Tashiro2014 PDF

Lighting Research and
Technology http://lrt.sagepub.com/
Discomfort glare for white LED light sources with different spatial arrangements
T Tashiro, S Kawanobe, T Kimura-Minoda, S Kohko, T Ishikawa and M Ayama
Lighting Research and Technology published online 24 April 2014
DOI: 10.1177/1477153514532122
The online version of this article can be found at:

http://lrt.sagepub.com/content/early/2014/04/24/1477153514532122
Published by:
http://www.sagepublications.com
On behalf of:
The Society of Light and Lighting
Additional services and information for Lighting Research and Technology can be found at:
Email Alerts: http://lrt.sagepub.com/cgi/alerts
Subscriptions: http://lrt.sagepub.com/subscriptions
Reprints: http://www.sagepub.com/journalsReprints.nav
Permissions: http://www.sagepub.com/journalsPermissions.nav
Citations: http://lrt.sagepub.com/content/early/2014/04/24/1477153514532122.refs.html
>> OnlineFirst Version of Record - Apr 24, 2014
What is This?
Downloaded from lrt.sagepub.com at CARLETON UNIV on June 2, 2014

XML Template (2014) [23.4.2014–10:06am] [1–22]
//blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/LRTJ/Vol00000/140021/APPFile/SG-LRTJ140021.3d (LRT) [PREPRINTER stage]
Lighting Res. Technol. 2014; 0: 1–22
Discomfort glare for white LED light

sources with different spatial
arrangements
a b b c a
T Tashiro MSc , S Kawanobe MSc , T Kimura-Minoda PhD , S Kohko BSc , T Ishikawa PhD
a
and M Ayama PhD
a
Graduate School of Engineering, Utsunomiya University, Tochigi, Japan
b
Stanley Electric Company Limited, Meguro-ku, Tokyo, Japan
c
Iwasaki Electric Company Limited, Chuo-ku, Tokyo, Japan
Received 4 November 2013; Revised 19 February 2014; Accepted 25 March 2014
Quantitative relations between discomfort glare evaluation and photometric

quantities such as illuminance at the observer’s eye, average luminance of the
source area, average luminance of the effective area and effective glare luminance
for white LED sources having a variety of spatial luminance distributions have
been investigated. Effective glare luminance, which is the sum of luminances in
the luminaire area divided by the effective area, explains the scaling results of all
sources in the same way. In addition, a new equation modified from the
Commission Internationale de l’Éclairage Unified Glare Rating formula using the
effective glare luminance showed a strong correlation with the scaling results. It is
thus suggested that effective glare luminance is a useful index of discomfort glare
for light sources having different spatial luminance distributions.
1. Introduction indoor or outdoor architectural lighting,

street lighting, retail display lighting, enter-
The rapid development of high-power light- tainment lighting and the headlights of auto-
emitting diodes (LEDs) has made it possible mobiles, where incandescent, fluorescent,
to manufacture white LEDs for which the mercury, metal halide and high-pressure
luminous efficacy exceeds 100 lm/W. LEDs sodium lamps have been used.
are now replacing other light sources not only Most of this general lighting, especially
in indicators, signage and decorative lighting, street lights, has to fulfil conflicting require-
but also as general illumination. The defin- ments, for example, to provide a satisfactory
ition of general illumination is rather ambigu- view for both pedestrians and drivers with-
ous, but it refers to any lighting that makes out either experiencing glare. A goal for
useful or functional an area where the illu- illumination engineering is to establish a
minance without lighting is insufficient. way of designing lighting installations that
Typical examples of general illumination are can provide good visibility for users without
glare.
1.1. Types of glare

Glare is defined as the sensation produced
Address for correspondence: Tomonori Tashiro, Ayama by luminance within the visual field that is
Laboratory (Room 9-410), Utsunomiya University, 7-1-2,
Yoto, Utsunomiya-shi, Tochigi 321-8585, Japan. sufficiently greater than the luminance to
E-mail: [email protected] which the eyes are adapted to cause
ß The Chartered Institution of Building Services Engineers 2014 10.1177/1477153514532122

XML Template (2014) [23.4.2014–10:06am] [1–22]
2 T Tashiro et al.
annoyance, discomfort, or loss of visual that relates these five factors as shown in
performance and visibility.1 Glare is generally equation (1).6
divided into two types, disability glare and
discomfort glare. Disability glare is related to A B2
GF ¼ ð1Þ
a masking effect that reduces the visibility of D2 ﬀ2 S0:6
the target and is often called ‘physiological
glare’. On the other hand, discomfort glare where A is the apparent area of the source in
produces discomfort or a bothersome feeling square inches, B is the brightness of the
for observers and is often called ‘psycho- source in footlamberts divided by 1000, D is
logical glare’. Although discomfort glare does distance in feet from the source to the eye
not necessarily impair visual performance or divided by 10, ﬀ is the angle in degrees above
visibility, sometimes it produces physiological the horizontal divided by 10 and S is the
and psychological symptoms such as head- surrounding brightness in footlamberts
ache or stress. Some migraine suffers are divided by 10. These five factors are based
reportedly more sensitive to discomfort glare on the reports or experiments of Holladay7,8
than normal controls.2,3 It is important to or Nutting,9 and GF is well-known as the base
prevent discomfort glare because the source of discomfort glare evaluation.
luminance at which discomfort glare occurs is In the wake of this, many researchers
usually lower than that of disability glare; undertook discomfort glare studies, and
thus, if the source luminance is set below the reviews of research progress at different
threshold for discomfort glare, it could also times have been reported by several
prevent disability glare. authors.10,11 This action led to the establish-
Some researchers employ three subdiv- ment of major discomfort glare indices, such
isions for glare, adding ‘blinding glare’ or as the Unified Glare Rating (UGR)12 by the
‘dazzling glare’ to the above two types.4,5 This Commission Internationale de l’Éclairage
glare occurs when we see a very bright field of (CIE), developed from the CIE glare index
view that is close to a hazardous level, makes (CGI).13 The UGR formula describes the
us squint and is extremely uncomfortable. combined effect of the luminance, size and
This glare is considered a part of disability position of glare sources, as well as the
glare. For this reason, blinding glare or background luminance, as shown in equation
dazzling glare is not addressed in this study. (2). The degree of discomfort glare varies
from 7 (insensitive) to 31 (unbearable).

1.2. Discomfort glare indices 0:25 X L2i !i
UGR ¼ 8 log ð2Þ
In the century-long history of the study of LB P2i
glare, various indices to express subjective
appraisals of discomfort glare in different where Li is the luminance of the luminous
situations have been proposed. In the mid- parts of each luminaire in the observer’s
20th century, Harrison introduced the five visual field (cd/m2), LB is the background
factors; area, brightness of glare source, luminance (cd/m2), !i is the solid angle of the
surrounding brightness of glare source, dis- luminous parts of each luminaire at the
tance between the glare source and observer observer’s eye (sr) and Pi is the Guth position
and angle between the observers line of sight index for each luminaire, this being a function
and the light beam from the glare source, of the displacement from the observer’s line of
which affect the glare rating (GR), and sight. This equation indicates plainly the basic
developed the glare factor (GF), a metric features of discomfort glare evaluation; the

