than computer-based.
In many cases, present guidelines do not meet the needs for a user’s new task types. Above
all, existing technical tools also have limited functionality to evaluate a user’s real-time
visual perception, which can be applied as an input to control a building lighting system.
This research estimated each individual participant's visual sensations by analyzing
pupil sizes and their change patterns under different lighting conditions since the human
body has the physiological regulation ability which naturally minimizes the adverse effects
of the surrounding environment.
This study adopted a series of human subject experiments which were performed in
an environmental chamber of USC. Based on a computer-based task which is most
commonly performed in current offices, various ranges of ambient lighting parameters,
including luminance (cd/m
2
), illuminance (lux), luminance contrast ratio, and Unified
Glare Rating (UGR), were generated and controlled while each subject’s pupil sizes were
recorded. The experimental result data were statistically analyzed to identify a relationship
between human visual sensations, lighting parameters, and also pupil sizes by ethnic origin
and myopia condition.
The research outcomes showed the potential use of pupil sizes for estimating an
individual’s visual sensation and confirmed the principle as an applicable technology to
integrate an environmental design and control system with the help of a real-time sensing
device, such as a wearable sensor.
KEYWORDS:
Visual sensation; Human-Building Integration; Bio-signal; Luminance ratios; Unified
glare rating; Indoor environment
4
TABLE OF CONTENTS
ABSTRACT ................................................................................................................................... 3
TABLE OF CONTENTS .............................................................................................................. 4
CHAPTER 1: INTRODUCTION ................................................................................................ 7
1.1 Problem ................................................................................................................................ 8
1.2 Previous study .................................................................................................................... 10
1.3 Physiological Response ...................................................................................................... 11
1.4 Pupil Size ............................................................................................................................ 11
1.5 Research Objective ............................................................................................................ 12
CHAPTER 2: RESEARCH BACKGROUND .......................................................................... 14
2.1 Productivity and Human Health associated with Lighting Environment ................... 14
2.2 Improving Design Methods and Environment of Lighting .......................................... 15
2.3 Pupil Sizes of Human Beings and Potential Application .............................................. 18
2.4 Conclusions for Background Research .......................................................................... 20
CHAPTER 3: METHODOLOGIES .......................................................................................... 21
3.1 Scope of Work .................................................................................................................... 21
3.1.1 Lighting Parameters ................................................................................................... 23
3.1.2 Task Types .................................................................................................................. 28
3.1.3 Questionnaire and Performance Test ....................................................................... 28
3.1.4 Pupil Size Parametric Data ........................................................................................ 31
3.2 Experimental Chamber Setup .......................................................................................... 33
3.2.1 Chamber Design ......................................................................................................... 33
3.2.2 Lighting Fixtures ........................................................................................................ 36
3.3 Research Tools and Sensor Devices ................................................................................. 36
3.3.1 Sensory Devices: Illuminance Meter, Luminance Meter and HDR Camera ........ 37
3.3.2 DAQ Device ................................................................................................................. 38
3.3.3 ASL Mobile Eye XG ................................................................................................... 39
3.4 Adopted Software .............................................................................................................. 40
3.4.1 LabVIEW .................................................................................................................... 40
3.4.2 Programming Logic .................................................................................................... 41
3.4.3 Photolux 2.1 ................................................................................................................. 42
3.4.4 Minitab ........................................................................................................................ 43
3.5 Experiment Rounds ........................................................................................................... 43
3.6 DATA Analysis .................................................................................................................. 45
5
CHAPTER 4: STUDY RESULTS ............................................................................................. 47
4.1 Demographic Information of Test Subjects .................................................................... 47
4.2 First Round Experiment ................................................................................................... 48
4.2.1 Illuminance Level per Visual Sensation by Individual ........................................... 48
4.2.2 Pupil Size Between Luminance Contrast Ratio Group by Individual ................... 49
4.2.3 Visual Sensation per UGR Group by Individual ..................................................... 53
4.3 Second Round Experiment ............................................................................................... 57
4.3.1 Illuminance Level per Visual Sensation by Individual ........................................... 57
4.3.2 Pupil Size Between Luminance Contrast Ratio Group by Individual ................... 58
4.3.3 Visual Sensation per UGR Group by Individual ..................................................... 62
CHAPTER 5: DATA ANALYSIS AND DISCUSSION ........................................................... 67
5.1 Pupil Sizes per Illuminance Level by Individual ............................................................ 67
5.2 Illuminance Range per Visual Sensation by Individual ................................................. 73
5.2.1. First Round Experiment ........................................................................................... 73
5.2.2 Second Round Experiment ........................................................................................ 75
5.2.3 Discussion Differences Between the Two-Round Experiment ................................ 76
5.3 Raw Data of Pupil Size Per Sensation by Individual ..................................................... 77
5.4 Data Processing of Raw Pupil Size ................................................................................... 82
5.4.1. Pupil Size Standardization ........................................................................................ 82
5.4.2 Moving Average Filtering .......................................................................................... 83
5.4.3 Gradient of Pupil Size ................................................................................................ 86
5.5 Processed Pupil Size per Visual Sensation by Individual .............................................. 87
5.5.1 Comparisons of Standardized Pupil Size per Visual Sensation Between Different
Contrast Ratio Group by Individual ................................................................................. 87
5.5.2 Comparisons of Gradient Pupil Size per Visual Sensation Between Different
Contrast Ratio group by Individual .................................................................................. 96
5.6 Comparisons of Pupil Sizes Between Different Subject Groups ................................. 101
5.6.1 Iris Color ................................................................................................................... 101
5.6.2 Myopic Condition ..................................................................................................... 103
CHAPTER 6: CONCLUSIONS ............................................................................................... 106
6.1 Illuminance Parameters, Visual Perception and Pupil Size ........................................ 107
6.2 Pupil Size and Physiological Characteristics ................................................................ 108
6.3 Applications for Physical Models and Experimental Data .......................................... 109
CHAPTER 7 FUTURE WORK ............................................................................................... 111
6
7.1 Possible Improvements In Terms Of Participants ....................................................... 111
7.2 Possible Improvements In Terms Of Software And Hardware .................................. 111
7.3 Development of Strategy ................................................................................................. 112
Bibliography............................................................................................................................... 114
7
CHAPTER 1: INTRODUCTION
Buildings are the main consumers of energy worldwide, responsible for roughly 40% of
the world’s primary energy consumption. More detailed data shows that building energy
consumption in the United States, the United Kingdom and the European Union accounts
for about 41%, 39% and 37% of the energy consumption respectively. (EIA 2015; IEA
2014)
Similarly, other studies have shown that the contribution of global construction to energy
consumption, including residential and commercial categories, has grown steadily in
developed countries by 20 to 40%, surpassing other major sectors, such as industry and
transport. (International Energy Agency, 2006) Therefore, building energy efficiency is the
primary goal of energy policy development at the regional, national and international levels
in today’s world. The increase in energy use in lighting systems is particularly pronounced
in construction services. In the United States, for example, the lighting system consumes
35% of the building's energy consumption and 10% of the total consumption. (U.S. Energy
Information Administration, 2012)
More importantly, people in America spend about 9/10 of their time inside a constructed
environment in contemporary society. Hence, the quality conditions of an indoor
environment are important. The work productivity and environmental health of occupants
in a building are influenced by and sensitive to perceived sensations on environment and
ambient conditions, such as conditions of acoustic comfort, air quality, lighting quality and
thermal setting. Among those factors, the one that exerts the most significant impact on the
visual comfort and sensation of an occupant is lighting (visual) quality, since it is
susceptible to and affected by the instantaneous impact of quality of lighting on users.
Therefore, as mentioned above, lighting is one of the most important considerations for
indoor environmental quality and is almost one of the most significant components of an
efficient and effective office environment. Lighting and building energy consumption are
closely related and have a significant impact on the health and productivity of occupants
as well.
8
1.1 Problem
At first, lighting just serves to illuminate the surrounding area. With the development of
technology, people gradually realize that lighting has a huge impact on an individual’s
visual health. Today, the requirements for lighting are not only aimed to illuminate the
room, but also to minimize impact on the circadian system of the body and increase
productivity and provide visual acuity when needed. Some specialized lighting systems
can also increase alertness and promote sleep. For example, the concept of glare is now
well known and under its influence, people may have various negative reactions, such as
nausea, discomfort, or temporary vision loss. However, according to some researchers, the
percentage of occupants in buildings who are exposed to improper lighting conditions
amounts to 65%.
Those defective conditions, consequently, could lead to visual stress and glare problems in
the environment of a workplace and potentially result in physical symptoms (like
headaches and eye fatigue) and lower work efficiency. It is estimated that the annual loss
resulting from defective environmental conditions reaches up to USD 2,700 in work
productivity/health of every user in a workplace or office. Hence, proper lighting
conditions are of great importance to ensure the healthy state of occupants in a building
and protect them from disease or eye related injuries.
Current lighting specifications and guidelines provide illumination recommendations for
different room types, which are derived from the general lighting requirements for typical
activities occurring in each room. These standards were developed by the IESNA technical
team of Illuminating Engineering Society to ensure good visual acuity in a variety of tasks
to avoid eye fatigue and minimize loss of productivity and headaches.
The problem is that the present industry guidelines, such as from the Illuminating
Engineering Society of North America (IESNA), have been determined by empirical and
experimental methods in monitored laboratory tests instead of actual building occupants’
needs; also, those recommended guidelines are mostly for tasks that are paper-based
instead of computer-based.
Despite the inadequacies of the existing lighting design standards, most of the office
9
buildings in North America were designed following guidelines based on practical results
developed by the-Illuminating Engineering Society of North America (IESNA). Therefore,
in the actual constructed environment, even if the parameters of lighting meet the design
standards, it is not uncommon to hear the residents complain about the uncomfortable
lighting. As a result, dizziness, low efficiency and other negative effects cannot be avoided.
Another key issue that cannot be overlooked is that diagnostic tools applied in the industry
for the quality of the static computational lighting, like the image of lighting contour (isolux
plots), and the program of computational simulation, has been hindered by the phenomenon
that floor layouts change frequently in the office. More importantly, no functional
characteristics of those existing technical tools can evaluate the real-time visual sensation
of a user in a workplace which could be applied to building controls for lighting systems.
Over the past three decades, in order to enhance the effect of architectural lighting, much
efficient simulation software has emerged to help designers and architects analyze and
understand the complex lighting conditions in design projects. Currently, the mainstream
daylight simulation software includes Ecotect, Radiance, Daysim and 3D Max.
As a highly visual architectural design and analysis tool, Ecotect can achieve a wide range
of performance analysis of a comprehensive 3d model. Radiance is a popular and advanced
lighting simulation and environment rendering software. With strong calculation capability,
it provides designers spectral irradiance and spectral radiance values for both interior and
exterior spaces considering daylight and inter-reflected artificial lighting. It is often used
by designers and architects to predict lighting conditions, visual quality, and visual effects
in design spaces. Daysim provides the ability to simultaneously execute performance to
control daylight. At the same time, 3ds Max's powerful rendering capabilities will create
excellent daylight visualization, thus explaining the lighting conditions more directly.
In addition, High Dynamic Range (HDR) imaging is also a prevalent tool in lighting design
field. The HDR achieves a higher dynamic range of brightness than usual digital images
and more accurately represents images by combining multiple low dynamic range (LDR)
or standard dynamic range (SDR) photos. HDR is better because it contains a lot of
10
information. However, collecting enough images takes too much time and effort, which is
not economical for wide use.
Just as mentioned before, the biggest drawback of these simulation tools is that none of
these tools can estimate the occupants’ visual perception by obtaining real-time data.
Therefore, it is necessary to find a new way to help lighting design and better lighting
control.
1.2 Previous Study
Many researchers have proposed various levels of lighting intensity preferred by
individuals, which were determined by factors like age and gender. One of the author’s
previous research results also corresponded with the findings of existing literature with
surveyed and measured data. Individuals in those instances were placed in similar
conditions of lighting, yet the visual sensations reported by them were different.
Furthermore, they also requested different levels of lighting, despite being engaged in the
same work within the office building.
Some research projects, based on the imperative to utilize the eyes’ natural physiological
conditions, have been carried out to identify the pupil’s dilation and contraction to various
visual objects, such as luminance, reflections and surface color of the wall. By taking the
“accuracy of reading” as the acuity measure, Berman et al. investigated how pupil size
affected letter size, or the acuity function and in order to explore the spectral response of
pupils to the production of different luminance conditions in a limited scope, they also
established different lighting spectrums. Mojtaba implemented a test on visual acuity under
different conditions of color temperature while supervising the research subjects’ pupil
sizes (Mojtaba Navvab, 2002). The relations between pupil size and visual acuity, in
particular, have been explored in the ophthalmology field for decades. Some research
projects conducted recently concentrate on the pupil size; yet most of them were related to
pupil size influenced by relevant ophthalmological treatments. Similarly, most current
research topics concerning pupil size have paid excessive attention to physiological
reactions to fatigue, visual performance and visual stimuli as a result of pupil dilation and
11
constriction, yet attached little importance to the response of the pupil to visual comfort (or
sensations), in spite of its potential to be adopted as an index of visual sensation.
1.3 Physiological Response
The human body is like a sophisticated instrument that has the physiological response to
help itself reduce the negative impact of the external environment, such as extreme
temperature conditions or excessively bright or dark ambient light, on the body.
For example, depending on the strength of various stressors, the human body may sweat
or control the skin surface temperature to balance heat loss or gain due to ambient thermal
conditions. Another example is the change in pupil size. In most cases, pupils constrict and
dilate to respond to light variations. The size of the pupil is controlled by sphincter muscles
and papillary dilator which are opposed to each other. Those muscles are severely
innervated by the parasympathetic system and the sympathetic system as a part of push/pull
in the autonomic nervous system. When the surrounding environment is too dark, the
sphincter muscles enlarge the pupil in order to facilitate more light into the eye, so as to
help us see the objects, and vice versa.
