Research of accommodative microfluctuations
caused by visual fatigue based on liquid
crystal and laser displays
Wei-De Jeng,1 Yuan Ouyang,2 Ting-Wei Huang,1 Jeng-Ren Duann,3
Jin-Chern Chiou,1,3 Yu-Shun Tang,1 and Mang Ou-Yang1,*
1
Department of Electrical and Computer Engineering, National Chiao Tung University, Hsinchu 30010, Taiwan
2
Department of Electrical Engineering, Chang Gung University, Kwei-Shan, Tao-Yuan 33302, Taiwan
3
Biomedical Engineering R&D Center, China Medical University, Taichung 40402, Taiwan
*Corresponding author: oym@cn.nctu.edu.tw
Received 5 May 2014; revised 8 July 2014; accepted 13 July 2014;
posted 14 July 2014 (Doc. ID 211202); published 14 August 2014
Different levels of visual fatigue in the human eye depend on different color-formation methods and
image quality. This paper uses the high-frequency component of the spectral power of accommodative microfluctuations as a major objective indicator for analyzing the effects of visual fatigue based on various
displays, such as color-formation displays and 3D displays. Also, a questionnaire is used as a subjective
indicator. The results are that 3D videos cause greater visual fatigue than 2D videos (p < 0.001), the
shutter-type 3D display causes visual fatigue more than the polarized type (p ˆ 0.012), the display of
the time-sharing method causes greater visual fatigue than the spatial-formation method (p ˆ 0.008),
and there is no significance between various light source modules of displays (p ˆ 0.162). In general, people with normal color discrimination have more visual fatigue than those with good color discrimination
(p < 0.001). Therefore, this paper uses the high-frequency component of accommodative microfluctuations
to evaluate the physiological stress or strain by overexerting the visual system, and can compare the level
of visual fatigue between various displays. © 2014 Optical Society of America
OCIS codes: (120.2040) Displays; (330.1720) Color vision; (330.4595) Optical effects on vision;
(330.7310) Vision.
http://dx.doi.org/10.1364/AO.53.000H76
1. Introduction
Currently, visual fatigue is a common affliction because people frequently view displays for a long time.
As these stimulate the eyes continuously, they also
cause a great number of visual problems. Therefore,
it is worth discussing the effects of such displays on
people’s vision [1].
From a visual standpoint, displays can be divided
into three parts: color-formation methods, 3D technologies, and different light sources. In a study on
the effects of visual fatigue from various colors and
image appearances, it is necessary to find a stressfree way for people to view things. Color-formation
methods involve time sharing and spatial formation [2], 3D display technology has shutter and polarized type, and the light source module usually
includes a light-emitting diode (LED) and a laser diode (LD). To date, the color breakup from the timesharing method has been known to cause visual
fatigue in the human eye according to [3–9], and it
is common for 3D displays to make people feel tired.
Therefore, the main focus of this research is to investigate the degree of visual fatigue based on various
displays.
To solve these problems, it is necessary to find a
suitable indicator for evaluating the effects of various
displays on the human eye. There are various indicators for evaluating visual fatigue. A subjective
method such as a questionnaire could be used, and
some objective methods, such as critical fusion frequency (CFF) [10], are suitable in certain cases. This
study proposes to evaluate visual fatigue through
accommodative microfluctuations of the ciliary body.
There is now a similar evaluation method in medical
research on visual fatigue called the high-frequency
component (HFC) of accommodative microfluctuations. For instance, Gray et al. [11] used the lowfrequency component (LFC) and HFC to analyze
visual variations after using visual display terminals
(VDTs). The HFC’s spectral power was analyzed by
Kajita et al. [12]. It showed that visual fatigue could
indeed be measured and quantified. This method can
effectively and objectively determine whether subjects are suffering from visual fatigue. Therefore, this
research uses it as an indicator for visual fatigue.
In visual fatigue research, there are various and
extensive indicators and measuring methods, which
can be divided into five types: (1) the measurement of
the oculomotor systems including eye movement
velocity, accommodation power, convergence, viewing
distance, pupil diameter, and blinking; (2) the measurement of visual acuity including visual acuity and
CFF; (3) the measurement of performance of visual
tasks, i.e., recognition speed and error detection rate;
(4) the report of asthenopia symptoms; and (5) brain
activity measurements such as functional magnetic
resonance imaging (FMRI), magneto encephalography (MEG), and electroencephalography (EEG) for
observing neural activity affected by visual fatigue,
both temporally (MEG and EEG) and spatially
(FMRI) [13–18].
There are objective or subjective indicators according to Chi and Lin [10], such as accommodation
power, pupil diameter, visual acuity, eye movement
velocity, CFF, the subjective rating of visual fatigue,
and visual task performance. The subjective rating
for visual fatigue is an indicator related to a
judgment on the subjects’ level of visual fatigue; it
will have more inaccuracies while subjects were
disturbed by the response of the visual cortex
neural activity. Overall, not all of the visual fatigue
Table 1.