XML Template (2014) [23.4.2014–10:06am] [1–22]
Discomfort glare and LED spacing 3
subjective rating increases with an increase in For road lighting, disability glare is con-
the luminance and solid angle of the glare sidered the first priority in order to maintain
source and decreases with an increase in the the driver’s visibility at a satisfactory level for
background luminance. safe driving. The level of disability glare is
Although the UGR has been recognized as quantified by the TI shown by equation (6).
a useful glare index for interior lighting, its
Lvl
application is limited to light sources within a TI ¼ 65 ð6Þ
solid angle in the range of 0.0003–0.1 sr. To Lave
overcome these limitations, the CIE has
published supplementary recommendations where Lvl is the equivalent veiling luminance
for the quantitative assessment of glare from produced by the luminaries (cd/m2) and Lave
small, large and complex sources.14 However, is the average road surface luminance (cd/m2).
there remain many situations in which the In both GR and TI, the glare source is
UGR is not able to assess discomfort glare, assumed to be either a point source or a light
e.g. the glare from windows in daylight.11,15 source that has a uniform luminance, and Lvl
Other glare evaluation formulae have been is calculated from the vertical illuminance at
used for a variety of situations; for example, the observer’s eye, Eov, as shown in equation
the GR is used for evaluating discomfort (4), which can be easily measured.
glare in outdoor sports lighting, and the Therefore, these indices have been used for
threshold increment (TI) is used for evaluat- a long time in the assessment of environmen-
ing of disability glare in road lighting.16,17 GR tal lighting design. However, the appearance
is expressed in equation (3) as a range from 10 of LEDs in outdoor or road lighting instal-
(unnoticeable) to 90 (unbearable). lations has made the simple use of GR or TI
irrelevant in some cases as mentioned in the
Lvl next section. Some results obtained here could
GR ¼ 27 þ 24 log 0:9 ð3Þ not be explained by the illuminance at the
Lve
observer’s eye alone, requiring luminance
information about the glare source such as
where Lvl and Lve are the equivalent veiling is included in the UGR formula. Therefore, in
luminance produced by the luminaires (cd/ the discussion section, we have focused on the
m2) and by the environment (cd/m2), modification of the UGR formula rather than
respectively. GR or TI.
Lvl and Lve are expressed by equations (4)
and (5). 1.3. Discomfort glare of coloured LEDs and
X non-uniform sources
Lvl ¼ 10 Eov 2 ð4Þ
With the rapid increase in high-power LED
Lve 0:035 Eh = ð5Þ usage in general lighting, the relationship
between a LED light source and discomfort
glare has been reported by many researchers.
where Eov is the vertical illuminance at the From a visual perception point of view, two
observer’s eye (lx), is the angle between the major advantages of LEDs are colour variety
observer’s line of sight and the direction of and freedom of spatial design. However, both
light from the source (8), Eh is the average could become demerits if they are not utilized
horizontal illuminance (lx) and is the properly. Therefore, several studies have been
reflectance of the area assuming diffuse reported on the effect of colour and spatial
reflection. non-uniformity on discomfort glare.

XML Template (2014) [23.4.2014–10:06am] [1–22]
4 T Tashiro et al.
Discomfort glare of coloured LEDs has array, and the degree of discomfort glare of a
been reported by Kimura-Minoda and light source with high maximum luminance is
Ayama in relation to the brightness sensitivity greater than that of low luminance. Their
mechanism of colour vision.18 They experi- ambient light was reflected light from the
mentally evaluated the discomfort glare white paper placed in the lower portion of the
caused by six types of coloured LED light glare source, which was called the ‘surround
sources with different spectral distributions in area’ in their paper. This is not the same as the
which the peak wavelength varied from background luminance in the UGR formula,
459 nm to 628 nm. They investigated the however it seems that the effect of varying the
brightness sensitivity of individual observers adaptation level of the observer’s eye resulting
and classified the observers into three types: in a change of discomfort GR value is similar.
red and blue chromatic, blue chromatic and They also reported intriguing results about
achromatic. The red and blue chromatic type the spatial factor. The illuminance from the
has a brightness sensitivity higher than V() source is useful to predict a degree of
at long and short wavelengths; the blue discomfort glare sensation when the source
chromatic type has a brightness sensitivity size is smaller than 0.38 at the eyes, but when
higher than V() at short wavelengths only, the light source is larger than that, the effect
and the achromatic type has the brightness of maximum luminance of the glare source
sensitivity similar to V() at all wavelengths. becomes non-negligible.
These results showed a positive correlation The effect of spatial non-uniformity on
between brightness sensitivity and discomfort discomfort glare using LEDs has not been
glare sensitivity for individual observers; i.e. reported in many studies yet. Therefore, we
chromatic observers who have a higher sen- introduce the following two experiments in
sitivity to brightness than achromatic obser- which other light sources were used because
vers are more sensitive to the discomfort glare the results could be very helpful in under-
of coloured lights. standing the relationship of discomfort glare
As mentioned earlier, the compactness of and influence of any spatial non-uniformity.
LEDs and hence their freedom of design are Kasahara et al. investigated the influence of
distinct advantages. The maximum size of the arrangement as well as the number of
commercially available LED chips is a few small light elements and the luminance distri-
square centimetres, and most LED light bution of the light source surface on discom-
sources are shaped according to the geomet- fort glare.21 They used a metal halide lamp
rical arrangements of common LED chips, covered by white painted aluminium board
such as matrix, linear or staggered patterns. with small holes covered by light diffusing
However, this produces a spatial non- paper that simulated a luminaire made of
uniformity. For the effect of a spatial non- LED elements, instead of an actual LED light
uniformity on discomfort glare, the following source. They conducted an experiment on the
studies have been reported recently. characteristics of discomfort glare caused by
Bullough and Sweater-Hickcox examined different numbers and intervals of point
the effect of the light source illuminance, sources, and measured the borderline between
luminance and size on discomfort glare comfort and discomfort (BCD). By examining
responses.19,20 In these studies, observers the correlation between the BCD luminance
viewed a LED array directly or through a and the number of light elements in the glare
diffusing element. The results showed that the source, they found that discomfort glare
presence of ambient light significantly reduces depends largely on the spatial arrangement
discomfort glare perception from the LED of the light source.

XML Template (2014) [23.4.2014–10:06am] [1–22]
Eble-Hankins and Waters examined the 2. Experiment 1

effects of the spatial frequency and position of
stimuli on discomfort glare.22 As the stimuli 2.1. Test stimuli
to be evaluated, their experiments used In experiment 1, the effect of the homo-
pseudo light sources consisting of plates geneity of an LED source was investigated.
with a sinusoidal pattern of various spatial The test stimuli were three types of LED
frequencies that reflect light from an incan- sources, as shown in Figure 1. Each source
descent theatrical source. They performed a consisted of an outdoor LED lighting unit
paired comparison experiment and a rating made by Stanley Electric Co., Ltd. (SEC)
scale experiment. Although they concluded whose colour temperature was about 5000 K.
that both the spatial frequency and the LED elements were arranged in a three-by-
position of stimuli are significant predictors three matrix within an area of 13 cm 18 cm,
of discomfort glare, the average rating value i.e. 2.58 3.48 in visual angle.
does not vary significantly with spatial fre- The first one, type I, was made of raw LED
quency, whereas the frequency of selection in elements; the second, type II, was made of
the pair comparison method increases slightly LED elements covered by a lens cap, and the
with spatial frequency at eccentricities of 108, third, type III, was the entire LED matrix
208 and 308. covered by frosted glass to obtain a roughly
As described above, the studies by uniform luminance surface. Relative spectral
Kasahara et al. and Eble-Hankins and power distributions of the LED sources are
Waters have shown that the spatial pattern shown in Figure 2. For each of the test
of the stimulus influences the degree of stimuli, seven brightness levels ranging from
discomfort glare. However, these studies do unnoticeable to unbearable were determined
not clarify the quantitative relation between in preliminary experiments, and the bright-
the degree of discomfort glare and the ness was controlled by inserting different
photometric properties of test stimuli having neutral density (ND) filters in front of the
a variety of spatial arrangements or patterns. LED source. The optical density of each of
the ND filters is shown in Table 1(a).
1.4. Purpose of this study Numerals in the left most column in
Different spatial arrangements or patterns Table 1(a) to (d) represent a nominal bright-
produce different spatial distributions of the ness, 1 and 7 corresponding to the lowest and
luminance within a luminaire area which the highest brightnesses, respectively.
seems to be an essential photometric factor
for discomfort glare. Thus, the problem is to 2.2. Experimental conditions
clarify the relation between the degree of Figure 3 shows the experimental apparatus.
discomfort glare and some photometric quan- The booth was covered by a black curtain to
tity of sources having a variety of spatial block stray light from the environment. The
luminance distributions. Therefore, the pur- reflectance of the inside surface of the curtain
pose of our study is to investigate the was about 3%, and no chromatic object was
relationship between discomfort glare evalu- placed inside so when the glare source was on,
ation and some photometric quantity of LED the colour of the surround is neutral. The test
light sources in order to establish a discomfort stimulus was at the upper side of the fixation
GR formula for LEDs. We have conducted point, which was set approximately at the
four experiments using white LED light height of the observer’s eye. The height of
sources to produce different spatial luminance the test stimulus was about 1.7 m, and the
distributions. observing distance was about 3 m. The visual