Due to such a natural physiological system, the physiological condition of humans could
be kept stable, as long as changes in ambient environment could be minimized. Therefore,
based on the principle of physiological response, this study uses pupil size as a
physiological feedback signal to predict visual sensory status, and subjective illumination
sensation can be used as a function of objective physiological signals. Since the
experimental results are derived from a true human physiological response, this will be a
new method of visual quality assessment, which eliminates the drawbacks of traditional
methods, such as lighting simulation and high dynamic images, which rely mainly on the
presumptive human environmental response that can lead to bias and irrationality.
1.4 Pupil Size
The pupil is an aperture in the center of the iris of the eye through which light directly hits
the retina. The pupil looks black because the light entering the pupil is absorbed after
12
diffused reflection in the eye, and the rest of the light is absorbed directly by the tissues in
the eye, most of which does not leave the pupil.
Pupils constrict in the light, dilate in the dark, and the normal range is 2-4 mm. There are
two kinds of tiny muscles in the iris- one is called the pupil sphincter, located around the
pupil and less than 1 mm wide and the other is called the dilated pupil, which is arranged
radially in the iris. The pupil sphincter is in charge of pupil contraction, which is dominated
by the parasympathetic nerve in the eyes. The dilated pupil is responsible for the
enlargement of the pupil and is dominated by the sympathetic nerves. The two muscles
coordinate, restrict each other, and shrink to adapt to a variety of different environments.
The size of the pupil controls the amount of light entering the eye. The range of pupil size
can be quite large, with the diameter of the pupil capable of being less than 1 mm when the
pupil contracts significantly and can be greater than 9 mm with extreme expansion. The
pupils of adults are generally shaped as a perfect circle with a diameter between 2-4mm
and the difference between the sides not exceeding 0.25mm. The size of the pupil constricts
and dilates with the intensity of light, but also with age, refraction, physiological status and
other factors. In general, older people have smaller pupils, while young children to adults
have larger pupils, especially during adolescence. The pupil of myopia is larger than that
of presbyopia. Pupils expand when people are stressed and excited and narrow when people
take deep breaths, do mental work, and sleep. In addition, pupils will dilate or constrict
when certain drugs are used or certain diseases happen. Therefore, the change of the pupil
diameter is of great clinical significance.
More discussion about the potential use of pupil size for the lighting control and improving
health will be opened up in the background research chapter.
1.5 Research Objective
As mentioned previously, visual quality, as an IEQ factor, plays a vital role in the field of
a building environment, particularly in the environment of a workplace where the working
productivity and health of occupants are crucial. Hence, this research explored the potential
use of pupil size to evaluate the visual sensation of users in response to the lighting
13
environment in the workplace, which could be considered as one principle of interaction
between human and building.
14
CHAPTER 2: RESEARCH BACKGROUND
This chapter aims to explore previous research as well as other sources related to the
discussion on lighting parameters while investigating the applications of pupil size to the
field of lighting design. By exploring potential advantages and issues relevant to pupil
size together with other crucial factors, the work scope of this study could be better
identified. In order to further understand the function and lighting parameters of the pupil
size and its potential application in lighting control and design, background information
regarding the major areas of the field is also studied.
2.1 Productivity and Human Health associated with Lighting Environment
Numerous studies have been carried out to explore the correlation between various
lighting conditions and human productivity and health. Research conducted by Hedge,
Sims Jr., and Becker (1990) from Cornell University demonstrated findings concerning
the relationship between employees working with computers- their visual health,
satisfaction, and productivity- and lighting conditions in the working environment used
in the background research. Based on information offered by The American Society of
Interior Designers, it can be found that 68% of office workers complained about lighting.
Moreover, research conducted by Silicon Valley indicates that 79% of users for Video
Display Terminal (VDT) demanded better lighting conditions. (Gauri Shankar
Shrestha,2011) Furthermore, through a study, Louis Harris (1989) found that eyestrain
surpassed asbestos and radiation and ranked first as an office health hazard. It is difficult
to ignore all abovementioned dissatisfactions, as they imply potential issues concerning
lighting. In order to ensure visual comforts of employees, it is imperative to build a better
lighting environment in offices.
Besides the dissatisfactions about lighting aforementioned in the surveys, some other
research also points out that productivity could be influenced by lighting. According to
Dilouie (2003), it is known that people can be affected by lighting both physiologically
and psychologically. In the whole process to learn about the outside world, visual
impression accounts for 80%-85%. Nevertheless, it is the perception which relies on
lighting that makes visual purpose possible. As people stay indoors for most of the day,
15
lighting could control their predominant perception of the world. For a long time, it has
been widely acknowledged that some methods of lighting design were better, as they
could improve the satisfaction of employees. In contemporary society, assessing the
productivity of workers in offices becomes more difficult; as a result, satisfaction is more
than just a metric as before. With regards to the crucial role of satisfaction in evaluating
the productivity and performance of participants in this experiment, this project applied
a satisfaction investigation to all participants.
2.2 Improving Design Methods and Environment of Lighting
In order to seek for more appropriate lighting methods for office environments, many
studies have been executed, such as making changes in workplace layout, illuminance,
color temperature, type of lamp, as well as other factors that could potentially influence
lighting control.
As mentioned previously, Hedge Sims Jr., and Becker (1990) from Cornell University
designed a study to determine whether any difference could be created in health,
satisfaction, productivity or visual comfort of computer workers by applying the
parabolic downlighting or the indirect up-lighting system. Compared with the group of
indirect up-lighting, the group of parabolic downlighting complained twice more
frequently about concentration problems and tired eyes. Additionally, the group of
parabolic downlighting reported to be more bothersome than the group of indirect up-
lighting and the productivity of workers was negatively influenced by the problem of
visual discomfort in the condition of parabolic lighting.
Craig Dilouie (2003) has conducted research to explore how various forms of practical
office lighting have an impact on the employees’ health and performance in offices.
There are many variables in the research, including brightness of the room surface,
personal control for lighting, etc. It was reported in this research that parabolic and lensed
troffers were less comfortable than indirect/direct fixtures. In addition, it was found that
the satisfaction, performance of productivity and attention, as well as the lighting quality
perceived by employees were greatly improved if a dimmer could be used by them.
16
Furthermore, if people became more content with the quality of lighting, they could form
a more positive attitude towards the working environment and job. Through enhancing
visibility in task conditions and lighting, the ability of performing tasks can be
improved.(Dilouie, 2003)
To sum up, Dilouie found in his study that people who work with brighter room surfaces
and could access a dimming control to freely set a preferred level of lighting by
themselves would be more satisfied. In the process of working, satisfaction of lighting
level could significantly improve productivity, positive attitude and focused attention.
Besides, based on the results, it could also be inferred that more than ¼ of employees
were dissatisfied with the standard lighting and lighting can bring about the most
comfort if indirect/direct lighting was combined with perimeter wall washing and a
private dimmer, which can in turn enhance motivation and satisfaction among most
employees to the greatest degree. For company owners, this is an optimal choice to
realize productivity and health as well as economic profits.
In addition to available personal control and lighting type, other kinds of research
concerning lighting parameters also attracted enormous attention in the field of lighting
studies. With regards to the research, color temperature and color spectrum have been
attached with the most importance, since it is thought that productivity has the closest
relationship with the visual comforts of humans.
According to a report offered by PP&L (Pennsylvania Power and Lighting Company),
the mistake rate made by the Department of N3 Drafting was high, since employees did
not make every effort due to insufficient cool-white fluorescent lighting. There is also
another finding, which can be described as follows. As the overhead fixtures were
created light that bounced off the task’s surface into the eyes of employees, a type of
indirect glare, or veiling reflection in another name, was thus created. By applying a
change, some newly chosen lighting with full spectrum could be more effective
compared with that in previous systems, because it lasts longer and consumes less
electricity. Compared with the improvements in productivity after modernization, these
costs were remarkably reduced. With the decrease of veiling reflections through the
17
application of new lighting, the working productivity of employees was increased by
13%. It was estimated that the benefits brought by improved productivity was USD
235,290 annually. More importantly, many errors were also avoided. Furthermore, after
the installment of a new lighting system, it seemed that the rate of absenteeism was
lowered as well, together with the decrease of headache rates and eyestrain. In PP&L,
baseline benefits could be reduced by USD 255,929 annually with the adoption of better
lighting (Deneen, 2004). Hence, the application of lighting with full spectrum and
indirect fixtures could be viewed as a means to improve employee productivity and
visual comforts.
Dr. Same Berman applied the theory that energy efficiency and visual accuracy can be
improved by increasing the blue light’s scotopic side. Besides, Berman (2000)
mentioned the results of one lighting experiment in his paper—in order to test how
brightness and vision were affected by new findings about rod sensitivity, one
demonstration was therefore set up. Traditional fluorescent lamps were compared to
lamps with high color temperatures (scotopically enhanced, higher degree of bluish
output) and lamps with low color temperatures (scotopically deficient, lower degree of
bluish output) in terms of vision effects. The facilities staff of Intel observed that the
visual ability of people can be improved in scotopically strong lighting. Meanwhile,
people could perceive such lighting to be brighter in spite of the opposite result
demonstrated by light meter and this is the result that PGE attempted to indicate. Intel
also recognized that energy consumption can be greatly reduced through decreasing the
amount of lamps while maintaining or improving brightness conditions and prior vision
after the application of scotopically strong lamps (Berman, 2000).
As mentioned by Doris Rapp (1997), a famous pediatrician of MD in her paper of “Is
This Your Child’s World”, natural lighting is definitely the optimal lighting for all
places in the world. One place of upmost importance is schools. Nevertheless, students
in many places spend at minimum six hours each day in cool-white color fluorescent
light. Due to X-rays and radiation emitted by fluorescent lights, students’ learning
efficiency can be negatively influenced; what is worse, some health problems, like
depression, fatigue and eyestrain among students may sometimes be caused as well.
18
According to a research study, such health problems can be prominently reduced by as
high as 33% if fluorescent lights were replaced by full-spectrum lighting. As a result,
lighting with full spectrum has been advocated widely in research aspects and daily
activities (Rapp, 1997). Dr. Rapp states that it is imperative to choose full spectrum
lighting when there is a need to use fluorescent lighting. Companies may find it possible
to improve their office lighting environment on the basis of the latest invention for
lighting commodities with blue spectrum.
2.3 Pupil Sizes of Human beings and Potential Application
During the last few years, researchers have realized how pupil size plays a crucial role
in vision, and how important it is to design proper interior lighting for maintaining visual
acuity. Spectrum controls have been widely explored in many studies. Scotopically
strengthened lamps that favor wavelength of blue light could decrease pupil size. As
mentioned by Berman (2000), at the level of distinct interior light, better vision can be
achieved through smaller pupils. In the present practice of lighting, it is frequently
advocated that pupil size could be reduced through increasing light levels; however, this
means is neither effective nor takes advantage of the rods’ response to controlling pupil
size.
A program jointly conducted by University of California in San Francisco and Lawrence
Berkeley Laboratory investigated the responses of humans to electric lighting. If
subjects were put in lighting of indirect HPS (high-pressure sodium), a more remarkable
difference in pupil size could be detected compared with that of indirect incandescent
lighting when light intensity was matched photopically. Both of the lighting systems
were roughly applied to an identical distribution of spatial luminance. Based on the
observation for differences in pupil sizes, it can be claimed that the distribution of
spectral power in both lighting systems could influence visual performance as well as
other aspects in the function of visual systems.
It was recognized that the impact of pupil size on the capability of the visual system to
realize higher resolution for details, or depth of field and visual acuity called by
Leibowitz (1952) and sensitivity function of spatial contrast identified by B.Y.F.W.
19
Campbell and Green (1965), was considerable. F.W. Campbell (1957) observed that the
depth of field inversely decreased with the increase in pupil diameter. Moreover, in an
environment with stable ambient luminance, retinal luminance can be enhanced by
larger pupils (Ferguson and Stevens, 1956; Luckiesh and Moss, 1934). Hence, by
controlling pupil sizes independently from light conditions that are determined by
specific factors concerning visual tasks, visual acuity can be improved (Eastman and
McNelis, 1963). With scotopically deficient lamps, pupils will become larger and it was
believed that the performance of employees in such a case would be better than that in
an environment with scotopically strong lamps.
In addition, according to studies on contrast sensitivity (B.Y.F.W. Campbell and
Gubisch 1966; B.Y.F.W. Campbell and Green 1965), the quantity of contrast sensitivity
reduced steadily with the increase of pupil size. Further research studies also indicated
that some pupil sizes were preferred in the daily visual tasks of the world and the
outcomes of that research should result in a novel dimension to improve the lighting
environment quality.
Nevertheless, in view of controversies concerning vision science and the crucial role of
pupillary responses to lighting design and vision, it is necessary to identify the effects
of other factors in lighting and people on pupils by conducting further tests on other
types of lighting. With additional available information, general principles that
determine visual efficiency could be further ensured.
Some other research also investigated the physiological and environmental factors that
could influence pupil size. It was found that pupil sizes linearly decreased at all levels
of illuminance as a function of the age. Meanwhile, pupil size could be more
significantly affected by age with the highest level of illuminance. The pupil diameter’s
annual change rate resulting from age was 0.015mm at a high level of illuminance and
0.043 mm at a low level of illuminance, respectively. Moreover, the variability in pupil
sizes of people of the same age reduced by approximately the factor of two with the
increase of luminance over the investigated range. It was found by Winn et al. (1995)
that pupil size was irrelevant with iris color, refractive error and gender (P >0.1).
20
An effective proof is still lacking despite that some previous research demonstrated the
irrelevance between pupil size and gender. Other factors of humans, like ethnic origin,
myopic conditions and age, need to be tested again and further in-depth in the hope of
identifying how they potentially affect pupil size.
2.4 Conclusions for Background Research
At present, studies concerning this topic are far from enough. Nonetheless, those studies
have already illustrated that appropriate lighting conditions play a significant role in
office environment, as lighting conditions could influence the working efficiency, health
and visual comfort of employees significantly. In future research on lighting, the
potential application of pupil size can be explored. In previous studies, some research
investigated physiological and environmental factors that affected the pupil size and
some research on lighting also took pupil size as a research object. However, that
research mostly concentrated on the improvement of visual comfort through controlling
the spectrum. There is no existing research that focuses on the correlation among pupil
size and luminance/illuminance while applying to controlling and designing lighting,
which is a crucial factor in the contemporary building environment. As people at present
are exposed to lighting for more than 9/10 of their daily time, lighting exerts a prominent
effect on their working efficiency and visual health.