Indicators
Accommodation power
Pupil diameter
Visual acuity
Comparisons between Seven Indicators for Evaluating the Level of Visual Fatigue [10]
Advantages
High sensitivity to near work
High sensitivity to dynamic condition
Suitability to evaluation of whole visual
system
High sensitivity to dynamic condition
Eye movement velocity
Critical fusion frequency
Subjective rating of visual
fatigue
Visual task performance
measurement indicators have high sensitivity in
every experiment. There are different restrictions or
conditions in the different experiments, as shown in
Table 1. The indicators for evaluating visual fatigue,
accommodation power, pupil diameter, visual acuity,
eye movement velocity, CFF, subjective rating of
visual fatigue, visual task performance, and brain
activity measurements have their own limitations.
For instance, accommodation power is best used to
evaluate visual fatigue because the lens curvature
becomes larger in close work and fatigue occurs over
long periods of time. Work station illumination is different from that of a target or dynamic information,
and the iris’s sphincter muscle, as a pupil constrictor,
has to control the amount of light entering the eyes.
Hence, the pupil’s diameter is a suitable evaluator
under these conditions. Moreover, the subjective rating does depend on the participants’ psychological
and physiological feelings, and hence has good face
validity yet large intersubject variations. Therefore
a large number of participants is required to yield
reliable results from the subjective ratings.
An appropriate evaluation indicator should be
chosen to investigate the relationship between the
color-formation methods, 3D technologies, different
light source modules, and visual fatigue. This indicator needs to satisfy every condition to compare these
relationships. This study uses dynamic information
as the experimental tool since displays are used to
show the dynamics of daily life. The moving frames
stimulate eyesight. Moreover, it is better to use
multiple visibility ranges than others that just
explore the influence of different ranges at short
distances. From the indicators mentioned in Table 1,
accommodation and CFF are highly sensitive at
short distances and low brightness contrast, respectively; both are unsuitable. Second, the pupil’s diameter is so easily influenced by exterior stimuli that it
is difficult to control variations. The speed of eye
movements can be adopted in a dynamic environment, but existing measurement methods are limited. Some instruments cause the eyes discomfort,
and the measurement of movement time is inaccurate. Because of such difficulties, this study uses
an advanced evaluation method of visual fatigue,
which is accommodative microfluctuations. The
accommodative microfluctuations can reflect the
High sensitivity to contrast work in low
brightness
High face validity and easy application
Direct and indirect show of visual fatigue
Disadvantages
Long time for stimulation
Unsuitability for static condition and many limitations
Long time for stimulation
Unsuitability for static condition and limitations of
device
Long time for stimulation
Low objectivity
Many limitations to apply
Table 2.
Abbreviation
Displays and Operation Modes of the Display Used
in Experiment
Luminance
…cd∕m2 †
Screen Size
(inches)
S3D
172.4
40
S2D
172.4
40
P3D
234.9
42
P2D
234.9
42
LP
250.7
44
Display
Type
Shutter 3D LCD
under 3D mode
Shutter 3D LCD
under 2D mode
Polarized 3D LCD
under 3D mode
Polarized 3D LCD
under 2D mode
Laser projector
ciliary body’s response. Because this indicator is not
restricted by the conditions and relates to the subjects’ physiology and the ciliary body microfluctuations, it can be more accurate when evaluating
visual fatigue. Finally, this research combines the
results of questionnaire tests to compare the relation
between subjective and objective rating methods.
2. Accommodative Microfluctuations
The eye lens’s curvature changes its dioptric ability
to maintain a clear image of objects as the distance
varies. This self-adjusting mechanism on the optical
system is called accommodation, which varies from
0.01 m (10 D) to infinity (0 D) for normal and young
eyes. The ciliary body around the lens controls the
curvature of the lens by contracting and relaxing,
and has a large number of suspensor ligaments
connected to the lens. Theoretically, the ciliary body
should be in static state when looking at an object,
but actually it adjusts repeatedly instead of adjusting immediately to a suitable curvature during the
accommodation process. As a result, it becomes unstable and fluctuates when focusing on an object,
and the fluctuation frequency is around a few hertz.
These are called accommodative microfluctuations
[10,11,18–24]. As people with normal vision look
at a close object, the activity of the ciliary body is
relatively low when compared to viewing a far object.
When people with visual fatigue look at a far object,
the activity of the ciliary body is significantly higher
compared to normal vision, and remains high while
looking at short distances, but it is not obvious when
compared to people with normal vision [11,20]. When
reading at short distances, people with myopia
exhibit a significant increase in the power of accommodative microfluctuations [25].
The behavior of accommodative microfluctuations
is complex, without rules, and nonlinear in time [1];
however, regular patterns exist while transferring
the waveform into the frequency domain. According
to the waveform, accommodative microfluctuations
consist of two components: LFC and HFC; the former
is defined below 0.6 Hz, and the latter between 1.0
and 2.3 Hz [11,12,18,21,22]. The following section
will introduce the relation between accommodative
microfluctuations and these two components.
Accommodative microfluctuations may be affected
by the distance of the target [20,26–28], the pupil
diameter [24,29–31], the form of the target [32,33],
the luminance of the target [11,34,35], the eye’s
age, astigmatism [36], visual fatigue [12,20,37,38],
biomonocular observation of the target [30,37], and
artifacts such as cardiopulmonary signals [39–41]
or other rhythmical physiological systems. Neurological control also affects the LFC’s wavelength, and
arterial signals correlate highly to the HFC [11,21].