XML Template (2014) [23.4.2014–10:06am] [1–22]
6 T Tashiro et al.
Type I Type II Type III
Figure 1 Test stimuli for experiment 1
1 light-blocking board, so as not to affect the

Type I glare evaluation. In the dark background
0.8 Type II condition, the fluorescent lamps were off. The
Type III
luminance of the background in the dark
0.6
condition was too low to measure using a CS-
Energy
0.4 2000 (Konica Minolta Inc.) in which the lowest

measurable luminance is 0.001 cd/m2 accord-
0.2 ing to the specification. In other background
conditions (0.1 cd/m2, 1.0 cd/m2 and 10 cd/m2),
0
400 450 500 550 600 650 700 luminance was measured at the three points on
Wavelength (nm) the both sides of the light source. The average
Figure 2 The spectral power distributions of types I, II
luminances were 0.097 cd/m2, 1.206 cd/m2 and
and III 10.59 cd/m2, respectively.
angle between the observer’s line of sight and 2.3. Procedure and observer
the direct light beam from the centre of the The evaluation scale is based on Matsuda
test stimulus was 8.58. This apparatus was et al.’s response scale,23 where 1, 5 and 9
constructed to simulate the outdoor condi- correspond to unnoticeable, beginning to feel
tions of a pedestrian looking straight ahead and unbearable discomfort glare, respectively.
under a street light at a height of 4.5 m and a It is similar to de Boer’s response scale,24
distance of 30 m. which is frequently used in studies of discom-
Four levels of background luminance were fort glare, but it differs in that greater
employed: dark, 0.1 cd/m2, 1.0 cd/m2 and discomfort glare corresponds to a larger
10 cd/m2. Each level of background luminance number, which is preferable for Japanese
is achieved using black (dark, 0.1 cd/m2 and observers.
1.0 cd/m2 conditions) or grey (in the 10 cd/m2 The experimental procedure was as follows:
condition) cloth and two fluorescent lamps the background luminance was set to one of
with ND filters. We hung the black or grey the four conditions and kept constant during
cloth 1 m behind the LED light source, and set each session. The observer entered the experi-
up two fluorescent lamps so as to illuminate the mental apparatus and became adapted to the
cloth as evenly as possible. The two fluorescent background luminance for about 5 minutes.
lamps were insulated from the observer using a Then, the experimenter presented the observer

XML Template (2014) [23.4.2014–10:06am] [1–22]

Table 1 Summary table for all the conditions in all the experiments
(a)
Ex1
Source type T1 T2 T3
Intensity level ND RS SD ND RS SD ND RS SD
1 4.19 1.17 0.796 3.86 1.58 0.400 2.05 2.21 0.236

2 3.71 1.90 0.623 3.39 2.38 0.633 1.82 2.79 0.398
3 3.39 2.33 0.836 2.83 3.13 0.840 1.34 4.42 0.643
4 3.00 3.54 0.563 2.37 4.27 1.133 1.02 5.25 0.739
5 2.53 4.98 0.720 1.82 5.54 0.835 0.63 6.65 0.826
6 2.20 6.44 0.906 1.34 6.65 0.898 0.30 7.58 0.759
7 1.82 7.73 0.742 0.86 7.85 1.025 0.00 8.35 0.840
(b)
Ex2
Pitch length 0.28 0.48 0.68 0.88 1.08 1.48 1.98 2.58
Intensity level ND RS SD RS SD RS SD RS SD RS SD RS SD RS SD RS SD
1 3.00 1.76 0.622 1.64 0.581 1.88 0.559 2.00 0.604 1.92 0.692 1.96 0.584 1.88 0.600 1.76 0.536
2 2.50 2.20 0.827 2.32 0.940 2.32 0.867 2.60 0.914 2.44 0.835 2.44 0.859 2.24 0.715 2.28 0.710
3 2.05 3.28 1.398 2.96 0.918 2.92 1.051 2.96 0.934 3.24 1.044 3.00 1.072 3.04 0.944 2.76 0.948
4 1.49 4.24 0.931 4.24 0.904 4.16 0.900 4.04 0.746 4.32 0.981 4.44 0.848 4.00 0.927 4.32 1.054
5 1.02 5.28 1.139 5.08 1.125 5.08 1.206 5.20 1.082 4.92 0.863 5.12 1.018 5.16 1.050 5.28 0.951
6 0.45 6.88 1.079 1.00 1.107 6.92 0.822 6.56 0.892 6.56 0.801 6.64 0.779 6.84 0.695 6.88 0.861
7 0.00 8.32 0.646 8.24 0.765 7.80 0.866 7.76 0.779 8.00 0.688 7.76 0.714 7.64 0.771 7.76 0.731
(c)
Ex3
Matrix type 22 33 55
Intensity level ND RS SD RS SD RS SD
1 3.00 1.36 0.518 1.80 0.867 2.36 1.291

2 2.50 1.77 0.826 2.39 0.950 3.10 1.265
3 2.05 2.47 0.935 3.39 1.177 4.09 1.359
4 1.49 3.68 1.276 4.42 1.278 5.42 1.265
5 1.02 4.54 1.458 5.34 1.225 6.36 1.110
6 0.45 5.97 1.346 6.92 1.063 7.84 1.053
7 0.00 6.98 1.109 8.06 1.108 8.67 0.887
(d)
Ex4
Diffusion angle 08 208 608
Intensity level ND RS SD ND RS SD RS SD
1 3.39 2.22 1.462 2.050 2.68 1.360 1.77 1.238

2 3.00 3.10 1.368 1.820 3.25 1.403 2.47 1.167
3 2.53 4.55 1.161 1.340 4.58 1.385 3.62 1.261
4 2.20 5.85 1.050 1.020 5.62 1.012 4.63 1.140
5 1.82 7.37 0.880 0.630 7.18 1.058 6.28 1.132
6 0.30 9.79 0.270 0.300 7.98 0.683 7.28 1.066
7 0.00 8.94 0.154 0.000 8.50 0.436 8.13 0.924
ND: optical density of neutral density filter(s), RS: the average discomfort glare rating of all observers, SD: standard
deviation of RS.
with the test stimulus at a certain brightness subjectively evaluated the degree of discomfort
for 3 seconds. During the experiment, the glare using Matsuda et al.’s response scale.
observer was instructed to use the chin rest Stimuli of different brightnesses were pre-
and to fixate the fixation point. The observer sented in a random order.