21
CHAPTER 3: METHODOLOGIES
This chapter describes the research methods in detail. First, to give a whole idea of the
experimental procedure, the main work flow of the experiment and the selected
variables will be explained. Secondly, the laboratory settings are described in detail to
help deliver a clear understanding of the experimental environment. Third, the research
apparatus and software used in the study will also be given a detailed description and
index. Finally, other preparatory work will be addressed, including participation in IRB
courses and preliminary studies introductions.
3.1 Scope of Work
Figure 3.1 Diagram of the research methodologies
The experiment will be carried out in the environmental chamber of Watt Hall at the
University of Southern California (USC) University Park campus. The chamber is on
the basement floor, so the experiment results will not be affected by daylight. The size
of the entire chamber is 112” x 120” x 95”, which is very close to the size of an ordinary
workplace and is the intended simulation of the experimental environment. Throughout
the experiment, the environmental parameters such as temperature and humidity
remained unchanged.
22
The workstation desk has been put in the laboratory along the wall on the south and
twenty-eight dimmable LED lamps were used as the light sources after being installed
over the desk. The light intensity generated by those lamps can be controlled manually
through remote control, so as to produce various lighting intensities above the desktop.
In addition, there were two chairs put beside the table, which were used by experiment
investigator and subjects, respectively.
The laboratory’s interior was refurbished in accordance with the requirements of the
experiment. The dimmable LED lights were changed according to demands for different
illuminance. A TV with changeable brightness was adopted to imitate the contiguous
screen behind the screen of the target computer, with the purpose to produce contrast
ratio. In addition, a mouse, a keyboard and a monitor were put at the center of this table,
for the sake of completing computer-related tasks. The maximum luminance of the
background screen is 318 cd/m2. It occupied 60% of the field of view based on the
comparison between the size of the TV and the wall in the subject’s view.
The environmental chamber has good heat insulation and noise absorption capabilities.
At the same time, through normal operation of the ceiling ventilation fan, constant
thermal conditions and good indoor air quality can be ensured. The ambient thermal,
acoustic and air quality conditions were carefully controlled throughout the experiment.
The air temperature, relative humidity, and CO2 during the experiment ranged from 24.5
± 0.5℃, 32 ± 2.5%, and 610 ± 35 ppm, respectively.
Human participants were required to partake in this experiment. As stipulated by
laboratory conditions, each experiment could only involve one participant. There were
no particular selection approaches, and participants were selected randomly.
Nevertheless, a balanced proportion of demographic features such as eye color, myopia,
age and gender were taken into consideration while selecting candidates. Participants
23
were told about the whole experimental process and major tasks in the experiment, as
well as limitations on drinking and eating in the experiment process. Every experiment
took about one and a half hours. Participants were required to perform specific
computer-based office work during the whole experiment process. At the constant ratio
of luminance contrast, the range of illumination was varied between 50 lux and 1400
lux every fourth minute with the pace of 150 lux. In regard to every level of illumination,
participants should answer a group of questions concerning their comfort conditions and
visual perception through a questionnaire. In the meantime, data about illumination
parameters and pupil size were stored and collected automatically by a computer.
Section 3.3 and Section 3.4 illustrate relevant details. Based on responses of participants
to comfort conditions and visual perception in the questionnaire, data of lighting system
and real-time size of the pupil were collected in the experiment. A model of visual
comfort was thus developed through statistical tools under the assistance of the real-
time sensing device, which was expected to be used in building integration for human
beings.
3.1.1 Lighting Parameters
1. Illuminance Level
In this chamber, 30 units of 730 lumen 9W dimmable LED lamps were installed on the
surface of the ceiling. As LED lamps do not produce any thermal radiation, they would
not influence the thermal environment in this laboratory. Color temperature in this
experiment was set at about 5,000K, which was quite common in the environment of
daily work. The lighting intensity generated on the surface of the workstation ranged
between 50 lux to 1400 lux. The analog regulator which was provided with a 150-lux
illumination interval was used to adjust illumination and this was the smallest
perceptible change in illumination displayed in the previous research (J.-H. Choi, Zhu,
and Johnson 2013).
24
In the experiment, overall luminance would also be taken into account and measured
through Photolux 2.1. The correlation index was estimated to be 0.99, and the p-value
of the correlation index was 0.000. Due to the prominent linear relation between
illuminance and overall luminance, all data of luminance were provided according to
the measured illuminance in the experiment.
Figure 3.2 The correlations between illuminance and average luminance of the
experimental settings
2．Contrast Ratio and UGR Value
The contrast ratio for the visual comfort in the experiment denotes the contrast between
luminance in the task area and the glare source—the TV screen right behind the screen
of the task computer.
25
Figure 3.3 The field of subjects’ view
The UGR value and contrast ratios used in the experiment were identified through
measuring the luminance of the TV, the computer screen as well as surrounding surfaces
(such as walls) by virtue of a camera with fish-eye lens (the model was Nikon Coolpix
8400). Photolux software was employed to calculate the computer screen’s luminance,
which was set to be constant at 75 cd/m
2
. As concluded in a wide range of research, the
desirable luminance ratio for the visual comfort between task and adjacent surroundings
is 3:1 (Osterhaus, 2002). UGR values were generally defined in the 28, 25, 22, 19, 16
and 13 step. As to the limits of UGR for lighting products in different environments, the
maximum value for the technical drawing as well as for meeting, training, writing,
reading and computer-related work should be no more than 16 and 19, respectively (Liht
2016). The maximum luminance of the background screen is 318 cd/m2. It occupied
60% of the field of view based on the comparison between the size of the TV and the
wall in the subject’s view. The rationality degree of recommendations should be
validated by experimental setting. On the basis of this principal and the consideration
for the practical situation of this laboratory, the outcome of exercisable contrast ratios
26
were picked to be 6.4/1, 1/1 and 1/4.25; while UGR values were examined and
demonstrated in the following Table 3.1.
Table 3.1 Realizable contrast ratio, related parameters and TV settings
NO. Contrast ratio Laptop luminance (cd/m
2
) TV luminance (cd/m
2
)
TV
Darkness
1 1/4.25 75 318 12%
2 1/1 75 73.2 60%
3 6.4/1 75 11.6 5%
Figure 3.4 TV and computer screen luminance false color
(contrast ratio of computer screen to TV: a. 1/4.25; b. 1/1; c. 6.4/1)
Table 3.2 Calculated Unified Glare Rating (UGR) value of first round experiments
First Round Experiment
No. Contrast ratio Illuminance (lux) Unified Glare Rating (UGR)
1 1/4.25 50 18.3
2 1/4.25 200 17.8
3 1/4.25 350 17.1
4 1/4.25 500 18.0
5 1/4.25 650 18.2
6 1/4.25 800 17.8
7 1/4.25 950 17.5
8 1/4.25 1100 17.2
9 1/4.25 1250 16.7
10 1/4.25 1400 17.0
Second Round Experiment
1 1/1 50 11.6
2 1/1 200 13.3
3 1/1 350 13.2
4 1/1 500 12.7
5 1/1 650 12.4
6 1/1 800 12.1
7 1/1 950 12.1
27
8 1/1 1100 11.2
9 1/1 1250 11.2
10 1/1 1400 11.3
Third Round Experiment
1 6.4/1 50 9.7
2 6.4/1 200 15.5
3 6.4/1 350 16.4
4 6.4/1 500 16.3
5 6.4/1 650 15.5
6 6.4/1 800 15.0
7 6.4/1 950 14.8
8 6.4/1 1100 15.0
9 6.4/1 1250 14.6
10 6.4/1 1400 14.6
Table 3.3 Calculated Unified Glare Rating (UGR) value of second round experiments
First Round Experiment
No. Contrast ratio Illuminance (lux) Unified Glare Rating (UGR)
1 1/4.25 1400 17.8
2 1/4.25 1250 17.5
3 1/4.25 1100 17.2
4 1/4.25 950 16.7
5 1/4.25 800 17.0
6 1/4.25 650 18.3
7 1/4.25 500 17.8
8 1/4.25 350 17.1
9 1/4.25 200 18.0
10 1/4.25 50 18.2
Second Round Experiment
1 1/1 1400 12.1
2 1/1 1250 12.1
3 1/1 1100 11.2
4 1/1 950 11.2
5 1/1 800 11.3
6 1/1 650 11.6
7 1/1 500 13.3
8 1/1 350 13.2
9 1/1 200 12.7
10 1/1 50 12.4
Third Round Experiment
28
1 6.4/1 1400 15.5
2 6.4/1 1250 15.0
3 6.4/1 1100 14.8
4 6.4/1 950 15.0
5 6.4/1 800 14.6
6 6.4/1 650 14.6
7 6.4/1 500 15.5
8 6.4/1 350 16.4
9 6.4/1 200 16.3
10 6.4/1 50 9.4
3.1.2 Task Types
In today's society, a lot of technical work is mainly accomplished on a computer. In
other words, computer-based tasks are the mainstream type of work. Therefore, this
study used computer-based task types to test subjects' visual perception and comfort
under various lighting conditions. Simplified performance tests were conducted under
simulated typical office lighting conditions to assess participants' visual acuity and their
productivity at each lighting condition.
The computer-based tasks of this experiment included reading and typing. The same
reading material was provided to each participant in the form of a Word file. At the same
time, the subject was asked to type the same content in another blank text window. These
two text files were displayed on the monitor with the same font size and background
color. Such work was assigned to each subject in order to imitate eye activities in the
office environment. Each participant was asked to complete the performance test at three
different illumination contrast ratios of 1 / 4.25, 1 / 1, and 6.4 / 1, respectively.
3.1.3 Questionnaire and Performance Test
In this experiment, the feedback of experiment participants about their comfort
conditions, visual perceptions and certain physiological data concerning these subjects,
such as eye condition, gender and age, would be recorded by requiring them to complete
29
one simple questionnaire. In previous research studies that involved the participation of
human subjects, ethnicity, age and gender were often considered as the most crucial
parameters, particularly in studies focusing on thermal comfort and visual comfort
(Sivaji et al., 2013). The other two parameters that were thought to be much more closely
relevant with light studies are myopia condition and iris color. It is thought that all those
parameters could influence a pupil’s physiological response (Sivaji et al., 2013).
Subsequently, whether the features of one subject and visual comfort as well as the
subjective lighting condition were significantly correlated with each other was checked
after relevant data had been recorded and analyzed.
The seven-point scale would be adopted to evaluate every question in the questionnaire.
According to the previous background research (Sauro, 2014), it can be found that the
seven-point scale could offer answers with higher resolution but less complexity in
contrast to the five-point scale. Evidence shows that participants tend to choose the
option adjoining to the prior one for the newer level with little judgment or thought if
an eleven-point scale or even a scale with more points is applied, leading to negative
influence on experiment results. In research studies concerning thermal comfort seeking
for satisfaction and perception, the survey with the seven-point scale has been applied
commonly. The seven-point scale is constituted by seven options from “very bright
(indicated by +3)” to “very dark (indicated by -3)”, together with a middle point of
“neutral (indicated by 0)”. That is to say, this scale includes a somewhat effective level
and an extremely high level and between them there is also one moderate level. Such a
scale could help participants to report their visual comfort and visual sensation. Figure
30
3.5 in the following demonstrates the questionnaire adopted in the study.
31
Figure 3.5 Sample Section of Designed Questionnaire
3.1.4 Pupil Size Parametric Data
As aforementioned in the last chapter, in order to collect the data of real-time pupil size
as well as other relevant parameter data concerning the human subjects, this study
adopted XG—a kind of mobile pupilometer ASL mobile eye. All participants were
required to wear the sensor devices during the complete experiment process.
32
Figure 3.6 Participant Wearing ASL Mobile Eye-XG Device
As a type of wearable sensor, this device combines two digital cameras with high
resolution. One of the digital cameras is used to record the eyes of participants and the
other is used for recording scene images. The data and images recorded would then be
processed by two tools—an independent video recording as well as an Excel document
where the pupil size of participants were recorded 30 times every second. In this study,
the sensing frequency for observing the significant changes of pupil sizes was one
second.
The statistical software packages of Minitab 17 and Microsoft Excel were applied to
processing and analyzing data. The details about data processing and analysis are
elaborated on in the fourth chapter. Previously, the radius was tried to be used to
measure the size of the pupil, but it is very challenge to identify a radius of the pupil
size due to its irregular shape depending on individuals.
Figure 3.7 A set of photos of the irregular shape of pupils
Thus, for the convenience of comparison and illustration, pupil size would be described
in pixels. As pupil size varies for each individual, data about every individual participant
33
collected in this study were standardized, so as to compare these physiological changes
among different participants or physiological groups and to estimate the change rate of
the pupil size under different lighting conditions. Chapter 4 describes the
abovementioned standardizing process.
3.2 Experimental Chamber Setup
In this experiment involving human participants, the environmental laboratory at the
Watt Hall’s basement floor was used. The environment of this laboratory simulated that
of an office for an individual with controllable lighting conditions, which could not be
affected by conditions of its adjacent workplace. Meanwhile, other major parameters
related to indoor environment quality, like air, acoustic conditions and thermal
conditions, are also controlled and monitored in this laboratory. Those parameters were
maintained at a constant level during the whole experiment, so that only the
physiological responses of the human on the basis of lighting were explored.
3.2.1 Chamber Design
With the strong support of the University of Southern California School of Architecture,
an environmental chamber was provided for the experiment. The room is approximately
15 square meters and the size of the entire chamber is 112” x 120” x 95”. Since the
chamber is on the basement floor, the experiment results were not affected by daylight.
The chamber is divided into two internal spaces. One is the participant's test room; the
other is the investigator's monitoring room. This internal layout helps participants focus
on testing without distraction.