However, the effect by the pupil’s diameter is the most
obvious.
The pupil changes with light, and when the pupil’s
diameter is smaller, the HFC fluctuations are imperceptible, although the LFC increases; when the
diameter increases, the HFC fluctuations become
obvious and the LFC decreases [21].
Geacintov and Peavler (1974) reported that there
is a connection between pupil instability and visual
fatigue [42]. Some recent research has studied eye
variations and visual fatigue after viewing VDTs;
it indicates the close connection between the pupil’s
variations and accommodative microfluctuations
[11]. It showed that patients with asthenopia can
indeed be diagnosed from changes in the HFC
[12]. Suzuki et al. (2001) tried using a color code to
show the position of targets, the accommodative
response amplitude, and the HFC value as figures
[21]. While viewing a distant stable target, the result
shows that a normal subject’s HFC is about 50–60,
labeled in green, while a subject with asthenopia
is above 60, labeled in red. After combining these results, it is concluded that the ciliary body’s tension is
low when viewing far objects, so its variation is large
if visual fatigue occurs, and slight with short distances. Thus, the ciliary body’s tension is recognized by
the HFC variations, and subjects are assessed to see
whether they suffer from visual fatigue.
As mentioned before, there is a high correlation
between the HFC and cardiopulmonary signals;
therefore, eyes suffering from visual fatigue can be
evaluated by the HFC variations because physiological aspects influence cardiopulmonary signals. In
addition to measuring accommodative microfluctuations and calculating the HFC, the device
can be used to measure accommodation power, which
has been more functional than optometers in the past.
3. Methodology and Experiments
In order to research the effects on the human eye
by various displays, three displays of different types
were used in the experiment: a shutter-type
3D liquid crystal display (LCD), a polarized-type 3D
LCD, which are using both an LED as a backlight
module and the color-formation method in spatial
formation. A laser projector uses a mix of LED and
LD for the light source module; the color-formation
method is a time-sharing method. Also, both of the
3D LCDs can change to the 2D display mode. The
resolution of the 3D LCD is full HD (1920 × 1080),
and the laser projector is XGA (1024 × 768). Table 2
shows the display’s luminance specifications under
different display types; all the luminance values of
display types were measured from a white screen.
In order to express the display type in a simple
way, the abbreviations are defined under different
display types.
An auto refract-keratometer Righton Speedy-K
can record a subject’s ciliary muscle microfluctuations through built-in targets. Subjects should gaze
steadily at a target, which varies from ‡ 0.5 to
−3.0 D with 0.5 D step, for a total of eight targets.
The software “MF-1” is then applied to analyze the
data, and calculations are made by the fast Fourier
transform (FFT). The mean spectral power is
integrated into the frequency domain between 1
and 2.3 Hz, and used to evaluate the research’s
visual fatigue level because it indicates ciliary body
fluctuations. In total, there are six HFC data for each
target.
The Farnsworth–Munsell 100-Hue Test is a direct
device used to examine chromatic discrimination,
and is used by this research as a basis for grouping
subjects according to their chromatic discrimination
abilities. In the standard reference, an error score
less than 20 is good in chromatic discrimination,
between 20 and 100 is normal, and more than 100
is bad [43]. In this study, the subjects will be grouped
into different chromatic discrimination abilities
according to the standard reference.
Figure 1 shows the flow chart for the experiment;
the whole experiment is carried out in a dark room.
Before viewing video subjects were asked about such
discomforts as eyestrain, headache, or body stiffness;
then they rested for 5 min to prevent affecting the
test from prior fatigue. In the beginning of the experiment, all subjects participated two times in the hue
test under a D65 light source and then divided into
two groups: good chromatic discrimination and
normal chromatic discrimination.
After finishing the hue test, subjects were asked to
take a 5 min break to relax their eyes, and then the
dipoters of the left eye were measured by the auto
refract-keratometer for 1 min, 45 s to get the HFC
data; then the subjects filled out the questionnaire
as shown in Table 3 [44,45]. The subject gave a score
for each symptom before and after viewing videos;
scores 1 to 5 indicate no feeling, slightly, medium,
deep, and serious, respectively. Upon completing
the questionnaire on 15 symptoms, subjects viewed
the video sitting on a chair at about a 2 m distance
from the screen for 15 min. This experiment chose
two videos, and each video was separated into four
parts and randomly displayed on a shutter 3D LCD
under 2D and 3D mode, a polarized 3D LCD under
2D and 3D mode, and a laser projector. After viewing
the videos, the eye measurements were taken again,
and the questionnaire was filled in immediately
afterwards. In order to prevent a large number
of subjects from suffering from an accumulation of
tiredness, each person was tested in only one mode
per day, and each person was tested four times in
Fig. 1. Flow chart of experimental process.
each mode; there are five modes in total in the experiments. Hence, it takes 20 days to finish a round for
each person. Finally, after all the data are collated
and analyzed, the HFC is used as the major indicator.
Table 3.
Questionnaire for Evaluating the Level of Visual Fatigue by
Different Symptoms
Questions
1. Do you feel eyestrain?eyestrain
2. Do your eyes feel dry?dry eyes
3. Do you feel the environment too bright
is too bright?