XML Template (2014) [23.4.2014–10:06am] [1–22]
8 T Tashiro et al.
Fluorescent lighting Background

LED source
LED source
Observer
8.5°
1.7 m
Fixation point
Chin rest Fluorescent lighting
Background Fixation point
Light–blocking board
1m 3m Light-blocking board
Figure 3 Experimental apparatus
Eight observers, six males and two females area and effective area were calculated from
in their twenties or thirties (the average age the HDR image, respectively. The illuminance
was 23.6 years) with normal colour vision, at the observer’s eye refers to the vertical
participated in the experiment. The order of illuminance at the eye position of the obser-
the different background luminance condi- ver, i.e. measured with the normal to the
tions was randomized for each observer. fixation point. The average luminance of the
source area is the sum of the luminance of
2.4. Photometry the pixels in the HDR image divided by the
According to recent studies of discomfort source area, which is represented by the
glare using non-uniform light sources, some number of pixels corresponding to the size
measure of the luminance distribution within of the LED light matrix. This area corres-
the luminaire seems to be a key factor.25 We ponds to the outer rectangular area of the
created a high dynamic range (HDR) image LED matrix and thus is constant in size
to obtain the detailed spatial luminance dis- despite the intensity level and types of light
tribution of the test stimuli. The HDR image source. The average luminance of the effective
with 980 980 pixels resolution was obtained area refers to the sum of the luminance of the
using a 2D colour analyzer (CA-2500, Konica pixels in the HDR image divided by the
Minolta Inc.) and ND filter. In experiment 1, effective area, which is defined as the number
185 pixels correspond to approximately 18 of of pixels that have a luminance value (¼non-
visual angle. zero pixels) in this study.
In this study, we examined the relationship
between the discomfort GR and three photo- 2.5. Results
metric quantities, the illuminance at the Average rating of discomfort glare for all
observer’s eye, the average luminance of observers and the associated standard devi-
the source area and the average luminance ations for all conditions are shown in
of the effective area. The illuminance at the Table 1(a). Types I, II and III are abbreviated
observer’s eye was measured using an illu- to T1, T2 and T3, respectively, for conveni-
minometer (T-10, Konica Minolta Inc.), and ence in the analysis described in later sections.
the average luminance values of the source Figures 4(a-1) and (a-2) show the

XML Template (2014) [23.4.2014–10:06am] [1–22]
relationships between the discomfort GR and Table 2 Chromaticity data of types I, II, and III
the illuminance at the observer’s eye and Source type Type I Type II Type III
average luminance of the source area, respect- Intensity level x y x y x y
ively, under the background luminance of 1 0.328 0.384 0.329 0.382 0.375 0.433
1.0 cd/m2. The results plotted against the 2 0.329 0.384 0.326 0.381 0.374 0.428
average luminance of the effective area are 3 0.328 0.381 0.326 0.378 0.374 0.422
4 0.320 0.365 0.325 0.376 0.378 0.420
not shown because their trend is quite similar 5 0.331 0.341 0.325 0.371 0.379 0.413
to that of the illuminance at the observer’s 6 0.330 0.364 0.326 0.371 0.379 0.707
eye. As shown in Figures 4(a-1) and (a-2), the 7 0.330 0.362 0.330 0.369 0.378 0.402
Ave 0.328 0.374 0.327 0.376 0.377 0.418
discomfort GRs for types I and II agree quite
well, but those for type III shift to larger
values along the horizontal axis.
Although the results under other back- This indicates that the discomfort GR of
ground luminance conditions exhibit almost non-uniform light sources, such as types I and
the same trend, the discomfort GR decreases II, is larger than that of uniform light sources,
as the background luminance increases as such as type III, when the illuminance at the
shown in Table 1(a). observer’s eye is about the same.
We tried to conduct a two-way factorial Another issue that should be mentioned
analysis of variance (ANOVA) for three here is the effect of light source colour.
curves in Figure 4(a-1), but it could not be Is the difference in the evaluation of type III
conducted directly because the values of hori- due to the chromaticity difference between
zontal axis of measured points for the three type III and type I or II? In order to answer this
types are not the same. This is because problem, we measured the chromaticity of
each source and obtained the results shown in
different ND filters were employed to control
Table 2. The chromaticities of three sources are
the brightness of the different light sources,
close to each other. x and y between types
resulting in different illuminances at the obser-
II and III are larger than those between types I
ver’s eye. To conduct an ANOVA, the values
and III, but they are still not very large (both
of the dependent variable under different
are less than 0.05). According to the study of
source types obtained at the same independent
Kimura-Minoda and Ayama, there was no
value(s) are necessary. Therefore, we calcu- difference in the discomfort GR between two
lated the approximate discomfort GR values different-coloured LEDs, amber [(x,y) ¼ (0.55,
for types I, II and III at the illuminances at the 0.45)] and white [(x,y ¼ (0.30, 0.31)].18 The
observer’s eye of 1.0 lx, 2.5 lx, 5.0 lx, 7.5 lx and chromaticity differences between our LED
10 lx, for each observer by linear interpolation. light sources are much smaller than that,
These illuminances were chosen within the implying the chromaticity difference between
range where results of all three types have the test stimuli of different types is unlikely to
measured points. After that, we conducted be the main cause of the discrepancy between
two-way factorial ANOVA using these type III compared to types I and II.
values. Statistically significant effects are
found for both the illuminance at the obser-
ver’s eye and the source type. The former is 3. Experiment 2
obvious but for the latter, we carried out
supplementary tests, and the results revealed 3.1. Test stimuli
that there is a significant difference (p50.05) In experiment 2, the effect of the pitch
between types I and III (p ¼ 1.28 106), and length of a three-by-three LED matrix on
types II and III (p ¼ 1.28 106). discomfort glare was investigated. Figure 5

XML Template (2014) [23.4.2014–10:06am] [1–22]
10 T Tashiro et al.
(a1) 9 (a2) 9
Background: 1 cd/m2 Background: 1 cd/m2
8 8
Discomfort glare rating

7 7
6 6
5 5
4 4
Type I Type I
3 3
Type II Type II
2 Type III 2 Type III
1 1
102 10–1 100 101 102 103 100 101 102 103 104 105 106
Illuminance at the observer's eye (lx) Average luminance of the source area (cd/m2)
(b1) 9 (b2) 9
8 8

7 7
6 0.2° 6 0.2°
0.4° 0.4°
5 0.6° 5 0.6°
4 0.8° 4 0.8°
1.0° 1.0°
3 1.4° 3 1.4°
1.9° 1.9°
2 2.5°
2 2.5°
1 1
10–2 10–1 100 101 102 103 100 101 102 103 104 105 106
(c1) 9 (c2) 9
8 8
7 7
6 6
5 5
4 4
2x2 2x2
3 3
3x3 3x3
2 5x5 2 5x5
1 1
10–2 10–1 100 101 102 103 100 101 102 103 104 105 106
(d1) 9 (d2) 9
8 8
7 7
6 6
5 5
4 4
0° 0°
3 3
20° 20°
2 60° 2 60°
1 1
10–2 10–1 100 101 102 103 100 101 102 103 104 105 106
Figure 4 Results of (a) experiment 1, (b) experiment 2, (c) experiment 3 and (d) experiment 4