34
Figure 3.8 Finished chamber setting
The picture illustrates the layout of the test room in detail. In order to reduce the impact
of indoor environmental quality parameters on subjects other than lighting, a separate
room was designed for testing, while investigators were stationed in a nearby monitoring
room to supervise the experiment and provide any assistance as needed. Taking into
account the requirements of fire safety, the space was divided between rooms using a
movable curtain partition with a transparent window so that investigators can observe
through the window. The curtain partition and indoor wall were pure white, which is a
typical color of interior walls in the office environment, which will not produce any
luster reflection to affect the experimental results.
35
Figure 3.9 Diagrammatic plan of Modifying Chamber
Figure 3.10 Diagrammatic plan of Modifying Chamber
36
3.2.2 Lighting Fixtures
Thirty dimmable LED light units (model: Coidak Dimmable LED, 730 lumens) were
installed on the laboratory ceiling to create the required lighting conditions indoors.
Each lamp emits a luminous flux of up to 730 lumens according to the lamp's technical
parameter table. Since the LED lamp does not release heat during operation, the use of
the LED lamp eliminates any effect of radiant heat from the lamp on the thermal quality.
The color temperature of the lamp can also be adjusted and the color temperature of
each lamp is set to 5000 Kelvin in the experiment, which is similar to the color
temperature of the white fluorescent lighting device in the traditional office
environment. The illuminance generated on the table surface is 50 lux to 1400 lux with
an interval of 150 lux. According to previous studies by the authors, the 150 lux
illumination interval is the smallest step of the perceived illumination change. In
addition, the experiment uses a programmable dimmer compatible with LED lamps to
produce the desired illumination level.
Figure 3.11 Dimmable LED Lamp
3.3 Research Tools and Sensor Devices
This section describes in detail all tools and sensor apparatuses used in the study.
37
3.3.1 Sensory Devices: Illuminance Meter, Luminance Meter and HDR Camera
Three illuminance meters were used to measure and collect illumination intensity data
(Figure 3.12). They are placed on the left, middle, and right sides of the table at equal
intervals so that the measured data can be more accurately reflected. The illuminance
meters were driven by batteries to output an analog signal of 0-5v to a data acquisition
device (DAQ), and at the same time, the data acquisition device converts the analog
signal into an illuminance value and stores them.
The Coolpix 8400 HDR camera (Figure 3.13) equipped with a fisheye lens is used in
conjunction with Photolux 2.1 software to estimate luminance data and UGR values.
The camera settings are summarized in Table 3.4.
Table 3.4 Settings for COOLPIX 8400 in “M” mode
38
Four photographs were taken at each illumination level to calculate luminance data in
Photolux 2.1. Each image was taken with different aperture values and exposure time
settings, as shown in Table 3.5.
Table 3.5 Aperture and Exposure Time Settings
The measured high-resolution luminance is compared with the luminance estimated by
the software. Studies have confirmed that the deviation between measured and
calculated values is small.
Figure 3.12 OMEGA HHLM-1 Figure 3.13 Coolpix 8400
3.3.2 DAQ Device
Another data acquisition device used in this experiment was NI USB 6008 (Figure 3.14).
It was connected to the illuminance meter and transmits the collected data to the
39
notebook computer. All data is automatically saved to the selected file through the
program developed by LabVIEW software and marked with detailed information
including date, time, etc.
Figure 3.14 NI USB-6008
3.3.3 ASL Mobile Eye XG
ASL mobile eye XG is the most essential and core device in this study to detect,
measure, and collect real-time pupil size data for participants (Figure 3.15).
Figure 3.15 ASL Mobile Eye XG
The device consists of three main parts: a monitor, specially-designed glasses with a
camera and a laptop with analysis software. After the connection according to the
instruction manual, the three components work together to collect the data of the pupil
size parameters of the participant. At the beginning of the experiment, each participant
was asked to adjust the glasses to his or her own circumstances and to keep track of the
glasses throughout the experiment. The glasses were equipped with two cameras, one
40
for recording a real-time image of the pupil and the other for monitoring the front view
seen by the human subject. Pupil size data is automatically saved to the computer and
the preset analysis software automatically saved the data as CSV files for future analysis
in Microsoft Excel and Minitab 17.
Figure 3.16 A photo taken by the front camera of the pupilometer during the experiment
3.4 Adopted Software
This research uses several programmable data collection software programs and data
statistical analysis software. The project first needed to determine an effective scheme
for collecting illuminance data. Subsequently, other software was used for luminance
analysis, determination of the uncertain glare level (UGR), and data processing.
3.4.1 LabVIEW
LabVIEW, developed by NI (National Instrument) Company of America, is a kind of
development environment and graphical program. It has some similarities with the
development environment of Basic and C, yet one of the most prominent differences
between other kinds of computer languages and LabVIEW is that LabVIEW writes
programs through the graphical editing language G and generates programs in the form
of block diagrams, whereas other kinds of computer languages generate code by
41
languages based on text. Moreover, as the core for the design platform of NI, LabVIEW
software is an optimal choice for controlling systems or developing measurements. The
development environment of the LabVIEW comprises all necessary tools for scientists
and engineers to build applications quickly, so that they could better improve
productivity, make innovation and solve problems.
3.4.2 Programming Logic
According to the operating instructions of the illuminance meter, the calculation formula
of illuminance was compiled by the author. The signal range transmitted by the data
acquisition sensor was 0-5v and the calculated illuminance is displayed on the notebook
computer according to the conversion of the illuminance formula. The user can also set
the time interval and select the desired channel according to the project purpose. The
file name is automatically created based on the date and time of the experiment. Figure
3.17 and Figure 3.18 show the appearance and graphical programming interface in
LabVIEW. The interface has been modified and developed according to the products
provided by Prof. Joon-Ho Choi.
Figure 3.17 Front Panel of Designed Program in LabVIEW
42
Figure 3.18 Block Diagram of Designed Program in LabVIEW
3.4.3 Photolux 2.1
Photolux 2.1 was used to calculate the luminance of the TV and the computer.
Luminance meters cannot measure large areas of brightness, so better methods are
needed to estimate the luminance level of a particular area. Photolux solves this problem
by combining and analyzing all four images taken with different aperture and exposure
settings into a single file.
Figure 3.19 Processed and analyzed image after combining four images taken at different
exposure settings for the same illuminance setting in Photolux.
43
3.4.4 Minitab
In this study, Minitab was used to statistically analyze data sets to determine the
correlation between parameters.
Minitab statistical software is a software package founded in 1972 at Pennsylvania State
University in the United States. So far, it has been widely used in more than 100
countries and more than 4800 colleges and universities around the world.
Minitab applications typically incorporate statistical processing methods such as six
sigma, capability maturity model integration (CMMI), and other process improvement
methods. It provides a large number of statistical analysis functions, including t-test,
correlation, regression, variance analysis, etc. It also has a powerful graphics generator
that helps better interpret and display data.
3.5 Experiment Rounds
The whole experiment took three rounds with 10 participants in each round.
44
Figure 3.20 Sample Section of Designed Questionnaire
The experiment’s first round consists of three sub-rounds of experiments and only one
subject at a time. One fixed contrast ratio was set in each round as illuminance level
changed from 50 lux to 1400 lux with a 150 lux interval. See the previous section for
specific parameter settings. During the experiment, light was changed from dark to
bright gradually. This procedure is known as light adaption which usually only takes
human eyes less than 1 minute to adapt to the new lighting condition. Thus, each
illuminance level step was designed with 2 minutes for stabilization and 1 minute for
data collection. The whole experiment lasted one hour and thirty minutes, all data, such
as illuminance and pupil size is automatically recorded by the devices. The subjects
were assigned some simple computer-based work during the test, such as reading and
typing. At the end of each illuminance level change, the subject was required to report
his or her visual sensation by filling the seven-point scale questionnaire mentioned in
3.1.3. All choices should be made based on the subjects’ perceptions about the lighting
45
condition during the final 1 minute.
The second round of experiments were conducted with opposite illumination level order
while other parameters remained unchanged.
Figure 3.21 part of pictures took for the human subjects during experiment
Participants were recruited through flyers and electronic postings with most of them
being on-campus students. Because of the diversity limitation of a campus population,
the demographic information of the subjects shows some limitations. Therefore, in order
to get as unbiased results as possible, efforts were made to balance the sample sizes by
critical human factors of visual sensation when creating the data pool, including gender,
age, ethnic origins and glasses-worn condition. Table 4.1 of Chapter 4 listed the
demographic information about participants.
3.6 DATA Analysis
46
After each round of testing, the collected data was summarized and analyzed in units of
individual subjects. Each participant is combined into a single Excel data set including
personal information, pupil size, illumination level, UGR, questionnaire responses to
visual perception and comfort conditions. Minitab is used for statistical analysis. Curve
fitting is carried out by using MATLAB software. The results and analysis would be
discussed in detail in Chapter 4.
47
CHAPTER 4: STUDY RESULTS
The experiment was conducted at an environmental chamber located in the basement of
Watt Hall on the University Park campus. After two rounds of experiments, all data was
collected and analyzed according to the experiment’s multiple round sequence. For the
lighting parameters, this study considered three variables: illuminance level, luminance
contrast ratio and Unified Glare Rating (UGR). In the analysis of each round, the three
lighting parameters were presented about each individual’s pupil size and visual sensation.
This chapter focuses on the presentation, preliminary analysis and discussion of all
experimental data. More detailed discussion is in Chapter 5. All data presented here is raw
and, thus, not processed.
4.1 Demographic Information of Test Subjects
The 20 participants of the experiment were recruited through flyers and electronic postings
with most of them being on-campus students. Thus, the demographic information of the
participants shows some limitations because of the diversity limitation of the campus
population. Therefore, in order to get as unbiased results as possible, efforts were made to
balance the sample sizes by critical human factors of visual sensation when creating the
data pool, including gender, age, ethnic origins and glasses-worn. Table 4.1 and table 4.2
list the demographic information concerning the participants.
Table 4.1 Demographic information about
first round of human subjects
Gender Age Iris Color Glass-worn
Male Female <25 ≥25 Blue Brown Yes No
Sample No. 4 6 7 3 2 8 5 5
Table 4.2 Demographic information about second round of human subjects
Gender Age Iris Color Glass-worn
Male Female <25 ≥25 Blue Brown Yes No
48
Sample No. 3 7 4 6 3 7 5 5
4.2 First Round Experiment
For the first round experiment, the light was changed from dark to bright over 10 steps.
The difference between every two steps was 150 lux. The outcome of exercisable contrast
ratios was picked to be 6.4/1, 1/1 and 1/4.25. The air temperature, relative humidity,
acoustic condition and CO2 during the experiment were ranged from 24.5 ± 0.5℃, 32 ±
2.5%, 30 dB and 610 ± 35 ppm, respectively. Ten participants attended the first round
experiment.
4.2.1 Illuminance Level per Visual Sensation by Individual
Pupil size changes with the illuminance level; the brighter the background light, the smaller
the pupil size, and vice versa. In the experiments, illumination and visual perception all
showed a positive correlation. That is, the lower the illuminance, the smaller the sensation
value reported by the participants; the higher the illuminance, the greater the sensation
value. Figure 5 was plotted based on the experimental results of the first round. For each
subject under testing, an explicit increasing pattern could be observed from Figure 4.1. This
observation is also consistent with common knowledge. However, for each participant, the
individual's perception of the same light intensity is very different. For example, 57 lux
was reported by subject 8 as a neutral sensation while 354 lux, 620% higher than the
previous value, was reported by subject 5 as a neutral perception.
49
Figure 4.1 Boxplot of Illuminance per Sensation in Each Subject’s Test
4.2.2 Pupil Size Between Luminance Contrast Ratio Group by Individual
Since contrast ratio is consistent during each sub-round experiment, we only discuss the
relationship between visual sensation and contrast ratio among three sub-round
experiments in Chapter 5. Below is Figure 4.2 (a)-(j), which is the interval plots of pupil
size between contrast ratio groups for the individuals of the first round experiment. For
each subject, the distinct difference of pupil size among the three contrast ratio groups can
be observed. Thus, the inference drawn can be that contrast ratio is one significant
influential factor that impacts the pupil’s dilation and contraction when the other lighting
parameters remain unchanged. Another finding is most subjects had the largest dilated
pupil size when the contrast ratio is 6.4/1 and had the smallest contrasted pupil size when
the contrast ratio is 1/4.25. However, the test results of two subjects who have high myopia
showed the opposite pattern. It cannot be asserted that this phenomenon is due to high
myopia, since the sample size is too small.
No
Visual Sensation
10 9 8 7 6 5 4 3 2 1
3 2 1 0 -1 -2 -3 3 2 1 0 -1 -2 -3 3 2 1 0 -1 -2 -3 3 2 1 0 -1 -2 -3 3 2 1 0 -1 -2 -3 3 2 1 0 -1 -2 -3 3 2 1 0 -1 -2 -3 3 2 1 0 -1 -2 -3 3 2 1 0 -1 -2 -3 3 2 1 0 -1 -2 -3
1600
1400
1200
1000
800
600
400
200
0
Illuminance (lux)
Boxplot of Illuminance per Visual Sensation
50
51
52
53
Figure 4.2 (a)-(j) Interval plot of pupil size between contrast ratio group of the
first round
experiment by individual
4.2.3 Visual Sensation per UGR Group by Individual
In the experiments, pupil size and Unified Glare Rating (UGR) did not show a strong
correlation. Figure 4.3 (a)-(j) was plotted based on the experimental result data of the first
round. Overall, pupil size and Unified Glare Rating (UGR) does not show a positive or
negative correlation. Also, for each subject, the individual's perception of the same UGR
value is very different. The inference is that UGR value might have a stronger correlation
with visual satisfaction instead of visual sensation. Based on this result, UGR will not be
discussed in Chapter 5 as there is a weak correlation between pupil size and UGR value.
54
3 2 1 0 -1 -2 -3
18
17
16
15
14
13
12
11
10
9
Visual Sensation
Unified Glare Rating (UGR)
95% CI for the Mean
Individual standard deviations were used to calculate the intervals.