4. Does your eyelid twitch?eyelid twitching
5. Do you feel stress infeeling of pressure
your eyes?in the eyes
6. Do you feel ache behindache behind the
the eyes?eyes
7. Do you feel everything isblurred vision
blurred?
8. Do you have a headache?headache
9. Does your head feel heavy? head feels heavy
10. Do you have a headachehead hurts when
when shaking your head?shaken
11. Do you feel dazed?dazed feeling
12. Do you feel irritated?irritated feeling
13. Do your shoulders feel stiff? stiff shoulders
14. Do you want to sleep?sleepy feeling
15. Do you feel it is difficult to difficulty
pay attention?concentrating
Symptoms
Before After
Table 5.
The questionnaire is used as a reference indicator for
the analysis and discussion.
3D
2D
LP
2D
S3D
P3D
Now the HFC is measured before and after viewing
videos, and the level of visual fatigue can be defined
by ΔHFC. The ΔHFC means the difference between
the HFC, which is measured before and after viewing
videos; it means ΔHFC ˆ HFCafter −HFCbefore . Each
subject is tested four times in each mode. Each time
is measured six HFCs at each target, and there are
eight targets in each measurement; thus, each subject has 6 × 8 × 4 ˆ 192 data for each mode.
All the data from this procedure are labeled with
different variables, and the analysis of variance
(ANOVA) and one-tailed t-test are used to find what
effects the different color-formation methods, 3D
technologies, and lighting sources have on visual
fatigue.
There are 30 subjects in total, and the mean age
is 25.0 3.8. Different chromatic discrimination
abilities are considered to determine the relationship
between them and visual fatigue. Besides, the ranking for mean scores is applied to analysis of questionnaires for finding what effects the different
color-formation methods, 3D technologies, and lighting sources have on visual fatigue.
Table 4 shows the t-test results and compares the
level of visual fatigue between each mode. Standard
error (SE) is the standard deviation of the sample
mean. SE is computed by the sample standard
deviation divided by the square root of the sample
size. The asterisk indicates that the t-test result is
significant (p value <0.05). The results indicate that
the ΔHFC by S3D is much larger than one by S2D
(p < 0.001) and the ΔHFC by P3D is much more than
the one by P2D (p ˆ 0.035). Therefore, the subjects’
Variables (Mode)
*
t-Test Results for Each Mode
Mean ΔHFC (SE)
S3D
S2D
S3D
P3D
P3D
P2D
S2D
P2D
S3D
P2D
P3D
S2D
S3D
LP
P3D
LP
S2D
LP
P2D
LP
1.43
0.63
1.43
0.87
0.87
0.44
0.63
0.44
1.43
0.44
0.87
0.63
1.43
1.03
0.87
1.03
0.63
1.03
0.44
1.03
(0.179)
(0.167)
(0.179)
(0.172)
(0.172)
(0.165)
(0.167)
(0.165)
(0.179)
(0.165)
(0.172)
(0.167)
(0.179)
(0.171)
(0.172)
(0.171)
(0.167)
(0.171)
(0.165)
(0.171)
p Value
<0.001*
0.012*
0.035*
0.215
<0.001*
0.317
(0.124)
(0.118)
(0.171)
(0.118)
(0.179)
(0.172)
<0.001*
0.008*
0.012*
ciliary body is tenser and more fatigued when viewing the 3D videos than when viewing the 2D videos.
The subjects’ ΔHFC is higher when viewing S3D
than P3D (p ˆ 0.012), so the shutter-type display
strains the subjects’ ciliary body more than the polarized system does. There is no significant difference
between viewing S2D and P2D (p ˆ 0.215), possibly
because both of them are in the 2D mode using a spatial formation. Finally, the ΔHFC of watching LP is
significantly different to S2D (p ˆ 0.045) and P2D
(p ˆ 0.006), but not to P3D (p ˆ 0.257). The ciliary
body tension is caused more by S3D than LP
(p ˆ 0.052), but LP is more than S2D (p ˆ 0.045) and
P2D (p ˆ 0.006). The latter result indicates that time
sharing causes the human eye to experience more
fatigue than the spatial-formation method does.
Table 5 shows the t-tests results for the 3D and 2D
modes; it shows that the 3D mode causes much more
serious tension than 2D (p < 0.001). Thus, subjects
strained the ciliary body more while viewing the
3D videos than the 2D videos. The t-tests results
for the LP and LCD show that subjects have much
more tension caused by the LP (p ˆ 0.008), indicating that the time-sharing method causes more
fatigue to the ciliary body than the spatial formation
does. For the two kinds of 3D displays, subjects who
viewed 3D videos with shutter glasses had significant ciliary body tension compared to those wearing
polarized glasses (p ˆ 0.012).
Table 6 shows that the two kinds of light sources
have no significant difference when compared to
each other (p ˆ 0.162). It indicates that the LED
backlight and LED mixed with LD light source
module have the same effect on fatigue and the
ciliary body.
In order to investigate the relation between
chromatic discrimination ability and visual fatigue,
all subjects were grouped into two groups, and
analyses of chromatic discrimination in each mode
were conducted to understand how it affects visual
fatigue.
From Table 7, some first test results with the
subjects have significant differences compared to
0.052
0.257
Table 6.