XML Template (2014) [23.4.2014–10:06am] [1–22]
Visual angle of the pitch length

0.2° 0.4° 2.5°
shows the test stimuli used in experiment 2. conditions. First, we plotted the relation
The source is a high-output small LED chip between discomfort glare and the pitch
made by SEC, and the pitch length can be length of the LED matrix (Figure 6). As
changed from 0.28 to 2.58 in visual angle by shown in Figure 6(a), all the lines for F1 to F7
using a switch. Pitch lengths of 0.28, 0.48, 0.68, stay roughly constant, indicating that the
0.88, 1.08, 1.48, 1.98 and 2.58 in visual angle spatial interval variation between LED elem-
were employed. In addition, seven bright- ents employed in this study (0.2–2.58 in visual
nesses ranging from unnoticeable to unbear- angle) does not affect the subjective rating of
able were determined in preliminary discomfort glare. However, when we compare
experiments, and they were controlled by the results for the same pitch length and ND
inserting different ND filters in front of the filter(s) but with different background condi-
test stimulus. The optical density of each of tions, the discomfort GR decreases as the
the ND filters is shown in Table 1(b). background luminance increases as shown in
the case of F7 in Figure 6(b).
3.2. Experimental conditions Next, the relation between the discomfort
The experimental apparatus, procedure GR and the illuminance at the observer’s eye,
and evaluation scale are the same as those and the relation between the rating score and
used in experiment 1. Three levels of back- the average luminance of the source area are
ground luminance were employed: 0.1 cd/m2, shown in Figures 4(b-1) and (b-2). These
1.0 cd/m2 and 10 cd/m2. The dark background results were obtained under a background
condition was omitted because the results of luminance of 1 cd/m2. All the curves agree
experiment 1 showed no significant difference quite well when plotted against the illumin-
from the background luminance of 0.1 cd/m2. ance at the observer’s eye. In the similar way
Twelve observers, eight males and four to experiment 1, the estimated discomfort GR
females in their twenties (the average age values were calculated using interpolation
was 21.7 years) with normal colour vision when the illuminance at the observer’s eye
participated. Each photometric quantity was were 0.08 lx, 0.12 lx, 0.21 lx, 0.64 lx, 1.58 lx,
measured using the same method as in 5.91 lx and 12.3 lx, and the two-way factorial
experiment 1. ANOVA was done. Illuminance at the obser-
ver’s eye has an obvious statistically signifi-
3.3. Results cant effect, but no significant effect was found
Table 1(b) shows the mean ratings of for pitch length (p ¼ 0.65). The results plotted
discomfort glare and the associated standard against the average luminance of the effective
deviations, for all observers, for all area are not shown because their tendency is

XML Template (2014) [23.4.2014–10:06am] [1–22]
12 T Tashiro et al.
quite similar to that of the illuminance at the reason is that the luminous flux radiated from
observer’s eye. On the other hand, when the the source is about the same because the
mean ratings are plotted against the average number of LED elements and the optical
luminance of the source area, the eight curves density of the ND filter(s) are the same,
disagree as shown in Figure 4(b-2). As the whereas the source area increases with
pitch length increases, the curve shifts increasing pitch length. In the cases of larger
towards smaller values along the horizontal pitch length, dark areas between LED elem-
axis. ents become larger than those of a small pitch
The average luminance of the source area length luminaire. Some pixels in such dark
decreases with increasing pitch length. The areas become zero-value in the HDR image.
(a) 9
Background: 1 cd/m2
8
7
6
F1
5 F2
4 F3
F4
3 F5
F6
2
F7
1
0 0.5 1 1.5 2 2.5 3
Visual angle of pitch length (degree)
(b) 9
7
3 Background: 0.1 cd/m2

Background: 1 cd/m2
2 Background: 10 cd/m2
1
0 0.5 1 1.5 2 2.5 3
Visual angle of pitch length (degree)
Figure 6 Results of experiment 2 – relation between discomfort glare and the pitch length of LED matrix. ((a) F1–F7
denote the results with the different ND-filters and (b) effect of background luminance in the case of F7)

XML Template (2014) [23.4.2014–10:06am] [1–22]
The number of zero-value pixels increases 4.2. Experimental conditions

with increasing pitch length. In other words, The conditions, experimental apparatus,
although the luminaire area increases with procedure, evaluation scale and method of
pitch length, there are some parts where the measuring each photometric quantity are the
luminance is practically zero. Taking this kind same as those in experiment 1, except that
of moth-eaten area as the luminaire area background luminances of 0.1 cd/m2, 1.0 cd/
caused an excessive decrease of average lumi- m2 and 10 cd/m2 were employed, as in experi-
nance, and the curves shift toward the left. ment 2. Nineteen observers, ten males and
These zero-value pixels should be eliminated, nine females in their twenties (the average age
and this leads us to the concept of ‘effective was 21.8 years) with normal colour vision,
area’ described in a later section. participated in the experiment.
The results under other background con- Each photometric quantity was measured
ditions also show a similar tendency to that using the same method as in experiment 1.
indicated in this figure.
4.3. Results
Table 1(c) shows the mean rating of
4. Experiment 3 discomfort glare and the associated standard
deviation for all observers, for all conditions.
4.1. Test stimuli The relationship between the mean discom-
In experiment 3, the effect of the density of fort GR and the two photometric quantities,
LED elements within a fixed area of illuminance at the observer’s eye, and the
3.88 3.88 was examined. The test stimuli average luminance of the source area, are
were three matrix types, 2 2, 3 3 and shown in Figures 4(c-1) and (c-2), respect-
5 5, as shown in Figure 7. They are ively. These results were obtained under a
composed of the same small LED chips used 1.0 cd/m2 background luminance. As shown
in experiment 2. In addition, for each of the in Figures 4(c-1) and (c-2), the results for the
test stimuli, seven brightnesses ranging from three different spatial conditions agree with
unnoticeable to unbearable were determined each other quite well. In the similar way to
in preliminary experiments, and they were experiment 1, the estimated discomfort GR
controlled by inserting different ND filters in values were calculated using interpolation
front of the test stimuli. The optical density of when the illuminances at the observer’s eye
each of the ND filters is shown in Table 1(c). were 1 lx, 2.5 lx, 5.0 lx, 7.5 lx and 10 lx, and the
two-way factorial ANOVA was done. There
Lighting pattern
2x2 3x3 5x5

XML Template (2014) [23.4.2014–10:06am] [1–22]
14 T Tashiro et al.
was no statistically significant interaction deviations for all observers, for all conditions.
(p ¼ 0.65) between the curves. Illuminance at Figures 4(d-1) and (d-2) show the relationship
the observer’s eye has an effect obviously, but between the discomfort GR and the two
lighting pattern does not have an effect. The photometric quantities under a 1.0 cd/m2
results for other background conditions are background luminance. The three curves
almost the same. This means that the spatial disagree; with increasing uniformity, the
density of LED elements in a fixed area does not curve shifts toward the larger values along
influence the evaluation of discomfort glare. A the horizontal axis. We conducted statistical
tendency similar to that in experiments 1 and 2 analysis in the similar way to experiment 1.
for the effect of the background luminance was Results showed that diffusion angle, as well as
observed: for the same density and same ND the illuminance at the observer’s eye, has an
filter(s), the discomfort GR decreases as the effect. We carried out the supplementary
background luminance increases. tests, and the results revealed that there is a
statistically significant difference (p50.05)
between 08 and 208 (p ¼ 0.92 104), and 08
5. Experiment 4 and 608 (p ¼ 1.82 106).
5.1. Test stimuli

Experiment 4 examined the effect of the 6. Discussion
uniformity of the light source. We used a
combination of type I in experiment 1 and a 6.1. Difference between uniform and
special diffuser panel, the light-shaping dif- non-uniform light sources
fuser (LSD, Optical Solutions Corp.), as test The results of all four experiments showed
stimuli. The LSD has a microlens array on the that the illuminance at the observer’s eye is a
surface, and the light is diffracted at given good photometric quantity for predicting the
diffusion angles. In this study, diffusion discomfort GR for non-uniform light sources.
angles of 208 and 608 and unfiltered light However, the results of uniform light sources,
(diffusion angle ¼ 08), were used. In addition, type III in experiment 1 and the LSD at 208
seven brightnesses were obtained using ND and 608 in experiment 4, deviate from those of
filters. The optical density of each of the ND non-uniform sources.
filters is shown in Table 1(d). The sizes of the test sources in the four
experiments were 2.58 3.48 in experiments
1 and 4, from 0.48 0.48 to 5.08 5.08 in
5.2. Experimental conditions experiment 2 and 3.88 3.88 in experiment
The conditions are the same as those 3. All the sources except type II in experi-
in experiment 1, except that only background ment 1 were arrays of raw LED elements
luminances of 0.1 cd/m2, 1.0 cd/m2 and 10 cd/ from SEC having similar luminous intensity
m2 were employed, as in experiments 2 and 3. distributions; the results for type II, with a
Eleven observers, six males and five females in covered lens, agree with those for type I.
their twenties (the average age was 21.9 years) Therefore, our results show that within the
with normal colour vision, participated. Each source size range 0.48 0.48 to 5.08 5.08,
photometric quantity was measured using the the illuminance at the observer’s eye might
same method as in experiments 1, 2 and 3. be the photometric quantity best exhibiting
a simple relation with the subjective rating
5.3. Results of discomfort glare. However, the devi-
Table 1(d) shows the mean ratings of ations of the uniform sources remain
discomfort glare and associated standard unsolved. According to Bullough et