Interval Plot of UGR Value per Visual Sensation by Individual_Subject A
3 2 1 0 -1 -2 -3
75
50
25
0
-25
-50
Visual Sensation
Unified Glare Rating (UGR)
95% CI for the Mean
Individual standard deviations were used to calculate the intervals.
nterval Plot of UGR Value per Visual Sensation by Individual_Subject C
2 1 0 -1 -2
25
20
15
10
5
0
Visual Sensation
Unified Glare Rating (UGR)
95% CI for the Mean
Individual standard deviations were used to calculate the intervals.
Interval Plot of UGR Value per Visual Sensation by Individual_Subject B
55
3 2 1 0 -1
30
25
20
15
10
5
0
Visual Sensation
Unified Glare Rating (UGR)
95% CI for the Mean
Individual standard deviations were used to calculate the intervals.
nterval Plot of UGR Value per Visual Sensation by Individual_Subject E
3 2 1 0 -1 -2
40
30
20
10
0
Visual Sensation
Unified Glare Rating (UGR)
95% CI for the Mean
Individual standard deviations were used to calculate the intervals.
nterval Plot of UGR Value per Visual Sensation by Individual_Subject D
3 2 1 0 -1
60
50
40
30
20
10
0
-10
-20
-30
Visual Sensation
Unified Glare Rating (UGR)
95% CI for the Mean
Individual standard deviations were used to calculate the intervals.
nterval Plot of UGR Value per Visual Sensation by Individual_Subject F
56
3 2 1 0 -1 -2
25
20
15
10
5
Visual Sensation
Unified Glare Rating (UGR)
95% CI for the Mean
Individual standard deviations were used to calculate the intervals.
Interval Plot of UGR Value per Visual Sensation by Individual_Subject I
3 2 1 0 -1 -2 -3
22
20
18
16
14
12
10
Visual Sensation
Unified Glare Rating (UGR)
95% CI for the Mean
Individual standard deviations were used to calculate the intervals.
Interval Plot of UGR Value per Visual Sensation by Individual_Subject H
57
Figure 4.3(a)-(j) Interval plot of UGR value per visual sensation by individual
4.3 Second Round Experiment
For the second round experiment, the light was changed from bright to dark over 10 steps.
The difference between two steps was 150 lux as well. The outcome of exercisable contrast
ratios was picked to be 6.4/1, 1/1 and 1/4.25. The air temperature, relative humidity,
acoustic condition and CO2 during the experiment were ranged from 24.5 ± 0.5℃, 32 ±
2.5%, 30 dB and 610 ± 35 ppm, respectively. Ten subjects attended the second round
experiment as well.
4.3.1 Illuminance level per visual sensation by individual
Just as the first round, the boxplot of illuminance per sensation in each subject’s test was
generated based on the experiment data of the second round. Contrasted with the
discoveries in the first round experiments, comparable discoveries about illuminance-
sensation patterns were made. The raw data is summarized in Figure 4.4.
2 1 0 -1
60
50
40
30
20
10
0
-10
-20
-30
Visual Sensation
Unified Glare Rating (UGR)
95% CI for the Mean
Individual standard deviations were used to calculate the intervals.
nterval Plot of UGR Value per Visual Sensation by Individual_Subject J
58
Figure 4.4 Boxplot of Illuminance per Sensation in Each Subject’s Test
4.3.2 Pupil Size Between Luminance Contrast Ratio Group by Individual
The same data process method of the previous round was adopted. The interval plot of
pupil size between contrast ratio group by the second round experiment by individual was
generated based on the experiment data of the second round, shown in Figure 4.5 (a)-(j).
The previous inference was confirmed since similar findings about the trend of pupil size-
contrast ratio pattern were revealed compared to the first round experiments.
59
60
61
62
Figure 4.5(a)-(j) Interval plot of pupil size between contrast ratio group by first round
experiment by individual
4.3.3 Visual Sensation per UGR Group by Individual
The same data process method of the previous round was adopted and the interval plot of
pupil size between UGR by the second round experiment by individual was generated
based on the experiment data of the second round, shown in Figure 4.6 (a)-(j). Similar
findings of the trend of pupil size-UGR pattern were revealed compared to the first round
experiments.
63
3 2 1 0 -1 -2 -3
75
50
25
0
-25
-50
Visual Sensation
Unified Glare Rating (UGR)
95% CI for the Mean
Individual standard deviations were used to calculate the intervals.
Interval Plot of UGR value per Visual Sensation by Individual_Subject C
3 2 1 0 -1 -2 -3
18
17
16
15
14
13
12
11
10
9
Visual Sensation
Unified Glare Rating (UGR)
95% CI for the Mean
Individual standard deviations were used to calculate the intervals.
Interval Plot of UGR value per Visual Sensation by Individual_Subject A
2 1 0 -1 -2
24
22
20
18
16
14
12
10
8
Visual Sensation
Unified Glare Rating (UGR)
95% CI for the Mean
Individual standard deviations were used to calculate the intervals.
Interval Plot of UGR value per Visual Sensation by Individual_Subject B
64
3 2 1 0 -1
30
20
10
0
-10
Visual Sensation
Unified Glare Rating (UGR)
95% CI for the Mean
Individual standard deviations were used to calculate the intervals.
Interval Plot of UGR value per Visual Sensation by Individual_Subject E
3 2 1 0 -1 -2
40
30
20
10
0
-10
Visual Sensation
Unified Glare Rating (UGR)
95% CI for the Mean
Individual standard deviations were used to calculate the intervals.
Interval Plot of UGR value per Visual Sensation by Individual_Subject D
3 2 1 0 -1
60
50
40
30
20
10
0
-10
-20
-30
Visual Sensation
Unified Glare Rating (UGR)
95% CI for the Mean
Individual standard deviations were used to calculate the intervals.
Interval Plot of UGR value per Visual Sensation by Individual_Subject F
65
3 2 1 0 -1 -2 -3
35
30
25
20
15
10
5
0
Visual Sensation
Unified Glare Rating (UGR)
95% CI for the Mean
Individual standard deviations were used to calculate the intervals.
Interval Plot of UGR value per Visual Sensation by Individual_Subject G
3 2 1 0 -1 -2 -3
22
20
18
16
14
12
10
Visual Sensation
Unified Glare Rating (UGR)
95% CI for the Mean
Individual standard deviations were used to calculate the intervals.
Interval Plot of UGR value per Visual Sensation by Individual_Subject H
3 2 1 0 -1 -2
22
20
18
16
14
12
10
8
6
Visual Sensation
Unified Glare Rating (UGR)
95% CI for the Mean
Individual standard deviations were used to calculate the intervals.
Interval Plot of UGR value per Visual Sensation by Individual_Subject I
66
Figure 4.6 (a)-(j) Interval plot of UGR value peri visual sensation by individual
2 1 0 -1
60
50
40
30
20
10
0
-10
-20
-30
Visual Sensation
Unified Glare Rating (UGR)
95% CI for the Mean
Individual standard deviations were used to calculate the intervals.
Interval Plot of UGR value per Visual Sensation by Individual_Subject J
67
CHAPTER 5: DATA ANALYSIS AND DISCUSSION
This study was designed to provide designers and engineers with insight into the visual
sensation influential factors that they use to improve the future lighting control system.
Specifically, it was intended to assist in optimizing lighting control in the office
environment in which defective condition can possibly bring work productivity down and
physical side effects, for example, the eye fatigue and headaches.
The purpose of this chapter is to analyze and discuss the data that were collected to estimate
the overall visual sensation as a function of human pupil size, different lighting parameters
and other physiological characteristics.
The whole experiment was designed regarding requisites to achieve the research goal,
which includes the qualitative data of human subjects’ physiological characteristic, the
objective physiological response data of pupil size, the subjective evaluation value of visual
sensation and visual satisfaction. All these data were summarized, and relative data display
graphics were plotted in this chapter for resultant discoveries.
5.1 Pupil Sizes per Illuminance Level by Individual
Figure 5.1(a)-(j) below was plotted based on the experimental result data of 1st round,
contrast ratio 1:1. The quartile distribution of measured pupil sizes (in pixels) that
corresponded to changes in illuminances from 50 lx to 1400 lx, is illustrated in each line.
68
69
1400 1250 1100 950 800 650 500 350 200 50
90
80
70
60
50
40
Illuminance(Lux)
pupil r_Subject D
Boxplot of pupil r_Subject D
1400 1250 1100 950 800 650 500 350 200 50
80
75
70
65
60
55
50
45
Illuminance(Lux)
pupil r_Subject E
Boxplot of pupil r_Subject E
70
1400 1250 1100 950 800 650 500 350 200 50
70
65
60
55
50
45
40
Illuminance(Lux)
pupil r_Subject H
Boxplot of pupil r_Subject H
1400 1250 1100 950 800 650 500 350 200 50
100
90
80
70
60
50
Illuminance(Lux)
pupil r_Subject I
Boxplot of pupil r_Subject I
71
Figure 5.1(a)-(j) Pupil size range per illuminance level by individual
In the experiments, an explicit decreasing pattern could be observed. Illumination inversely
increased with the decrease in pupil size, that is, the higher the illuminance, the smaller
size of the pupil (in pixels) recorded by the pupilometer. The lower the illuminance, the
greater the pupil size. This observation is also consistent with common sense. However,
the pupil size under the same light intensity varied from person to person.
For example, 58 pixels of average pupil size was reported by subject E at 500 lux
illuminance level while 72 pixels, 24% higher than the previous value, was reported by
subject G at the same illuminance level. But overall, the trend of pupil size decreasing as
the illuminance increasing is consistent.
72
Figure 5.2 Average Pupil Sizes per Illuminance Level
The coefficient of correlation between illuminance and the average pupil size is -0.89.
Statistically, A coefficient of correlation of +0.8 or -0.8 indicates a strong correlation
between the independent variable and the dependent variable. Therefore, there is a strong
correlation between illuminance and pupil size. Illuminance is the most important factor
affecting pupil size compared to other factors considered by this experiment, such as UGR
value of which coefficient is 0.588.
Figure 5.3 The coefficient of correlation between illuminance and average pupil size
1400 1250 1100 950 800 650 500 350 200 50
70.0
67.5
65.0
62.5
60.0
57.5
55.0
Illuminance(Lux)
Avr pupil r (pixel)
Boxplot of Avr pupil r (pixel)
73
5.2 Illuminance Range per Visual Sensation by Individual
5.2.1. First Round Experiment
The illuminance level setting of the experiment ranged from 50lux to 1400lux with steps
of 150lux, stepwise increased. This section only analyzes the relationship between
illumination and visual perception; only one fixed contrast ratio was picked since no more
variables were needed other than illuminance level.
The contrast ratio setting selected was 1 to 1, which satisfied the Illuminating Engineering
Society of North America (IESNA) recommended value between task and adjacent
surrounding screen. Figure 5.4 illustrated illuminance range per sensation in each subject’s
test by the boxplot. It was plotted based on the experimental result data of first round.
First, illumination and visual perception showed a positive correlation, that is, the lower
the illuminance, the smaller the sensation value reported by the subjects; the higher the
illuminance, the greater the sensation value. For each subject under test, an explicit
increasing pattern could be observed from the figure 5.4. This observation is also consistent
with common sense.
Second, 65% of the whole visual sensation selection was marked as the overall bright
sensation which includes ‘slightly bright’ (+1), ‘bright’ (+2) and ‘very bright’ (+3); 15%
were marked as neutral, and 20% were considered as overall dark sensation. This result
reminded us that the dominant visual sensation of the subjects was a bright sensation during
this research illuminance test range, which covers from 50 lux – 1400 lux.
Last but not least, for each subject, the individual's perception of the same light intensity is
very different at the same time. For example, 57 lux was reported by subject ten as a neutral
sensation while 354 lux, 620% higher than the previous value, was reported by subject nine
as a neutral perception.
74
Figure 5.4 Boxplot of Illuminance per Sensation in Each Subject’s Test
Figure 5.5 is a dot plot of all illuminance data by visual sensation. This graphic displays
the illuminance distribution range by each visual sensation in a very visual way. The
illuminance range of the subjects’ neutral (0) sensation widely ranged from 50 lux to 1000
lux and ‘bright’ (+2) sensations were reported with conditions between 200 lux and 1400
lux. Illuminance level 375 lux was marked as ‘dark’ (-2) to ‘bright’ (+2) by different subject.
This once again strengthens that the visual perception of people is very different, lighting
needs to be adjusted from person to person.
Figure 5.5 Dotplot of Illuminance Distributions per Sensation in Each Subject’s Test
No
Visual Sensation
10 9 8 7 6 5 4 3 2 1
3 2 1 0 -1 -2 -3 3 2 1 0 -1 -2 -3 3 2 1 0 -1 -2 -3 3 2 1 0 -1 -2 -3 3 2 1 0 -1 -2 -3 3 2 1 0 -1 -2 -3 3 2 1 0 -1 -2 -3 3 2 1 0 -1 -2 -3 3 2 1 0 -1 -2 -3 3 2 1 0 -1 -2 -3
1600
1400
1200
1000
800
600
400
200
0
Illuminance (lux)
Boxplot of Illuminance per Visual Sensation
1400 1200 1000 800 600 400 200
-3
-2
-1
0
1
2
3
Illuminance (lux)
Visual Sensation
Each symbol represents up to 33 observations.
Dotplot of Illuminance by Visual Sensation
75
5.2.2 Second Round Experiment
The illuminance level setting of the experiment ranged from 50lux to 1400lux with steps
of 150lux, stepwise decreased. The contrast ratio setting selected for second round
experiment analysis was also 1 to 1.
This time, although the order of illumination adjustment is changed, illumination and visual
perception still showed a positive correlation. The visual sensation from ‘slightly bright’
(+1) to ‘very bright’ (+3) are still most prevalent which accounted for 69% of whole visual
sensation selection.
Figure 5.6 Boxplot of Illuminance per Sensation in Each Subject’s Test
There is no doubt that the dominant visual perception of some illuminance can be seen at
a glance in Figure 5.7 below. For example, all the subject marked ‘bright’ (+2) or ‘very
bright’ (+3) at the illuminance level around 1400 lux. Some illuminance levels were marked
by multiple visual sensations, for example, 500 lux was regarded as ‘slightly dark’ (-1),
‘neutral’ (0), ‘slightly bright’ (+1) or ‘bright’ (+2) by different subject. The dominant visual
sensation of each visual sensation can be seen as the sample size increases.