0.006*
t-Test for LED Backlight and Mixed LED and LD
Light Source
Mean ΔHFC (SE)
Variables
Light sources
A significant result (p < 0.05).
1.15
0.53
1.03
0.53
1.43
0.87
p Value
A significant result (p < 0.05).
0.045*
*
Mean ΔHFC (SE)
Variables (Mode)
4. Analysis and Results
Table 4.
t-Test for 3D/2D, LP/LCD, and Shutter/Polarized
LED
LED+LD
0.84 (0.085)
1.03 (0.171)
p Value
0.162
Table 7.
Subjects
01
02
03
04
05
06
07
08
09
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Chromatic Discrimination Ability of All Subjects
Total Error
Scores
12, 08
16, 04
40, 08, 20
28, 24
48, 16, 20
36, 12
96, 40, 44
56, 20, 28
12, 24
32, 20
40, 12, 24
20, 20
20, 20
88, 88
40, 08, 20
20, 28
20, 20
28, 16
36, 32
32, 36
24, 16
08, 18
16, 24
08, 20
16, 28
72, 48, 32
12, 08
52, 36
24, 16
08, 12
Mean Error
Scores
10
10
14
26
18
24
42
24
18
26
18
20
20
88
14
24
20
22
34
34
20
13
20
14
22
40
10
44
20
10
Table 8.
Chromatic
Discrimination
good
good
good
normal
good
normal
normal
normal
good
normal
good
normal
normal
normal
good
normal
normal
normal
normal
normal
normal
good
normal
good
normal
normal
good
normal
normal
good
the second test results; this situation should use
hypothesis testing to filter out the singular data.
We compute the difference of two hue test scores (second test score minus first test score) in all normal
subjects, so the total score difference is 23 (because
seven subjects’ hue tests the first time may be wrong
due to unfamiliarity). The mean score difference is
7.74, and the standard deviation is 5.9. Then we
can find all seven subjects’ score differences are successful to reject the null hypothesis, which means if
the score of the first two tests has a significant difference, this situation has very low probability of occurrence. Hence, the subject should do the third hue
test. Also, we can find that after the subject finishes
the third test, the score difference (between the second and third tests) becomes smaller than the score
difference between the first two tests. This is why we
delete the first test score. We also asked those subjects about the first hue test, and subjects said that
because this test only allows a few seconds to finish,
they became very anxious when the time was running, so they did not perform well on the first test,
but they felt better on the second test. Finally, the
numbers of subjects are 11 and 19 in good and normal chromatic discrimination, respectively.
Table 8 shows a detailed analysis of chromatic
discrimination power. All mode means all the 3D,
2D, and LP samples were combined together. An
t-Test for Chromatic Discrimination under Each Condition
Variables
(Chromatic Discrimination)
All
Mean
ΔHFC (SE)
good
normal
good
normal
good
normal
good
normal
good
normal
good
normal
good
normal
3D
2D
S3D
P3D
LCD
p Value
<0.001*
0.64 (0.097)
1.24 (0.124)
0.87 (0.144)
1.8 (0.240)
0.56 (0.123)
0.91 (0.157)
1.15 (0.226)
1.86 (0.290)
0.65 (0.219)
1.21 (0.276)
0.36 (0.151)
0.79 (0.187)
0.69 (0.213)
1.54 (0.285)
0.005*
0.040*
0.024*
0.055
0.039*
LP
*
0.009*
A significant result (p < 0.05).
interesting consistency, found in the results above,
is that the subjects’ ΔHFC with normal chromatic
discrimination is higher than that with good chromatic discrimination power under All mode
…p < 0.001† , under the 3D mode (p ˆ 0.005), the 2D
mode (p ˆ 0.04), the shutter system (p ˆ 0.024), the
polarized system (p ˆ 0.055), LCD (p ˆ 0.039), and
the laser projector (p ˆ 0.009). It is understood that
individual differences affect the level of visual
fatigue.
Figure 2 shows the mean scores for each symptom
corresponding to each mode; different intensities in
the radial direction mean different mean scores. The
mean score for each symptom in each mode is calculated by taking the score difference from each symptom; the score difference means the score’s difference
between before and after viewing the display. Then
we sum the scores from the four different trials
5
4
6
3
7
2
8
1
-0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6
9
15
10
14
11
12
13
Fig. 2. Scores of symptoms in each mode. The outer point means
that the discomfort level of the symptom is more serious.
S3D
S2D
P3D
P2D
LP
corresponding to the 15 symptoms, then minus
15 × 2.5 ˆ 37.5 (average), which was used to compare
the effect on visual fatigue caused by different
variables.
Figure 3 shows the total ranking for each mode.
The rankings for S3D and P3D are higher than for
S2D, P2D, and LP. Then S2D, LP, and P2D are in
descending order from high to low. But there are
no differences between S3D and P3D. Figure 4 shows
the total rank in the 3D and 2D modes, the LCD and
LP, and the shutter and polarized. The total rank
in the 3D mode is higher than in the 2D mode, but
there are no differences between the LCD and LP,
or the shutter and polarized. It shows that 3D videos
cause more visual fatigue than 2D ones. The error
bar in Figs. 3 and 4 comes from the SEs in the total
rank from all participants.