XML Template (2014) [23.4.2014–10:06am] [1–22]
al.,19,20 when the size of the source is those of the study by Eble-Hankins and
greater than 0.38, it is necessary to consider Waters.
not only the illuminance at the eye, but
also the luminance of the LED source. The 6.2. Effective glare luminance
minimum size of the test stimulus in our The ultimate goal of this study is to
study is 0.48. Therefore, we need to inves- propose a discomfort GR formula that is
tigate the effect of the luminance of the unaffected by differences in the spatial lumi-
each test stimuli. This will be discussed in nance distribution of the glare source.
detail in the next section. To accomplish this, it is necessary to
In our study, discomfort GRs for non- determine some photometric quantity that
uniform light sources were greater than those can be used to evaluate the discomfort glare
for uniform light sources at the same photo- for LED light sources with various spatial
metric condition. This trend indicates dis- luminance distributions. We re-plotted the
agreement with the results of Eble-Hankins results of all the experiments for background
and Waters.22 The discomfort GRs for luminance of 1.0 cd/m2 against three photo-
sinusoidal stimuli of different spatial fre- metric quantities, the illuminance at the
quencies shown in the study of Eble-Hankins observer’s eye, the average luminance of
and Waters do not vary greatly with spatial the source area and the average luminance
frequency, and a uniform stimulus is selected of the effective area, as shown in Figures 8(a),
most frequently as a more glaring stimulus. (b) and (c), respectively.
In their study, all the stimuli were diffused In outdoor lighting or road lighting, the
light sources reflected by the same paper, rating of discomfort glare is often plotted
even though they had non-uniform patterns against the illuminance at the observer’s eye.
owing to the different spatial frequencies It might be a good index if limited to uniform
printed on the papers. Although it was not sources or non-uniform light sources.
indicated in their study, the illuminance at However, it is not a proper metric when
the observer’s eye of all the stimuli might not uniform and non-uniform light sources are
vary greatly because the average luminance compared directly with each other as shown
of all the stimuli remained almost the same, in Figure 8(a).
and the viewing distance was kept constant. For conventional light sources, such as
Alternatively, the illuminance at the obser- fluorescent lamps, incandescent lamps and
ver’s eye might increase slightly for papers mercury lamps, the luminance of the light
having uniform or nearly uniform patterns, source is often used as a measure of bright-
because loss of light from those papers ness. Because the luminance within the source
would be less than that from papers with area is so inhomogeneous for the stimuli used
low-spatial-frequency patterns that have in this study, we employed the average
wide blackish parts. Also, light reflected by luminance at the source area as a measure
the white parts of the papers in the Eble- of the brightness of the source area and
Hankins and Waters study was diffused, plotted the rating results against it, as shown
whereas light from the LED sources used in in Figure 8(b). However, this quantity is not
our study had a specific luminous intensity appropriate either, because it does not explain
distribution. Thus, the proportion of the the results of experiment 2 in particular.
light beam entering the pupil is considered to Then, as a new candidate, we defined the
be larger than that of diffused light. We average luminance of the effective area. As
consider that these are the main causes of described in Section 3.3, the effective area is
the disagreement between our results and defined as the sum of non-zero pixels in the

XML Template (2014) [23.4.2014–10:06am] [1–22]
16 T Tashiro et al.
(a) 9 Ex1-T1
Background: 1 cd/m2 Ex1-T2
8 Ex1-T3

Ex2-0.2
7
Ex2-0.4
6 Ex2-0.6
Ex2-0.8
5 Ex2-1.0
Ex2-1.4
4
Ex2-1.9
3 Ex2-2.5
Ex3-2x2
2 Ex3-3x3
Ex3-5x5
1
Ex4-0
102 10–1 100 101 102 103
Ex4-20
Illuminance at the observer,s eye (lx) Ex4-60
(b) 9 Ex1-T1
8 Ex1-T3
Ex2-0.2
7
Ex2-0.4
6 Ex2-0.6
Ex2-0.8
5 Ex2-1.0
Ex2-1.4
4
Ex2-1.9
3 Ex2-2.5
Ex3-2x2
2 Ex3-3x3
Ex3-5x5
1
Ex4-0
100 101 102 103 104 105 106 Ex4-20
Average luminance of the source area (cd/m2) Ex4-60
(b) 9 Ex1-T1
8 Ex1-T3
Ex2-0.2
7
Ex2-0.4
6 Ex2-0.6
Ex2-0.8
5 Ex2-1.0
Ex2-1.4
4
Ex2-1.9
3 Ex2-2.5
Ex3-2x2
2 Ex3-3x3
Ex3-5x5
1
Ex4-0
100 101 102 103 104 Ex4-20
Average luminance of the effective area (cd/m2) Ex4-60
Figure 8 Relation between discomfort glare and the photometric quantities (a) the illuminance at the observer’s eye,
(b) the average luminance of the source area and (c) the average luminance of the effective area

XML Template (2014) [23.4.2014–10:06am] [1–22]
luminaire area. The above quantity is thus strong non-linear contribution of luminance
calculated from the sum of luminance in the to discomfort glare.
luminaire area divided by the effective area. The scaling results obtained in all experi-
When the rating results are plotted ments for the background luminance of
against this quantity, the results for uniform 1.0 cd/m2 are plotted against the effective
light sources deviate from the results for glare luminance with WF in Figure 9(a). As
non-uniform light sources, as shown in shown in Figure 9(a), all the curves become
Figure 8(c). closer compared with the results plotted
The marked characteristic observed in against other photometric quantities. The
Figures 8(a) and (c) is the displacement results of other background luminance con-
between the results for non-uniform and ditions have similar trend. Very high coeffi-
uniform sources. To compensate for this cients of determination, 0.938, 0.937
gap, either the results for uniform sources and 0.917 for background conditions of
must be shifted toward smaller values or the 0.1 cd/m2, 1.0 cd/m2 and 10 cd/m2, respect-
results for non-uniform sources must be ively, are obtained between the discomfort
shifted toward larger values along the hori- GR and logarithmic value of effective glare
zontal axis. We chose the latter. We thought luminance with WF. This means the effective
that pixels with high luminance values glare luminance calculated using WF is a
in the luminaire area are more effective candidate to explain the degree of discomfort
than pixels with a low luminance. Thus, glare. However the discrepancy between the
we assumed a weighting factor that is a results of different experiments increases with
monotonically increasing function of the background luminance. So, employing differ-
luminance and defined the value ‘effective ent weighting functions for different back-
glare luminance’ using this weighting ground luminances might bring us even better
function. fits. In addition to that, increasing back-
As explained in Section 2.4, we obtained ground luminance reduces the discomfort GR
the luminance value of each pixel in the as shown in all the results of experiments 1 to
source area using the HDR image. The 4. This suggests that the weighting factor
luminance of each pixel was multiplied by should be decreased with increasing back-
WF(Lij), and then the product values of ground luminance. Thus, we also tried to
Lij WF(Lij) were summed for all pixels in obtain the different exponential functions for
the source area (ij denotes the pixel position the weighting factor for different background
in the HDR image). This value was divided by luminance.
the effective area. We tried to obtain the best
fit for all the results and obtained an expo- WF01 ¼ 7:0 104 L0:605
ij ð8Þ
nential function for the weighting factor as
shown in equation (7). WF1 ¼ 6:0 104 L0:588
ij ð9Þ
WF ¼ 4:0 105 Lij0:811 ð7Þ WF10 ¼ 4:0 104 L0:471

ij ð10Þ
where WF is the weighting factor function for where WF01, WF1 and WF10 are the weighting
all background conditions, and Lij is the factor functions for background conditions of
luminance value of each pixel in the HDR 0.1 cd/m2, 1.0 cd/m2 and 10 cd/m2, respect-
image. The value of WF becomes close to the ively, and Lij is the luminance value of each
unit 0.95 at Lij ¼ 2.5 105 cd/m2, whereas it pixel in the HDR image. It is worth noting
becomes 2.94 at Lij ¼ 106 cd/m2, indicating that the exponent decreases with the increase