76
Figure 5.7 Dotplot of Illuminance Distributions per Sensation in Each Subject’s Test
5.2.3 Discussion Differences Between the Two-Round Experiment
All the discussion in this section once again strengthens that the visual perception of people
is very different, lighting needs to be adjusted from person to person.
Since the two round experiments have different subjects, it is inappropriate to compare
visual sensation in each round experiment by individual due to the physiological difference.
The overall illuminance distribution pattern by visual sensation between the two rounds of
the experiment was compared to see if lighting intensity sequence of adjustment impact
visual perception since this is the only difference in the lighting of these two-round
experiment.
All of us know the critical effect of illumination on perception, but it seems the order of
illumination adjustment has little impact in the range of 50 lux - 1400 lux. For each
illuminance interval covered by the same visual sensation, the dominant illuminance level
picked by the most subjects is always the same compared between the two rounds of the
experiment. For example, 200 lux was within neutral perception and chosen by the highest
number of participants in both two round experiments. In other words, in most cases, there
1400 1200 1000 800 600 400 200
-3
-2
-1
0
1
2
3
Illuminance (lux)
Visual Sensation
Each symbol represents up to 33 observations.
Dotplot of Illuminance by Visual Sensation
77
are going to have identical sensation mark to the same subjects no matter the light intensity
adjust sequence is from dark to bright or from bright to dark in the range of 50 lux to 1400
lux.
5.3 Raw Data of Pupil Size Per Sensation by Individual
Previous research has discussed the impact of individual difference on pupil size under the
same lighting condition. Rui concluded physiological characteristics as the reason of the
fact observed that some subjects showed relatively small ranges of pupil size changes
between the visual sensations while others generated large variations in their pupil sizes
across the visual sensations. (Rui, 2014)
For this study, the researcher also noted this phenomenon during the experiment. For a
better explanation, random subjects were selected from the first to compare their raw pupil
size changes with visual sensation at contrast ratio 1 to 1. The pupil is measured in pixel
and contrast ratio 1 to 1 is under the Illuminating Engineering Society of North America
(IESNA) recommended luminance ratios range between task and adjacent surrounding
screen.
78
Figure 5.8 Interval Plot of Raw Pupil Size of Subject A
Figure 5.9 Interval Plot of Raw Pupil Size of Subject B
3 2 1 0 -1 -2 -3
77.5
75.0
72.5
70.0
67.5
65.0
Visual Sensaton
Raw Pupil Size (Pixel)
95% CI for the Mean
Individual standard deviations were used to calculate the intervals.
Interval Plot of Raw Pupil Size_Subject A
3 2 1 0 -1 -2 -3
65.0
62.5
60.0
57.5
55.0
Visual Sensaton
Raw Pupil Size (Pixel)
95% CI for the Mean
Individual standard deviations were used to calculate the intervals.
Interval Plot of Raw Pupil Size_Subject B
79
Above Figure5.8 and Figure5.9 conclude two subjects’ experimental results under the
contrast ratio 1 to 1, first round experiment. First, both of the subjects didn’t consider the
illuminance level as very dark or very bright. This echoes the discovery gotten from the
dot plot of illuminance distribution figure in the previous section. Second, another finding
is how absolute pupil size is distinct from person to person under the same lighting
condition. Subject A has 76 pixel of pupil size at the dark sensation and 65.6 pixel of pupil
size at bright sensation while subject B has 63.5 pixel and 54 pixels, relatively.
Figure5.10 summarized raw data of pupil size of random selected subject from 1
st
round
experiment per visual sensation. A decreasing pattern of pupil size can be observed. For
each individual, the overall tendency of pupil size change is to decrease as visual sensation
increased from very bright’ (+3) to ‘very dark’ (-3). For pair comparison, for example, even
without the use of statistical tools, it was clear that the pupil size absolute values of subjects
five and six were completely different under the same visual sensation. The analysis of
variance reported a p-value of 0.00 that is lower than 0.05 in the level of 95% confidence.
Figure 5.10 Boxplt of Raw Pupil Size per Visual Sensation (first round)
80
Figure 5.11 Dotplot of Raw Pupil Size by Visual Sensation (first round)
Figure 5.12 illustrated raw data of pupil size of randomly selected subject from 2
nd
round
experiment per visual sensation. The second-round experiment has opposite lighting
adjustment sequences compared to the first-round experiment.
Figure 5.12 Boxplt of Raw Pupil Size per Visual Sensation (second round)
99 90 81 72 63 54 45
-3
-2
-1
0
1
2
3
Raw Pupil Size (Pixel)
Visual Senstaion
Each symbol represents up to 14 observations.
Dotplot of Raw Pupil Size by Visual Sensation
81
Figure 5.13 Dotplot of Raw Pupil Size by Visual Sensation (second round)
Similarly, for each individual, visual sensation increased from very bright’ (+3) to ‘very
dark’ (-3) while pupil size decreased in the meantime. For pair comparison, with the
different illumination adjustment sequences, subjects No.5 and No.6’s pupil size absolute
values still have different change range under the same visual sensation but have the same
pattern.
To conclude above, measuring visual perception in terms of the absolute value of pupil size
can only represent the individual’s visual sensation. Otherwise, the evaluation results can
be very inaccurate due to physiological difference. Thus, new metrics need to be considered
since the absolute value of pupil size is not a good indicator.
Meanwhile, another finding is that if every subject’s pupil size of neutral sensation was
taken as a base value, to every single subject, the pupil size change percentage varied in a
certain range and the order of increasing and decreasing illuminance has little impact on
the base value for every subject, which reminds us to utilize the change percentage of pupil
size to evaluate visual sensations in the large population.
99 90 81 72 63 54 45
-3
-2
-1
0
1
2
3
Raw Pupil Size (Pixel)
Visual Senstaion
Each symbol represents up to 14 observations.
Dotplot of Raw Pupil Size by Visual Sensation
82
5.4 Data Processing of Raw Pupil Size
5.4.1. Pupil Size Standardization
Just as the perception of thermal condition might differ significantly between individuals,
contract and dilate ratio of the human pupil varies depending on the individual. In other
words, the absolute value of pupil size of different people might distribute in different
scopes even under the same lighting condition. In our case, this statement was also proven
through the discussion above in section 4.1.
Thus, since raw data of pupil size has little comparability, the collected pupil size data
needed to be standardized by adopting the following equation (1) to diminish the bias due
to individual physiological characteristics.
Standardized Pupil Size (%)
＝(
𝑃𝑢𝑝𝑖𝑙 𝑆𝑖𝑧𝑒 (𝑖 )−𝑃𝑢𝑝𝑖𝑙 𝑆𝑖𝑧𝑒 ( 𝑛𝑒𝑢𝑡𝑟𝑎𝑙 𝑠𝑒𝑛𝑠𝑎𝑡𝑖𝑜𝑛 )
𝑃𝑢𝑝𝑖𝑙 𝑆𝑖𝑧𝑒 ( 𝑛𝑒𝑢𝑡𝑟𝑎𝑙 𝑠𝑒𝑛𝑠𝑎𝑡𝑖𝑜𝑛 )
) × 100% …………………………… (1)
where i is visual sensation of human subjects to illuminance level.
83
Figure 5.14 Dotplot of Standardized Pupil Size per Visual Sensation (first round)
Overall speaking, standardized pupil size is distributed in each kind of significant
difference in visual perception, and this is consistent throughout each of the two rounds of
experiments. There was at least a roughly 4% difference between visual perception and
pupil size.
5.4.2 Moving Average Filtering
The frequency of the pupilometer we adopted in this study is 30 Hz which means the
apparatus recorded pupil size for 30 times in just one second and generated 30 pieces of
data in csv file at the meantime. Although a large number of data can improve the reliability
of the experimental results and reduce the sample errors, it often brings more data
fluctuations.
This study adopted a moving average which is commonly used with time series data to
smooth out short-term fluctuations and highlight longer-term trends or cycles. For example,
84
it is often used in technical analysis of financial data, like stock prices, returns or trading
volumes. It is also used in economics to examine gross domestic product, employment or
other macroeconomic time series. Mathematically, a moving average is a type of
convolution and so it can be viewed as an example of a low-pass filter used in signal
processing.
There is no fixed time window setting when the moving average was applied, since the
threshold between short-term and long-term depends on the application, and the parameters
of the moving average should be set accordingly.
In this thesis, three-time window setting, 10s, 20s and 30s, were tested on each subject’s
pupil size data in order to select the best threshold setting which filters the element of noise
and at the meantime preserves the curve characteristics in our case.
An example of a simple moving average for an n-second sample of pupil size is the mean
of the previous n seconds' pupil size. If those pupil size are PM, PM-1, ……, PM-(n-1), then
the formula is:
𝑃𝑢𝑝𝑖𝑙 𝑆𝑖𝑧𝑒 𝑀𝐴
=
𝑃 𝑀 +𝑃 𝑀 −1
+⋯+𝑃 𝑀 −(𝑛 −1)
𝑛 = ………………………………………… （2）
Figure 5.15(a)-(c) illustrated the comparison between the curve of one subject’s pupil size
raw data and processed data after 10s, 20s and 30s moving average smooth processing,
respectively. It was empirically found that 20s is the best option to choose for estimating a
gradient of pupil sizes. It seems to show the pattern of the original data without
compromising the sensitivity. Threshold 20s had the best noise elimination and curve
feature retention among three and was adopted for follow-up analysis.
85
a) 10s threshold
b) 20s threshold
86
c) 30s threshold
Figure 5.15(a)-(c) Threshold 10s, 20s and 30s moving average smooth processed pupil size
data, respectively.
5.4.3 Gradient of Pupil Size
One of the variables considered by this study is the change rate of the pupil size, which is
also known as gradient. The gradient was calculated from pupil size data which we
processed by moving average filtering with 20s threshold. In order to pick an appropriate
time window, 10s and 30s threshold has also been tested. The gradient of pupil size during
each 10-s interval was too small for a gradient to be estimated and the gradient of pupil
size during each 30-s interval was too big for a gradient to be estimated.
As a test result and also based on an empirical experience, 20s were selected to estimate
the gradient by using the formula below:
Gradient of pupil size (t) = Pupil size (t) – Pupil size (t - 19)
where t, present time point, and t-19: past time point by 19 units of 1 s (i.e., 19 s).
87
Thus, the change rate of the pupil size can be estimated as another parameter for the
following more detailed analysis.
5.5 Processed Pupil Size per Visual Sensation by Individual
5.5.1 Comparisons of Standardized Pupil Size per Visual Sensation Between Different
Contrast Ratio Group by Individual
Based on the analysis results in the chapter above, the conclusion can be made that pupil
size shows a significant difference for each subject between different luminance contrast
ratio groups, although the difference in values varied from person to person. Also, because
of the limitation of raw pupil size data, previous section 5.2 introduced another
measurement metric, standardization pupil size which eliminates deviations caused by
physiological factors. Thus, figure 5.16(a)-(j) was generated to illustrate comparisons of
standardized pupil size per visual sensation between different contrast ratio group by the
individual of the first-round experiment.
In figure 5.16(a)-(j), the y-axis displayed standardization of pupil size, in other words,
change in the range of pupil size taking pupil size at the neutral sensation as a base point.
X-axis displayed subject reported visual sensation from a seven-point scale questionnaire,
in which +3 represents very bright, +2 represents bright, +1,0 stand for slightly bright and
neutral sensation, and -3, -2, -1 represent very dark, dark and slightly dark, respectively.
Some subjects reported that they do not have very bright visual sensation or very dark
visual sensation during the experiment setting illuminance level 50 lux -1400 lux. Different
contrast ratio was represented by distinctive color and line shape style as well. Blue color
plus solid line stand for contrast ratio 1 to 1 group; Red color plus dash line stand for
contrast ratio 1 to 4.25 group; the green color with the dotted line represents contrast ratio
6.4 to 1 group.
First, all of the three contrast ratio groups show the distinct pattern that standardized pupil
88
size decrease as visual sensation changes from “very dark” (-3) to “very bright” (+3). Also,
each contrast ratio group showed positive standardized pupil sizes with a dark sensation
and negative pupil sizes with a bright sensation.
To simplified further analyses with subject groups, the collected pupil size and illuminance
data for individual sensations were regrouped from a seven-point scale to a three-point
scale to keep the scope of data at a level for statistical significance in the data analysis.
Therefore, visual sensations of _3 (very dark), _2 (dark) and _1 (slightly dark) were
clustered into ‘‘dark’’, and visual sensations of +1 (slightly bright), +2 (bright) and +3 (very
bright) were collected into ‘‘bright’’ for simplification
The two-sample t-tests reported significant differences in pupil size at each sensation in the
all three contrast ratio groups. The test reported a p-value of 0.000 in most of the cases,
which is less than 0.05 in the level of 95% confidence (Table 5.1).