Fig. 3. Total rank for mean scores of each mode.
5. Conclusions
and then divide by 4 (each mode repeats four times),
so the mean score of each symptom can be obtained.
After computing the mean score of all the symptoms,
a radar plot is used to express the analysis results by
questionnaire survey. The red line labels the highest
scores for each symptom: (1) eyestrain, (2) dry eyes,
which are much more serious and common, and (3)
too bright, (5) feeling of pressure in the eyes, (7)
blurred vision, (11) dazed feeling, (14) sleepy feeling,
and (15) difficulty concentrating, which are also
common symptoms. However, it is still hard to compare the effects of each variable. So the total ranking
of mean scores for each symptom in the modes was
used in the following analyses of the relationship
between each mode and visual fatigue.
According to Fig. 2, each mode in every symptom
can be ranked by their mean scores. The ranking for
mean scores of each symptom in each mode is ranked
5 (high) to 1 (low), which means that if a mode in the
first symptom (eyestrain) has the largest mean score,
it can get five points. Hence, after ranking all the
symptoms in each mode, the total ranking points
for each mode is derived by summing all the points
Fig. 4. Total ranking for mean scores under 3D/2D mode (white),
LCD/LP (gray), and shutter/polarized (black).
According to the results of the analysis of the ΔHFC
and questionnaire, the subjects suffer much more
ciliary body strain and visual discomfort while viewing 3D videos than 2D videos (p < 0.001). One possible reason could be that the brain has to make
more effort to adjust to the 3D variations, so it
may affect the neural processing and lead to visual
fatigue. Another reason could be that maybe the
stereo of the 3D video is not as real when compared
with natural vision. The binocular parallax that gives
the depth perception lets people feel that the stereo
effect is excessive, so the accommodation system is
overloaded and tension occurs in the ciliary body.
The subjects suffer much more ciliary body tension
and visual discomfort while viewing 3D videos with
shutter glasses than with polarized glasses (p ˆ
0.012). Because of the shutter system theory, brightness is relatively lower and the frames might glitter
so that the subjects’ ciliary body is relatively tenser.
In the marketplace, the polarized system camp
claims that their system has no disadvantages over
the shutter but the stereo is quite the same, and
market research from DisplaySearch shows that
the polarized system’s share of the market is increasing. Nevertheless, further research is needed to
confirm the above claims and conjectures.
Third, the ciliary body of subjects suffers more
fatigue and visual discomfort while viewing 2D videos
on a LCD TV than on the laser projector (LP)
(p ˆ 0.008). Taking time sharing for example, color
breakup may occur and affect people’s vision. One conclusion is that the cone and rod cells that discriminate
between color and gray levels, and sense the resolution, are discrete in the spatial domain. This is similar
to the spatial-formation method, so this method
makes the ciliary body less tense. Another conjecture
is that the ciliary body may have been stimulated
more slowly by the spatial-formation method than
the other method, during the 15 min experiment.
So, those questions are worthy of discussion.
Finally, subjects with normal chromatic discrimination power strain their ciliary body more than those
with good chromatic discrimination power under
each condition (All, p < 0.001; 3D, p ˆ 0.005; 2D,
p ˆ 0.04; shutter, 0.024; polarized, p ˆ 0.055; LCD,
p ˆ 0.039; LP, p ˆ 0.009). This indicates that individual differences do affect the levels of visual fatigue.
The reason might be that some parts or mechanisms
of the visual system in people with worse chromatic
discrimination are weaker, hence the different states
of the ciliary body. Generally, visual comfort is more
serious for subjects with normal chromatic discrimination than those with the good, and while watching
3D videos and especially wearing shutter glasses, or
watching 2D videos and especially by LP. However,
there are no differences between viewing 2D content
on LCD TVs and 3D content with polarized glasses.
In order to compare the subjective and objective
results, the design of the questionnaire is a key
point to enhance effectiveness. The rankings for
S3D and P3D are higher than for S2D, P2D, and
LP. Then S2D, LP, and P2D are in descending order
from high to low. But there are no differences between S3D and P3D. The total rank in the 3D mode
is higher than in the 2D mode, but there are no
differences between the LCD and DLP, or the shutter
and polarized. However, these results again show
that the level of visual fatigue is more serious after
viewing 3D videos.
It is obvious that some mismatches occurred in the
ΔHFC result analysis and questionnaire under
certain conditions. Though different ciliary body
conditions were detected, the visual discomfort, as
evaluated by the questionnaire, is the same. One
explanation is that the questionnaire is a subjective
method dependent on subjective cognition for
evaluations; hence, the reliability and validity of this
method need to be considered. Perhaps those
mismatches are due to this indicator’s inherent
inaccuracies, so it is not effective in diagnosing
and comparing the factors that cause visual fatigue
or their significance.
In conclusion, this paper discovers the following
fact: 3D videos cause people more visual fatigue than
do 2D videos (p < 0.001). The shutter glasses afflict
people with more visual fatigue than do polarized
glasses (p ˆ 0.012). The time-sharing method causes
people more visual fatigue than does the spatialformation method (p ˆ 0.008). There is no difference
between LED backlighting and mixed LED and LD
light sources (p ˆ 0.162). People with normal color
discrimination power suffer more visual fatigue than
those with generally good color discrimination power
(p < 0.001), but the levels of visual discomfort are the
same while viewing 2D on LCD TVs, and 3D contents
with polarized glasses. The HFC can indeed evaluate
physiological stress or strain resulting objectively
from exertion of the visual system, leading to visual
fatigue.