XML Template (2014) [23.4.2014–10:06am] [1–22]
18 T Tashiro et al.
(a) 9 Ex1-T1
Background: 1cd/m2 Ex1-T2
8 Ex1-T3
7 Ex2-0.2

Ex2-0.4
6 Ex2-0.6
Ex2-0.8
5 Ex2-1.0
4 Ex2-1.4
Ex2-1.9
3 Ex2-2.5
R2=0.937 Ex3-2x2
2 Ex3-3x3
1 Ex3-5x5
Ex4-0
102 10–1 100 101 102 103 104 105
Ex4-20
Effective glare luminance with WF (cd/m2) Ex4-60
(b) 9 Ex1-T1
Background: 1cd/m2 Ex1-T2
8 Ex1-T3
7 Ex2-0.2
Ex2-0.4
6 Ex2-0.6
Ex2-0.8
5 Ex2-1.0
4 Ex2-1.4
Ex2-1.9
3 Ex2-2.5
R2=0.945 Ex3-2x2
2 Ex3-3x3
Ex3-5x5
1
Ex4-0
10–2 10–1 100 101 102 103 104 105 Ex4-20
Effective glare luminance with WFBL (cd/m2) Ex4-60
Figure 9 Relation between discomfort glare and the effective glare luminance (a) with WF and (b) with WFBL
of the background luminance, indicating that WFBL can also adequately predict the degree
the same luminance is less effective in the of discomfort glare.
higher background condition. In our study, the coefficients of deter-
We call them WFBL here. In Figure 9(b), mination between the discomfort GR and
discomfort GRs for background luminance of the effective glare luminance with WF or
1.0 cd/m2 are plotted against the effective with WFBL are very high. If an effective
glare luminance calculated using WFBL. All glare luminance using a single WF could be
results indicate very high coefficients of a good predictor of discomfort GR for a
determination, 0.947, 0.945 and 0.936 variety of cases, it would be simpler than
for background conditions of 0.1 cd/m2, using WFBL which depends on the back-
1.0 cd/m2 and 10 cd/m2, respectively, between ground luminance. However, better correl-
the discomfort GR and effective glare lumi- ations are obtained using WFBL as shown in
nance. Scatter of the points becomes obvi- Figure 9(b). Which is the more appropriate
ously smaller than in Figure 9(a). Therefore, weighting factor, WF or WFBL, is an issue
the effective glare luminance calculated using to be considered in the future based on

XML Template (2014) [23.4.2014–10:06am] [1–22]
experiments under a wider variety of with WFBL (mUGR2) shown in equation

conditions. (12).
2
Leff 2 !
6.3. Modified UGR mUGR2 ¼ 0:69 log þ 5:51
P2
As indicated in Section 1.3, the degree of
ð12Þ
discomfort glare is quantitatively evaluated
by the UGR, which is defined by equation (2).
We tried to derive a new formula from the where Leff2 is the effective glare luminance
UGR that can fit our experimental results. with WFBL (cd/m2), ! and P are the same as
First, Li in equation (2) is replaced by equation (11).
the effective glare luminance with WF, and The coefficient of determination between
the coefficients are adjusted to give the the experimental values of the discomfort GR
best fit for the scale values obtained in this and the mUGR2 values derived using equa-
study. As a result, we obtained the modified tion (12) is highly positive at 0.879, indicating
UGR with WF (mUGR1) shown in equation that the degree of discomfort glare for an LED
(11). source can be predicted well using the mUGR2
or any other equation based on the same
2
Leff1 ! concept as the UGR. This means that the
mUGR1 ¼ 0:60 log þ 5:41 discomfort glare of light sources with various
LB P2
luminance distributions within a small area
ð11Þ (55.08 5.08) can be predicted by this index
defined using the effective glare luminance.
where Leff1 is the effective glare luminance It is worth noting that the coefficients of
with WF (cd/m2), ! is the solid angle of the determination in the results using WFBL are
luminaire from the observer’s eye (sr), LB is higher than those using WF, while the correl-
the background luminance (cd/m2) and P is ation becomes higher between the discomfort
the Guth position index for the centre point GR and mUGR calculated using WF and
of the luminaire. background luminance LB. This suggests that
The relation between the results of this establishing an index for discomfort GR,
study and the calculated mUGR1 value employing a single WF and keeping LB is
is shown in Figure 10. The coefficient of slightly better than using WFBL and eliminat-
determination between the experimental ing LB.
values of the discomfort GR and the Our results were obtained using a single
mUGR1 values derived using equation (11) luminaire for all the experimental conditions.
is highly positive at 0.887, indicating that Akashi et al. experimented with the discom-
the degree of discomfort glare for an LED fort glare in an office environment and
source can be predicted well using the examined the relationship between the UGR
mUGR1. glare index and the experimental evaluation
Next, Li in equation (2) is replaced by the value.26 In their results, the UGR glare index
effective glare luminance with WFBL, and the and the experimental evaluation value agreed
coefficients are adjusted to give the best fit well with each other when a single glare
for the scale values obtained in this study. In source was used. On the other hand, when
this case, we also removed the LB term several glare sources were used, the experi-
because the effect of the background lumi- mental evaluation value was always
nance is reflected in the weighting factor. As lower than the UGR glare index. Whether
a result, we obtained the modified UGR mUGR1 or mUGR2 is applicable to

XML Template (2014) [23.4.2014–10:06am] [1–22]
20 T Tashiro et al.
6
3
R2=0.887
1
1 2 3 4 5 6 7 8 9
Calculated value from mUGR with WF
Figure 10 Relation between the results of this study and mUGR with WF
discomfort glare evaluation for cases of mul- Second, if the number of elements in a
tiple luminaires should be examined by luminaire is kept constant, changes in the
appropriate experiments. luminaire size within visual angles from about
0.48 to 5.08 do not influence the discomfort
glare.
7. Summary Third, variation in the spatial density of
LED elements in a fixed area (2 2, 3 3 and
We conducted four experiments to investigate 5 5 in an area 3.88 3.88 square) does not
the quantitative relation between the subject- affect the discomfort glare.
ive rating of discomfort glare and the photo- The results of this study showed that for
metric values of LED light sources with relatively small LED sources having a
various spatial arrangements. highly non-uniform spatial luminance distri-
The results are summarized as follows: bution, the illuminance at the observer’s eye
First, the discomfort GR of a non-uniform is a good measure for the discomfort GR.
light source is larger than that of a uniform This is also true for light sources with a
light source if the illuminance at the obser- uniform surface, but the values of the
ver’s eye or the average luminance of the illuminance at the observer’s eye differ,
effective area is at the same level. unfortunately.