Table 5.1 the pair t-test comparisons of standardized pupil size at dark, neutral and bright
visual sensation point for 1:1 contrast ratio
CR 1:1
Average Standardized Pupil
Size (%)
Two Sample T-test
Dark Neutral Bright
P-Value (neutral
vs. dark)
P-Value (neutral
vs. bright)
Subject A 6.4240 0.0074 -5.4521 0.000* 0.000*
Subject B 3.8822 0.0052 -8.6247 0.000* 0.000*
Subject C 4.0961 0.0408 -6.8569 0.000* 0.000*
Subject D 11.1148 0.0047 -6.0818 0.000* 0.000*
Subject E 8.3679 0.0051 -10.5885 0.000* 0.000*
89
Subject F 5.3593 0.0107 -9.7674 0.000* 0.000*
Subject G 7.3061 0.0033 -10.5695 0.000* 0.000*
Subject H 12.2325 0.0057 -6.7218 0.000* 0.000*
Subject I ˗˗˗ 0.0106 -23.9967 ˗˗˗ 0.000*
Subject J 19.7768 0.0048 -15.9288 0.000* 0.000*
*Asterisks indicate statistical significance (P < 0.05)
Table 5.2 the pair t-test comparisons of standardized pupil size at dark, neutral and bright
visual sensation point for 1:4.25 contrast ratio
CR 1:4.25
Average Standardized Pupil
Size (%)
Two Sample T-test
Dark Neutral Bright
P-Value (neutral
vs. dark)
P-Value (neutral
vs. bright)
Subject A 5.7739 0.0129 -4.0890 0.000* 0.000*
Subject B 5.7514 0.0109 -6.4685 0.000* 0.000*
Subject C 5.1176 -0.0334 -6.7054 0.000* 0.000*
Subject D 12.4747 0.0091 -4.8823 0.000* 0.000*
Subject E 9.4587 0.0194 -9.8299 0.000* 0.000*
Subject F 10.8901 0.0081 -11.6063 0.000* 0.000*
Subject G 15.4295 0.0098 -15.7028 0.000* 0.000*
Subject H 8.0869 0.0092 -5.5681 0.000* 0.000*
90
Subject I ˗˗˗ 0.0049 -15.2101 ˗˗˗ 0.000*
Subject J 16.9430 0.0049 -14.5634 0.000* 0.000*
*Asterisks indicate statistical significance (P < 0.05)
Table 5.3 the pair t-test comparisons of standardized pupil size at dark, neutral and bright
visual sensation point for 6.4:1 contrast ratio
CR 6.4:1
Average Standardized Pupil
Size (%)
Two Sample T-test
Dark Neutral Bright
P-Value (neutral
vs. dark)
P-Value (neutral
vs. bright)
Subject A 9.7806 0.0142 -2.0541 0.000* 0.000*
Subject B 7.4771 0.0120 -6.4978 0.000* 0.000*
Subject C 10.9563 -0.0367 -2.8591 0.000* 0.000*
Subject D 14.4894 0.0100 -4.6839 0.000* 0.000*
Subject E 8.2569 0.0213 -14.5670 0.000* 0.000*
Subject F ˗˗˗ 0.0089 -12.7231 0.000* 0.000*
Subject G 15.1984 0.0108 -9.0218 0.000* 0.000*
Subject H 6.7375 0.0101 -2.0969 0.000* 0.000*
Subject I 10.4111 0.0054 -7.4196 0.000* 0.000*
Subject J 15.7231 0.0054 -13.3564 0.000* 0.000*
*Asterisks indicate statistical significance (P < 0.05)
Second, each contrast ratio group has different Standardized Pupil Size per visual sensation.
91
Table 5.1, table 5.2 and table 5.3 concludes results of the pair t-test comparisons of
standardized pupil size at dark, neutral and bright visual sensation point for 1:1, 1:4.25 and
6.4:1 conditions. For example, for subject A, the comparison of the changes in average
pupil size with the neutral level showed about a 6.42% change for dark sensations and -
5.45% for bright sensations. All of the comparisons were statistically significant with P-
values lower than 0.05. These findings support the concept that visual sensations
correspond to standardized pupil sizes in test subjects, and that common decreasing
patterns are characteristic to each contrast ratio group.
Last but not least, table 5.4 summarized the average standardized pupil size between dark
and bright sensation per contrast ratio for each subject. When the contrast ratio is 1:1, the
average standardized pupil size at the neutral sensation was 0.0 while it was 8.72 when
dark sensation, and -10.45 when bright sensation, respectively. When contrast ratio is
1:4.25, the average standardized pupil size at the neutral sensation was 0.0 while it was
9.99 when dark sensation, and -9.45 when bright sensation, respectively. When contrast
ratio is 1:1, the average standardized pupil size at the neutral sensation was 0.0 while it was
11.00 when dark sensation, and -7.52 when bright sensation, respectively.
92
93
94
95
Figure 5.16 (a)-(j) Interval plot of standardized pupil size per visual sensation between
contrast ratio group by individual
Table 5.4 Average standardized pupil size between dark and bright sensation per contrast
ratio by individual
Standardized Pupil Size (%)
Contrast Ratio 1:1 1:4.25 6.4 :1
Dark Bright Dark Bright Dark Bright
Subject A 6.4240 -5.4521 5.7739 -4.0890 9.7806 -2.0541
Subject B 3.8822 -8.6247 5.7514 -6.4685 7.4771 -6.4978
Subject C 4.0961 -6.8569 5.1176 -6.7054 10.9563 -2.8591
Subject D 11.1148 -6.0818 12.4747 -4.8823 14.4894 -4.6839
Subject E 8.3679 -10.5885 9.4587 -9.8299 8.2569 -14.5670
Subject F 5.3593 -9.7674 10.8901 -11.6063 ˗˗˗ -12.7231
Subject G 7.3061 -10.5695 15.4295 -15.7028 15.1984 -9.0218
96
Subject H 12.2325 -6.7218 8.0869 -5.5681 6.7375 -2.0969
Subject I ˗˗˗ -23.9967 ˗˗˗ -15.2101 10.4111 -7.4196
Subject J 19.7768 -15.9288 16.9430 -14.5634 15.7231 -13.3564
Average 8.728856 -10.4588 9.991756 -9.46258 11.00338 -7.52797
5.5.2 Comparisons of Gradient Pupil Size per Visual Sensation Between Different
Contrast Ratio group by Individual
For a more comprehensive analysis, the change rate of pupil size (i.e., gradient) is
calculated as another analysis parameter. As discussed in the section 5.1, pupil size
decreased as the illuminance level increased. Even though there were some variations in
gradients, the gradients of those body points were all negative. These negative gradients
indicated there were continuous decreases across the pupil size, from “dark” through
“bright”.
Also, the calculation of the gradient of pupil sizes provided interesting facts of pupil size
change rate as illuminance level changed from “dark” through “bright”. The shape of the
gradients of pupil size is “/‾‾\”, which indicated the gradients of the pupil size were the
lowest values in the dark sensation, and the values increased and kept relatively stable
during neutral sensation, then dropped hugely in the bright sensation. Most of the gradient
of the pupil size was nearly a zero, with a very narrow confidence interval. This value
indicated that the pattern was consistent for all of the subjects.
The gradient is the slope of the pupil size curve as a function of illumination. Thus we
know that as the illuminance increases, the decrease pattern of pupil size is a non- linear
curve instead of linear curve (figure5.17(a)-(j)).
97
98
99
100
Figure 5.17 (a)-(j) Interval plot of gradient pupil size per visual sensation between contrast
ratio group by individual
101
5.6 Comparisons of Pupil Sizes Between Different Subject Groups
Previous studies indicate that eye color and myopic condition are two significant human
factor which should be taken into consideration when we studied on the influential factors
of visual sensation. Therefore, in this section, group comparisons were made by iris color
and myopic condition to see whether or not these two influential factors should be taken
into consideration in our study.
5.6.1 Iris Color
Iris color has been verified by various current studies that it is one of significant human
factors which impact occupant’s visual sensation. Recent studies have also demonstrated
that there are significant change rate differences in the visual sensations of blue eyes and
brown eyes in the same lighting environment.
Figure 5.18 Interval plot of comparisons of overall standardized pupil size per visual
sensation between eye color groups (first round, CR1:1).
102
Regardless the iris color, both groups showed bigger standardized pupil sizes at the darker
visual sensation and smaller pupil sizes at the brighter sensation. Overall, the blue group
showed a clear and distinct decreased stepwise pattern without any overlap where
standardized pupil size decreased as the overall visual sensation rose. However, even
though the brown eyes showed a pattern similar to that of the blue eyes group, there was
more overlapped pattern in the brown group.
In addition, the most compacted cubic can always be observed at neutral sensation. This
verified the conclusion in the section 5.4 that people have relative stable pupil size change
rate at neutral sensation and also provided support for pupil size standardization process.
Figure 5.19 Interval plot of comparisons of overall standardized pupil size per visual
sensation between eye color groups (second round, CR1:1).
The ANOVA test showed a p-value of 0.000 which defended significant differences of pupil
size change ratio in each iris color group. The average change ratio of pupil size had a 5.5%
103
plunge in the blue iris group and 13% drop in the brown iris group compared to neutral
level at dark. Meanwhile, for bright sensation, the average change ratio of pupil size had a
-4% plunge in the blue iris group and -5% drop in the brown iris group compared to neutral
level. These statistical results support the idea that the individual's visual perception can
matched with his or her pupil size change ratio, or in other words, standardized pupil size
across the test subject. The statistical analysis results of iris color group added evidence to
this idea.
5.6.2 Myopic Condition
Although there are many causes of myopia, the mainstream generally recognized one of
the main causes is that long-term inappropriate use of the eye that resulting in ciliary
muscle spasm. For this reason, myopia can be considered to have a significant impact on
pupil size regulation and visual perception as well. Recent studies have also demonstrated
that there are significant differences in the visual sensations of myopic and non-myopic
subjects in the same lighting environment.(Guido Maiello, 2017)
Myopia has often been classified into low, moderate and high myopia based on the
refractive error present. Low myopia would be from 0.25D (25 degrees) to –3.0D
(300 degrees); Moderate myopia would be from –3.0D (300 degrees) to –6.0D (600
degrees); High myopia would be from –6.0D (600 degrees) and upwards. In this study, half
of the subjects were myopic, two of them were high myopic, four of them were moderate
myopia, and the rest of them were low myopia.
104
Figure 5.20 Interval plot of comparisons of overall standardized pupil size per visual
sensation between myopic or non-myopic groups (first round, CR1:1).
Subjects’ myopic condition is also selected as a parameter for group analysis. For the
subject who wears glasses, the pupilometer provides an auxiliary frame which helps to
attach the tracking camera to the subject’s own glasses from the original goggle frame.
105
Figure 5.21 Interval plot of comparisons of overall standardized pupil size per visual
sensation between myopic or non-myopic groups (second round, CR1:1).
The ANOVA test showed a p-value of 0.000 which defended significant differences of pupil
size change ratio in the myopic group. The average change ratio of pupil size had a 3.8%
plunge in the myopic group and 9.7% drop in the non-myopic group compared to the
neutral level at dark. Meanwhile, for bright sensation, the average change ratio of pupil
size had a -5% plunge in the myopic group and -5.4% drop in the non-myopic group
compared to neutral level. These statistical results support the idea that the individual's
visual perception can match with his or her pupil size change ratio, or in other words,
standardized pupil size across the test subject. The statistical analysis results of the myopic
group added evidence to this idea.
106
CHAPTER 6: CONCLUSIONS
The intent of this study was using pupils to assess the possibilities for visual perception,
along with discussing their potential applications to future illuminance control operations.
The illumination condition and parameters considered in this study were divided into
illuminance level, contrast ratio, and unified glare rating. This study selected human beings
as research participants and was supplemented with survey questionnaires as a research
methodology. This helped investigate the connection between pupil size and illumination
conditions while also uncovering other potential influences. The school’s architecture
department provided an underground laboratory with a controllable indoor environment,
and the laboratory staff set up the experiment based on the experimental design and
requirements. The experiment was completed for a total of two rounds, and each round had
ten volunteer participants. These volunteers were recruited through email and flyers. They
were randomly sampled based on their ability to rule out the influence of subjective factors
and on maintaining the objectivity of the sample selection. Each experiment involved only
one participant, with an average continuous testing time of 1 hour and 30 minutes. During
the experiment process, the participants were required to complete a computer-based task
under different illuminance conditions. They were also required to complete a
questionnaire. The visual reactions of their pupil sizes to different illuminance conditions
were also collected. The experiment participants were required to wear a pupilometer
throughout the investigation process. This apparatus used a 30Hz frequency to
automatically record pupil data and generate a document. Illuminance control circuits also
automatically detected and stored data on illuminance parameters. Besides investigating
the connection between pupil size and illumination conditions, this study also looked into
the influence of objective physiological factors on pupil size. The participants were divided
into different groups based on physiological characteristics. The differences between these
groups were also summarized into important discoveries.
107
6.1 Illuminance Parameters, Visual Perception and Pupil Size
Illuminance level, contrast ratio and unified glare rating were chosen as the three main
illuminance parameters, along with the participants’ visual perception and pupil size. These
reflected the illuminance conditions during the experiment process, along with subjective
visual perception and objective physiological responses. The first round of the experiment
was comprised of three small experiments. Each small experiment was arranged with one
fixed contrast ratio, divided into 1:1, 1:4.25, and 6.4:1. At the same time, each smaller
experiment included ten illuminance levels with a range of 50 lux to 1400 lux, with
intervals of 150 lux. As a subjective vision measurement strategy, a 7-point-scale
questionnaire was administered to participants to survey their perception of each
illuminance level. The second round of the experiment was the same as the first one, but
the adjusted order of the ten illuminance levels was from light to dark and from 1400 lux
to 50 lux. Using a comprehensive pupilometer to collect data on pupil size can analyze and
summarize illuminance level and the connection between contrast ratio, unified glare rating,
visual perception, and pupil size.
Standardized pupil size is distributed in each kind of significant difference in visual
perception, and this is consistent throughout each of the two rounds of experiments. There
was at least a roughly 4% difference between visual perception and pupil size.
In terms of the different adjusted orders of illuminance from light to dark and of dark to
light, brightness was the most prevalent form of visual perception. During the two rounds
of experiments, the marker of the participants’ perception of brightness was divided into
65%-69% of the total amount of markers. Contrasting the two rounds of experiments, the
intervals between each area of illuminance level that corresponds to visual perception were
extremely close. Furthermore, the highest illuminance level value marked by participants
within each interval was the same. During both of the two rounds of experiments, most of
the participants marked a 200 lux illuminance level as neutral visual perception. As one
108
can see, the illuminance of the light-dark adjusted order within the 50 lux to 1400 lux range
had a minimal influence on participants’ visual perception.
The experiment also adopted gradient pupil size to research the rate of change of pupils
within identical time intervals. During the experiment, the illuminance level was based on
an interval adjustment of 150 lux due to the illuminance level’s increased linearity. At the
same time, the pupil shrinkage appeared to be nonlinear. Its rate of change was also
relatively stable as it neared a neutral visual perception. When it was in the bright,
extremely bright, dark, and extremely dark visual perception intervals, the rate of change
increased rapidly. At this time, the inflection point was clearly visible.