6. Discussion
In this work, using chromatic discrimination as a
grouping index is very successful. From the results
it can be found that people with normal color
discrimination have more visual fatigue than those
with good. In general, if someone’s chromatic discrimination is very good, he is supposed to distinguish the red, green, and blue subpixels easily when
viewing a display, which means he will feel the image
is blurred because RGB are separated. Viewing a
blur image or video for a long time causes people
to feel discomfort, especially in the eyes. On the
contrary, the results are different, as we expected.
People with normal color discrimination have more
visual fatigue than those with good (p < 0.001). This
part requires more verification to support the
results.
Another issue that should be addressed is the relation between luminance and accommodative microflutuation. As mentioned in Table 2, it can be found
that different modes have each luminance; the luminance can affect the pupil size. Also, the pupil size is
associated with accommodation, which means the
pupil size can affect the accommodation microflutuations. However, the HFC fluctuations are imperceptible when the pupil changes with light: this only
causes LFC fluctuation increases; it does not affect
our results [21]. Hence, accommodative microflutuation can effectively and objectively determine
whether subjects are suffering from visual fatigue
when compared with other indicators.
This paper was partially supported by the Aim for
the Top University Program of the National Chiao
Tung University, the Ministry of Education of
Taiwan, the Ministry of Science and Technology
of Taiwan (NSC 102-2220-E-009-016), and the
Industrial Technology Research Institute. The
authors also thank them for providing experimental
assistance and related information.
References
1. M. Lambooij, W. Ijsselsteijn, M. Fortuin, and I. Heynderickx,
“Visual discomfort and visual fatigue of stereoscopic displays:
a review,” J. Imaging Sci. Technol. 53, 030201 (2009).
2. E. H. Stupp and M. S. Brennesholtz, Projection Displays
(Wiley, 1999).
3. T. Jarvenpaa, “Measuring color breakup of stationary images
in field-sequential-color displays,” J. Soc. Inf. Display 13,
139–144 (2005).
4. D. L. Post, P. Monnier, and C. S. Calhoun, “Predicting color
breakup on field-sequential displays,” Proc. SPIE 3058,
57–65 (1997).
5. X. Zhang and J. E. Farrell, “Sequential color breakup
measured with induced saccades,” Proc. SPIE 5007,
210–217 (2003).
6. E. Umezawa, T. Shibata, T. Kawai, and K. Ukai, “Ergonomic
evaluation of a projector using field sequential color projection
system,” in Proceedings of the 11th International Display
Workshops (Society for Information Display, 2004),
pp. 1531–1534.
7. D. L. Post, A. L. Nagy, P. Monnier, and C. S. Calhoun, “Predicting color breakup on field-sequential displays: Part 2,” SID
Symp. Dig. Tech. Pap. 29, 1037–1040 (1998).
8. S. H. Kim, T. Shibata, T. Kawai, and K. Ukai, “Physiological
effects of color breakup in field-sequential color projection
system,” SID Symp. Dig. Tech. Pap. 37, 314–317 (2006).
9. M. Ogata, K. Ukai, and T. Kawai, “Visual fatigue in congenital
nystagmus caused by viewing images of color sequential
projectors,” J. Display Technol. 1, 314–320 (2005).
10. C. F. Chi and F. T. Lin, “A comparison of seven visual fatigue
assessment techniques in three data-acquisition VDT tasks,”
Human Factors 40, 577–590 (1998).
11. L. S. Gray, B. Gilmartin, and B. Winn, “Accommodation microfluctuations and pupil size during sustained viewing of visual
display terminals,” Ophthalmic Physiol. Opt. 20, 5–10 (2000).
12. M. Kajita, M. Ono, S. Suzuki, and K. Kato, “Accommodative
microfluctuation in asthenopia caused by accommodative
spasm,” Fukushima J. Med. Sci. 47, 13–20 (2001).
13. J. R. Wilson and E. N. Corlett, Evaluation of Human Work: A
Practical Ergonomics Methodology (Taylor & Francis, 1995),
Chap. 28.
14. N. Pouratian, S. A. Sheth, N. A. Martin, and A. W. Toga,
“Shedding light on brain mapping: advances in human optical
imaging,” Trends Neurosci. 26, 277–282 (2003).
15. M. J. Nichols and W. T. Newsome, “The neurobiology of
cognition,” Nature 402, C35–C38 (1999).
16. B. A. Wandell and R. F. Dougherty, “Computational neuroimaging: maps and tracts in the human brain,” Proc. SPIE
6057, 605701 (2006).
17. A. M. Dale and E. Halgren, “Spatiotemporal mapping of brain
activity by integration of multiple imaging modalities,” Curr.
Opin. Neurobiol. 11, 202–208 (2001).
18. J. D. Bullough, Y. Akashi, C. R. Fay, and M. G. Figueiro,
“Impact of surrounding illumination on visual fatigue and
eyestrain while viewing television,” J. Appl. Sci. 6, 1664–
1670 (2006).