XML Template (2014) [23.4.2014–10:06am] [1–22]
To explain the results of uniform and non- 8 Holladay LL. Action of a light-source in
uniform sources together, a new photometric the field of view in lowering visibility. Journal
quantity is introduced in which the luminance of the Optical Society of America 1927; 14:
distribution within a luminance area is con- 1–15.
sidered. Therefore, we propose an effective 9 Nutting PG. Effects of brightness and contrast
in vision. Transactions of the Illuminating
glare luminance that correlated well to the
Engineering Society 1916; 11: 939–946.
subjective degree of discomfort glare and 10 Vos JJ. Glare today in historical perspective:
modified UGR, mUGR1 and mUGR2 using Towards a new CIE glare observer and a new
the effective glare luminance with WF and glare nomenclature: Proceedings of the CIE
with WFBL, respectively as an index for the 24th Session. Warsaw, 1999, pp. 38–42.
discomfort GR. 11 Osterhaus WKE. Discomfort glare assessment
and prevention for daylight applications in
office environments. Solar Energy 2005; 79:
Funding 140–158.
12 Commission Internationale de l’Eclairage.
Discomfort Glare in Interior Lighting. Vienna:
This study was supported by the research CIE, 1995.
fund endowed by the Iwasaki Electric Co., 13 Commission Internationale de l’Eclairage.
Ltd. and Stanley Electric Co., Ltd. Discomfort Glare in the Interior Working
Environment. Vienna: CIE, 1983.
14 Commission Internationale de l’Eclairage.
References CIE Collection on Glare. Vienna: CIE, 2002.
15 Iwata T, Tokura M. Examination of the
1 Rea MS. The IESNA Lighting Handbook limitations of predicted glare sensation vote
Reference and Application. 9th Edition, New (PGSV) as a glare index for a large source:
York: Illuminating Engineering Society of Towards a comprehensive development of
North America, 2000. discomfort evaluation. Lighting Research and
2 Main A, Vlachonikolis I, Dowson A. The Technology 1998; 30: 81–88.
wavelength of light causing photophobia in 16 Commission Internationale de l’Eclairage.
migraine and tension-type headache between Glare Evaluation for Use Within Outdoor
attacks. The Journal of Head and Face Pain Sports and Area Lighting. Vienna: CIE, 1994.
2000; 40: 194–199. 17 Commission Internationale de l’Eclairage.
3 Tatsumoto M, Kajiyama N, Eda T, Kaneko Glare and Uniformity in Road Lighting
M, Kobayashi S, Ayama M, Hirata K. Installations. Paris: CIE, 1976.
Discomfort glare from brake lamps in patients 18 Kimura-Minoda T, Ayama M. Evaluation of
with migraine: Proceedings of the International discomfort glare from color LEDs and its
Symposium on Automotive Lighting. Darmstadt, correlation with individual variations in
Germany, pp. 116–120. brightness sensitivity. Color Research and
4 Vos JJ. Reflections on glare. Lighting Research Application 2010; 36: 286–294.
and Technology 2003; 35: 163–176. 19 Bullough JD. Luminance versus luminous
5 Schreuder D. The human observer; visual per- intensity as a metric for discomfort glare. SAE
ception. Outdoor Lighting: Physics, Vision and Technical Paper. 2011-01-0111, 2011.
Perception. New York: Springer, 2008, pp. 273– 20 Bullough JD, Sweater-Hickcox K. Interactions
311. among light source luminance, illuminance and
6 Harrison W. Glare ratings. Illuminating size on discomfort glare. SAE International
Engineering 1945; Sep: 525–557. Journal of Passenger Cars - Mechanical
7 Holladay LL. The fundamentals of glare and Systems 2012; 5: 199–202.
visibility. Journal of the Optical Society of 21 Kasahara T, Aizawa D, Irikura T, Moriyama
America 1926; 12: 271–319. T, Toda M, Iwamoto M. Discomfort glare

XML Template (2014) [23.4.2014–10:06am] [1–22]
22 T Tashiro et al.
caused by white LED light sources. Journal 24 de Boer JB. Visual perception in road traffic
of Light and Visual Environment 2006; 30: and the field of vision of the motorist. Public
95–103. Lighting. Eindhoven: Philips Technical
22 Eble-Hankins ML, Waters CE. Subjective Library, 1967.
impressions of discomfort glare from sources 25 Hirning M, Coyne S, Cowling I. The use of
of non-uniform luminance. Leukos 2009; 6: luminance mapping in developing discomfort
51–77. glare research. Journal of Light and Visual
23 Matsuda S, Horaguchi K, Yoshikawa T. Environment 2010; 34: 101–104.
Experimental studies on British IES Glare 26 Akashi Y, Muramatsu R, Kanaya S. Unified
Index system. Journal of the Illuminating glare rating (UGR) and subjective appraisal of
Engineering Institute of Japan 1969; 53: 51–55 discomfort glare. Lighting Research and
(in Japanese). Technology 1996; 28: 199–206.

Tashiro2014 PDF

Uploaded by

Copyright:

Available Formats

Tashiro2014 PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Tashiro2014 PDF

Uploaded by

Copyright:

Available Formats

Lighting Research and

The online version of this article can be found at:

The Society of Light and Lighting

Email Alerts: http://lrt.sagepub.com/cgi/alerts

>> OnlineFirst Version of Record - Apr 24, 2014

Downloaded from lrt.sagepub.com at CARLETON UNIV on June 2, 2014

Lighting Res. Technol. 2014; 0: 1–22

Discomfort glare for white LED light

Received 4 November 2013; Revised 19 February 2014; Accepted 25 March 2014

Quantitative relations between discomfort glare evaluation and photometric

1. Introduction indoor or outdoor architectural lighting,

1.1. Types of glare

Downloaded from lrt.sagepub.com at CARLETON UNIV on June 2, 2014

Downloaded from lrt.sagepub.com at CARLETON UNIV on June 2, 2014

Discomfort glare and LED spacing 3

Downloaded from lrt.sagepub.com at CARLETON UNIV on June 2, 2014

Downloaded from lrt.sagepub.com at CARLETON UNIV on June 2, 2014

Discomfort glare and LED spacing 5

Eble-Hankins and Waters examined the 2. Experiment 1

Downloaded from lrt.sagepub.com at CARLETON UNIV on June 2, 2014

Type I Type II Type III

Figure 1 Test stimuli for experiment 1

1 light-blocking board, so as not to affect the

0.4 2000 (Konica Minolta Inc.) in which the lowest

Downloaded from lrt.sagepub.com at CARLETON UNIV on June 2, 2014

Discomfort glare and LED spacing 7

1 4.19 1.17 0.796 3.86 1.58 0.400 2.05 2.21 0.236

1 3.00 1.36 0.518 1.80 0.867 2.36 1.291

1 3.39 2.22 1.462 2.050 2.68 1.360 1.77 1.238

Downloaded from lrt.sagepub.com at CARLETON UNIV on June 2, 2014

Fluorescent lighting Background

Figure 3 Experimental apparatus

Downloaded from lrt.sagepub.com at CARLETON UNIV on June 2, 2014

Discomfort glare and LED spacing 9

Downloaded from lrt.sagepub.com at CARLETON UNIV on June 2, 2014

Discomfort glare rating

Discomfort glare rating

Discomfort glare rating

Discomfort glare rating

Lighting Res. Technol. 2014; 0: 1–22

Downloaded from lrt.sagepub.com at CARLETON UNIV on June 2, 2014

Discomfort glare and LED spacing 11

Visual angle of the pitch length

Figure 5 Test stimuli for experiment 2

Downloaded from lrt.sagepub.com at CARLETON UNIV on June 2, 2014

Visual angle of pitch length (degree)

3 Background: 0.1 cd/m2

Visual angle of pitch length (degree)

Lighting Res. Technol. 2014; 0: 1–22

Downloaded from lrt.sagepub.com at CARLETON UNIV on June 2, 2014

Discomfort glare and LED spacing 13

The number of zero-value pixels increases 4.2. Experimental conditions

Figure 7 Test stimuli for experiment 3

Lighting Res. Technol. 2014; 0: 1–22

Downloaded from lrt.sagepub.com at CARLETON UNIV on June 2, 2014

5.1. Test stimuli

Downloaded from lrt.sagepub.com at CARLETON UNIV on June 2, 2014

Discomfort glare and LED spacing 15

Downloaded from lrt.sagepub.com at CARLETON UNIV on June 2, 2014

WF ¼ 4:0 105 Lij0:811 ð7Þ WF10 ¼ 4:0 104 L0:471