The contrast ratios included in the experiment were 1:1, 1:4.25, and 6.4:1. Under identical
illuminance conditions, different contrast ratio groups affected the same visual perception
of pupil size in different ways and with statistical significance. This shows that contrast
ratio is directly related to visual perception.
The experiment also researched the connection between unified glare rating and visual
perception. It proved that there is no apparent influence of unified glare rating on visual
perception. What can be confirmed is that pupil size can be regarded as an optimal
indicator of illuminance conditions and that pupil size has potential applications in
illuminance control.
6.2 Pupil Size and Physiological Characteristics
Based on previous scholars’ research conclusions, this experiment collected participants’
physiological characteristics such as age, gender, ethnicity, iris color, and nearsightedness.
It then selected these indexes as subjects for analyzing physiological characteristics. The
comparison of the standardized pupil size between each visual perception group (eye color,
age, nearsightedness and gender) was due to the distribution of the standardized pupil size
109
for the differences between participants’ physiological characteristics. According to the
conclusions of previous studies, the influence of iris color and nearsightedness on changes
in pupil size is relatively significant. This experiment divides the sample into groups based
on these two indices. The results prove that iris color and nearsightedness significantly
influence changes in pupil size. The average range of change in the standardized pupil size
of participants with brown eyes was 4.2% less than those with blue eyes, and the average
pupil size of the myopic group was 4.5% less than the non-myopic one. During follow-up
studies, iris color and nearsightedness should be considered as important physiological
characteristics when assessing visual perception.
6.3 Applications for Physical Models and Experimental Data
Current experimental results can be used to operate and develop a prototype for an
individual user’s eye-pupil size data-driven adaptive lighting environmental control
module. This module will activate energy conservation during the operation of a building’s
illuminance system along with creating the environmental conditions for economical and
efficient vision control. This prototype can, in an office workstation environment, use a
user’s pupil expansion and contraction function and apply it to business purposes. This can
involves a low-cost interpupillary gauge that perceives participants’ pupil changes and
adjusts according to the surrounding light. This application can also potentially conserve
resources. Future studies can include questionnaires to follow up on data such as research
participants’ attitude towards effectiveness of adjustments and their level of satisfaction
and larger and more diverse subject pools.
According to the experimental conclusions, before the process was implemented, the
participants’ physiological data was collected. This was due to the conclusions of previous
experiments, as it was understood that different characteristics influenced the same
illuminance of pupil size change pattern in different ways. Collecting these physiological
110
characteristics made it easier to follow up and focus on different physiological
characteristics using different calculation methods.
As one of this experiment’s observed conclusions, 200 lux was within neutral perception
and chosen by the highest number of participants. Therefore, it was used as the baseline
for the illuminance level. Excessive illuminance adjustment steps would reduce the
significance of the changes in illuminance. As a result, as a prototype, the process was set
up with a total of five illuminance adjustment levels. Before starting to operate the system,
adjustments were used to ensure that an additional objective for illuminance would suit its
level of comfort. At the same time, the system would seize this time’s pupil size as a
baseline. When the pupil size was equivalent to the baseline, the objective light’s
illuminance level also reached the baseline, or 200 lux. In conclusion, there was a
difference of at least 4% between each visual perception interval and the pupil size. This
was used to calculate the suitable illuminance level that the system should display.
111
CHAPTER 7 FUTURE WORK
7.1 Possible Improvements In Terms Of Participants
The sample size of 20 participants is supposed to be sufficient for an effective analysis;
however, for research involving human subjects, the results could be accessed more stably
and precisely if the sample size is larger. That is to say, with a larger sample size, the
statistical result will be more reliable with a narrower range of confidence interval, which
is conductive to producing automatic control logic for lighting in the future.
Moreover, in this experiment, the diversity of participants could be balanced and achieved.
It is optimal to equally divide participants into groups based on physiological features, so
as to minimize potential negative impact resulted from uneven sample size in control
groups. Most participants of this experiment are undergraduate and graduate students at
similar ages, since this study was carried out on campus. Hence, it could be better if more
age groups, such as old people and middle-aged people, can be involved. Besides, in order
to diversify the category of eye color, it is feasible to invite more blue-eye people to take
part in this experiment. People with myopia ought to be taken into account, since different
degrees of glasses may also lead to various behaviors.
7.2 Possible Improvements In Terms Of Software and Hardware
A large investment has been made for researches on software, device and sensors. For the
purpose of task monitoring and executing, two computers and thirty LED lamps are used;
and a pupilometer is adopted to track pupil size. Nevertheless, there is still room for
improvements.
The type of pupilometer used in this experiment needs a cable to transfer the signal to a
laptop. The wired pupilometer somewhat interfered with the comfort of participants. Some
participants complained that the dragging of cable distracted their attention on the
112
computer-based task. Even if cables were fixed under the assistance of tools like clamps,
this issue still existed. Hence, in future experiment, the application of wireless pupilometer
could better serve those participants.
Participants sometimes also complained of the problem of glare when the light source
irradiated the reflective glass of the pupilometer. This issue was temporarily solved by
using a small piece of paperboard or a cap to cover the pupilometer’s top. Nevertheless,
some views over the pupilometer were shaded, which was not severe but may unfavorably
influence the experiment. This issue may be ascribed to the position of LED lamps; to be
specific, the distance between LED lamps and heads of participants was comparatively
smaller than that in traditional office. Hence, the light source needs to be designed better
in the future.
At present, a number of devices put on table may distract the attention of participants. Two
laptops—one for the illuminance meter and the other for the pupilometer, were adopted in
this experiment. Relevant software was installed in both laptops to collect data. However,
the process of collection and monitoring would be easier if the two programs of data
collection were integrated.
In addition, the major equipment to realize luminance contrast currently is the liquid-crystal
color TV . As it has practical use in daily life, the luminance contrast range created by the
liquid-crystal color TV is limited. In other words, the existing liquid-crystal color TV does
not meet the condition of extreme brightness when necessary. In future experiments, this
issue should be addressed by applying more professional equipment for background
luminance.
7.3 Development of Strategy
This study summarized the calculated change of pupil size based on various conditions.
113
Nonetheless, detailed formula and control strategies should be developed in the future
based on the data collected in this experiment. In order to ensure the effectiveness of any
proposed strategy, it should be validated through relevant research. Moreover, the sample
size and diversity should be further increased to guarantee the representativeness of
experimental data.
Based on an elaborate formula and control strategy, software can be developed and
programmed to control lighting devices or to be used for self-designed physical
applications. Nevertheless, it is quite necessary to test all those physical applications and
software. Test runs are of great importance in real projects, and both long-term and short-
term test run should be carried out.
This strategy could be applied to other purposes under further relevant studies instead of
being limited to the control of office lighting. It is believed that the control strategy based
on pupil size could be extensively used in this industry. More relevant studies are highly
valued and recommended.
114
Bibliography
“6 Tips for Taking Better Macro Photos with the iPhone Camera.” 2012.
http://osxdaily.com/2012/10/07/tips-better-iphone-macro-photos/.
Berman, B Y S A M. 2000. “The Coming Revolution in Lighting Practice.”
Choi, Joon-Ho, Vivian Loftness, and Azizan Aziz. 2012. “Post-Occupancy Evaluation of 20 Office
Buildings as Basis for Future IEQ Standards and Guidelines.” Energy and Buildings 46
(March): 167–175. doi:10.1016/j.enbuild.2011.08.009.
http://linkinghub.elsevier.com/retrieve/pii/S0378778811003434.
Campbell, B Y F W, and D G Green. 1965. “Optical and Retinal Factors Affecting Visual
Resolution.” J.Physiol 181: 576–93.
Campbell, B Y F W, and R W Gubisch. 1966. “Optical Quality of The Human Eye.” J.Physiol
186: 558–78.
Campbell, F.W. 1957. “The Depth of Field of the Human Eye.” Optica Acta 4: 157–64.
Deneen, S. 2004. Seeing the Light.
Dilouie, Craig. 2003. “Lighting and Productivity: Missing Link Found ?” Architectural Lighting
18,6: 39.
Eastman, A.A., and J.F. McNelis. 1963. “An Evaluation of Sodium, Mercury and Filament Lighting
for Roadways.” Illum Eng.
Energy Information Administration, 2012, Commercial Building Energy Consumption Survey: Energy
Usage Summary, Table 1
https://www.eia.gov/energyexplained/?page=us_energy_commercial
Energy Information Administration (2015). Total Primary Energy Consumption, 1980-2012.
International Energy Statistics.
“Event Horizon Volume 3 6 Aging Eyes and Pupil Size.” 2014. Accessed June 22.
http://amateurastronomy.org/Events/EH361.html.
Ferguson, H.M., and W.R. Stevens. 1956. “Relative Brightness of Colored Light Sources.” Trans IES.
115
Gauri Shankar Shrestha, Fathimath Nestha Mohamed, Dev Narayan Shah,2011, “Visual problems
among video display terminal (VDT) users in Nepal” J Optom. 2011 Apr; 4(2): 56–62.
Hedge, Alan, William R Sims Jr., and Franklin D Becker. 1990. “CUergo: Cornell University
Lighting Research Study.” http://ergo.human.cornell.edu/lighting/lilstudy/lilstudy.htm.
International Energy Agency (IEA), (2014). World Energy Outlook 2014 Factsheet.
World Energy Outlook 2014.
Irlen, Helen. 1991. “Reading By The Colors Overcoming Dyslexia and Other Reading
Disabilities Through the Irlen Method” 27 (Chapter 3).
Leibowitz, H. 1952. “The Effect of Pupil Size on Visual Acuity for Photometrically Equated Test
Fields at Various Levels of Luminance.” Journal of the Optical Society of America 42 (6):
416–22. http://www.ncbi.nlm.nih.gov/pubmed/14939110.
“Light+Architecture: Lighting Simulation Software: Autodesk 3DS MAX.” 2011.
http://blog.lightingvanguard.com/2011/08/lighting-simulation-software-autodesk.html.
“Lighting Analysis in BIM | Sustainability Workshop.” 2014. Accessed June 22.
http://sustainabilityworkshop.autodesk.com/buildings/lighting-analysis-bim.
Luckiesh, M., and F.K. Moss. 1934. “Seeing in Sodium Vapor Light.” J Opt Soc Am.
Manav, Banu. 2007. “An Experimental Study on the Appraisal of the Visual Environment at Offices
in Relation to Colour Temperature and Illuminance.” Building and Environment 42 (2)
(February): 979–983. doi:10.1016/j.buildenv.2005.10.022.
http://linkinghub.elsevier.com/retrieve/pii/S036013230500452X.
Mojtaba Navvab. 2002. “Visual Acuity Depends on the Color Temperature of the Surround
Lighting” LEUKOS The Journal of the Illuminating Engineering Society of North America
31(1):70-84
Oi, Naoyuki, and Hironobu Takahashi. “PREFERRED COMBINATIONS BETWEEN
ILLUMINANCE AND COLOR TEMPERATURE IN SEVERAL SETTINGS FOR DAILY
LIVING ACTIVITIES”: 2–5.
Sauro, Jeff. 2014. “Should You Use 5 or 7 Point Scales?” Accessed March 12.
https://www.measuringusability.com/blog/scale-points.php.
116
Sivaji, Ashok, Sajidah Shopian, Zulkifle Mohd Nor, Ngip-Khean Chuan, and Shamsul Bahri. 2013.
“Lighting Does Matter: Preliminary Assessment on Office Workers.” Procedia - Social and
Behavioral Sciences 97 (November): 638–647. doi:10.1016/j.sbspro.2013.10.283.
http://linkinghub.elsevier.com/retrieve/pii/S1877042813037282.
Asset Metadata
Creator
Lin, Xiaoxin (author)
Core Title
Human-building integration: Investigation of human eye pupil sizes as a measure of visual sensation in the workstation environment
Contributor
Electronically uploaded by the author
(provenance)
School
School of Architecture
Degree
Master of Building Science
Degree Program
Building Science
Publication Date
07/26/2020
Defense Date
04/26/2018
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
bio-signal,human-building integration,indoor environment,luminance ratios,OAI-PMH Harvest,unified glare rating,visual sensation
Format
application/pdf
(imt)
Language
English
Advisor
Choi, Joon-Ho (
committee chair
), Narayanan, Shrikanth (
committee member
), Schiler, Marc (
committee member
)
Creator Email
linxiaoxin1986@gmail.com,xiaoxinl@usc.edu
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c89-26977
Unique identifier
UC11671509
Identifier
etd-LinXiaoxin-6486.pdf (filename),usctheses-c89-26977 (legacy record id)
Legacy Identifier
etd-LinXiaoxin-6486.pdf
Dmrecord
26977
Document Type
Thesis
Format
application/pdf (imt)
Rights
Lin, Xiaoxin
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Abstract (if available)
Abstract
Lighting is the most crucial factor impacting an occupant's visual comfort in a building environment. However, most prevailing current lighting guidelines deriving from empirical values are designed primarily for paper-based tasks, rather than computer-based. In many cases, present guidelines do not meet the needs for a user's new task types. Above all, existing technical tools also have limited functionality to evaluate a user's real-time visual perception, which can be applied as an input to control a building lighting system. ❧ This research estimated each individual participant's visual sensations by analyzing pupil sizes and their change patterns under different lighting conditions since the human body has the physiological regulation ability which naturally minimizes the adverse effects of the surrounding environment. ❧ This study adopted a series of human subject experiments which were performed in an environmental chamber of USC. Based on a computer-based task which is most commonly performed in current offices, various ranges of ambient lighting parameters, including luminance (cd/m2), illuminance (lux), luminance contrast ratio, and Unified Glare Rating (UGR), were generated and controlled while each subject's pupil sizes were recorded. The experimental result data were statistically analyzed to identify a relationship between human visual sensations, lighting parameters, and also pupil sizes by ethnic origin and myopia condition. ❧ The research outcomes showed the potential use of pupil sizes for estimating an individual's visual sensation and confirmed the principle as an applicable technology to integrate an environmental design and control system with the help of a real-time sensing device, such as a wearable sensor.
Tags
bio-signal
human-building integration
indoor environment
luminance ratios
unified glare rating
visual sensation
Linked assets
University of Southern California Dissertations and Theses