19. W. N. Charman and G. Heron, “Fluctuations in accommodation: a review,” Ophthalmic Physiol. Opt. 8, 153–164 (1988).
20. T. Iwasaki and A. Tawara, “Reduction of asthenopia related to
accommodative relaxation by means of far point stimuli,” Acta
Ophthalmol. Scand. 83, 81–88 (2005).
21. S. Suzuki, M. Kajita, and K. Kato, “Evaluation of accommodative function by high frequency component of accommodative microfluctuation,” Jpn. J. Vis. Sci. 22, 93–97 (2001).
22. B. Winn and B. Gilmartin, “Current perspective on microfluctuations of accommodation,” Ophthalmic Physiol. Opt. 12,
252–256 (1992).
23. E. C. Huang and V. H. Barocas, “Accommodative microfluctuations and iris contour,” J. Vis. 6(5), 653–660 (2006).
24. M. Day, N. C. Strang, D. Seidel, L. S. Gray, and E. A. H. Mallen,
“Refractive group differences in accommodation microfluctuations with changing accommodation stimulus,” Ophthalmic
Physiol. Opt. 26, 88–96 (2006).
25. F. W. Campbell, J. G. Robson, and G. Westheimer, “Fluctuations of accommodation under steady viewing conditions,”
J. Physiol. 145, 579–594 (1959).
26. E. Harb, F. Thorn, and D. Troilo, “Characteristics of accommodative behavior during sustained reading in emmetropes and
myopes,” Vis. Res. 46, 2581–2592 (2006).
27. P. Denieul, “Effects of stimulus vergence on mean accommodation response, microfluctuations of accommodation and
optical quality of the human eye,” Vis. Res. 22, 561–569
(1982).
28. C. Miege and P. Denieul, “Mean response and oscillations of
accommodation for various stimulus vergences in relation
to accommodation feedback control,” Ophthalmic Physiol.
Opt. 8, 165–171 (1988).
29. J. C. Kotulak and C. M. Schor, “Temporal variations in accommodation during steady-state conditions,” J. Opt. Soc. Am. A
3, 223–227 (1986).
30. W. N. Charman and H. Radhakrishnan, “Accommodation,
pupil diameter and myopia,” Ophthalmic Physiol. Opt. 29,
72–79 (2009).
31. F. W. Campbell and G. Westheimer, “Dynamics of the accommodation responses of the human eye,” J. Physiol. 151, 285–
295 (1960).
32. L. R. Stark and D. A. Atchison, “Pupil size, mean accommodation response and the fluctuations of accommodation,”
Ophthalmic Physiol. Opt. 17, 316–323 (1997).
33. L. J. Bour, “The influence of the spatial distribution of a target
on the dynamic response and fluctuations of the accommodation of the human eye,” Vis. Res. 21, 1287–1296 (1981).
34. S. Phillips and L. Stark, “Blur: a sufficient accommodative
stimulus,” Doc. Ophthalmol. 43, 65–89 (1977).
35. C. M. Schor, C. A. Johnson, and R. B. Post, “Adaptation of tonic
accommodation,” Ophthalmic Physiol. Opt. 4, 133–137 (1984).
36. L. R. Stark, N. C. Strang, and D. A. Atchison, “Dynamic accommodation response in the presence of astigmatism,” J. Opt.
Soc. Am. A 20, 2228–2236 (2003).
37. M. Alpern, “Variability of accommodation during steady
fixation at various levels of illuminance,” J. Opt. Soc. Am.
48, 193–197 (1958).
38. A. Mira-Agudelo, L. Lundstrom, and P. Artal, “Temporal
dynamics of ocular aberrations: monocular vs binocular
vision,” Ophthalmic Physiol. Opt. 29, 256–263 (2009).
39. T. Iwasaki and S. Kurimoto, “Objective evaluation of eye
strain using measurements of accommodative oscillation,”
Ergonomics 30, 581–587 (1987).
40. D. R. Iskander, M. J. Collins, M. R. Morelande, and Z. Mingxia,
“Analyzing the dynamic wavefront aberrations in the human
eye,” IEEE Trans. Biomed. Eng. 51, 1969–1980 (2004).
41. B. Winn, J. R. Pugh, B. Gilmartin, and H. Owens, “Arterial
pulse modulates steady-state ocular accommodation,” Curr.
Eye Res. 9, 971–975 (1990).
42. T. Geacintov and W. S. Peavler, “Pupillography in industrial
fatigue assessment,” J. Appl. Psych. 59, 213–216 (1974).
43. H. S. Rigby, B. F. Warren, J. Diamond, C. Carter, and J. W. B.
Bradfield, “Colour perception in pathologists: the Farnsworth–Munsell 100-hue test,” J. Clin. Pathol. 44, 745–748
(1991).
44. M. Collins, B. Davis, and J. Wood, “Microfluctuations of
steady-state accommodation and the cardiopulmonary
system,” Vis. Res. 35, 2491–2502 (1995).
45. M. Emoto, T. Niida, and F. Okano, “Repeated vergence
adaptation causes the decline of visual functions in watching
stereoscopic television,” J. Display Technol. 1, 328–340 (2005).