Luminance Uniformity Analysis Based On Digital Camera Measurements

Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

Luminance Uniformity Analysis

Based on Digital Camera Measurements


Alexandru Viorel Rusu, Dorin Dumitru Lucache,
Gheorghe Livint Cătălin Daniel Gălățanu
Faculty of Electrical Engineering Faculty of Civil Engineering
Technical University “Gh.Asachi” Technical University “Gh.Asachi”
Iasi, Romania Iasi, Romania
[email protected] [email protected]
[email protected]

Abstract — This paper presents a new method for calculating For some time digital cameras luminance measurement are
luminance uniformity on the field. The first step of the study is more known and used in lighting measurements, although this
doing the luminance measurements by using a Digital Camera. type of measurement is not standardized yet. There are many
The second step of the study consists in editing the acquired image papers that have proved the acuity and the superior speed of the
to separate the calculation area for which the uniformity is to be
method [3], [4] in comparison to the classical spot
studied. Last but not least the final step of the study is to calculate
the uniformity. A graphical programme was used for editing of the measurements.
picture, while the calculations are done with a script in MATLAB One parameter that must be measured is the overall
R2020. The pictures have been acquired on the Independence uniformity (U0) which measures the variation of luminance and
Boulevard from Iasi, Romania. indicates how well the road surface serves as a background for
road markings, objects and other road users.
Keywords—calculation surface; average luminance; minimum
luminance = (1)
Even if not exactly in the same way, a similar value of
I. INTRODUCTION surface uniformity appears in most of the street lighting
standards [5-8]. The luminance surface uniformity is one of the
conditions that needs to be satisfied for road lighting straight
In street lighting design the most important parameter taken road design. Because of this, it is important to validate the
into consideration is the luminance. This value is the result of uniformity value by measuring it on the field.
the incident light intensity and the reflection properties of the This paper presents a method of calculating uniformity in
road surface. The average luminance (Lav) reflects the general street lighting measurements by using images made with a
luminance level at which the driver performs. The problems calibrated DSLR camera. The camera used for the study is a
regarding uniformity and discomfort glare are an important Nikon D5300 but any DSLR camera can be used after a specific
subject to study even for indoor lighting. For this reason some calibration. The method was exemplified with a study case in
authors have tried to find solutions to improve as much as which, the influence of the vegetation (more general,
possible these parameters even from the design stage [1]. landscape) on the luminance uniformity values was proved.
Considering that for indoor lighting point by point luminance
measurement are easy to do and that the surface and obstructive II. UNIFORMITY CALCULATION METHOD BASED ON
light are reduced, it is not necessary to do the entire surface LUMINANCE MEASUREMENTS DONE WITH DIGITAL
measurement. CAMERAS
In road lighting, the energy consumption criteria lead to low
lighting levels compared to other spaces. To increase visual Using Digital Cameras for Luminance measurements are
available for quite some time. In the present paper, for the
acuity and ameliorate glare, a proper solution is to increase the
measurements on the field, the digital camera used was Nikon
luminance contrast.
D5300. The calibration function given in (2) was previously
At the moment there are standardized measuring equipment, determined in the laboratory using a luminance standard and an
and methods [2] that are used for a long time for spot luminance integrating sphere [9].
measurements. This way of measuring is very difficult to do
because of the lack of access. One should wait for the k(x)NIKON=6.9676E-12x5-3.2643E-09x4+5.1211E-07x3-
carriageway to be free of any cars from any directions. For some 2.5702E-05x2+5.1953E-06x+8.4553E-02 (2)
roads, which are very crowded, this would be almost impossible In (2), x represents the gray level value, obtained in
because doing luminance measurements correctly involves MATLAB R2020 by using rgb2gray or a different
closing the traffic. transformation function.
Knowing the optoelectronic conversion function (OCF), the photo editor one can transform all the surrounding area in white,
file transformation from gray to luminance was performed, for as seen in Fig.2. So, the color code of the calculation surface can
the next parameters: be easily separated from the rest of the area.
t = 2 – Exposure time, seconds; Choosing a photo editor must be done very carefully. Some
of the photo editors make a blurry font on the edges of the brush.
f = 5 – Aperture number (f-stop); Thus, very small values of Lmin appear at the limit with white and
S = 100 – ISO sensitivity of the film; compromise the results. We opted for Microsoft Paint, which
helps to do the work at a reasonably fast speed.
Using the presented input data, one can see the night scene
of Independence Boulevard from Iasi, Romania, was obtained
and represented in Fig. 1.
A common situation that leads to differences between design
calculation values and the measured ones is the presence of
vegetation or street furniture that blocks the light from the street
lighting.
This situation is also found in Fig. 1 where one can see how
the vegetation significantly reduces the luminance level near the
border of the carriageway. According to the paper scopes, the
influence of the tree branches on the luminance level in this Fig. 2. Calculation surface 1
particular area will be assessed. Taking into account that
uniformity is calculated based on the minimum and average The second step for calculating the uniformity is
value, the inclusion of this shadow area into the calculation will transforming the RGB matrix into a luminance matrix by using
considerably decrease the uniformity value. For a better an optoelectronic conversion function one can get the luminance
understanding of the luminance and uniformity on the values of the entire picture taken with the camera.
carriageway, two situations will be studied by means of two case But the simple conversion it is not enough to have good
studies considering different calculation surfaces. results. There are very small values of luminance that appears on
In the first case, the surface taken into account represents the the luminance map. The minimum value measured is
entire carriageway, from border to border. In the second case, 0.021 cd/m2 which for sure is interpretable. This value appears
considering the big shadows in the left side of the carriageway, due to some errors of the camera (noise in pictures). The solution
the surface taken into account will rule out this area, to evaluate to remove this error is a mean filter that calculates the mean
only the surface lit directly. As consequence, the uniformity value of a matrix of pixels (kernel) around the calculated value.
results will not be influenced by shadows. After several tests by applying this method using different
number of elements, the number of pixels considered in kernel
was established to 25. Figs. 3 show different luminance maps
before and after filtering with different number of pixels in the
kernel.
The number of elements that should be taken into calculation
is variable and dependent on the picture itself and the asphalt
quality and usage. However, if the number is too small, the error
will not be removed. Analysing Figs. 3, it appears that 121 pixels
offer a judicious filtering of the studied night scene, while 441
pixels filtering led to a very blurry and practically unusable
picture.
Fig. 4 presents the luminance map (cd/m2) for the night scene
of the calculation surface 1 after conversion and filtering. It is
obvious that the luminance significant differs on the left side,
near de border, with the rest of the area. This results in a low
Fig. 1. Night scene of Independence Boulevard, Iasi City luminance uniformity level, which does not comply with the
standards. This non-compliance could be determined by at least
The first step in any lighting surface calculation, regardless
one of the following factors: poor design, bad equipment,
of the method used, for design or measurement purpose, is to
obstructive landscape vegetation etc.
separate the calculation surface. In this work, the separation has
been done by editing the pictures taken by the camera. Using a
Fig. 3.a. Luminance map (cd/m2) for the night scene of Fig. 3.b. Luminance map (cd/m2) for the night scene of
Independence Boulevard (original) Independence Boulevard (round off with 25 elements)

Fig. 3.c. Luminance map (cd/m2) for the night scene of Fig. 3.d. Luminance map (cd/m2) for the night scene of
Independence Boulevard (round off with 121 elements) Independence Boulevard (round off with 441 elements)

limits are set properly, this value can be determined just by


examining the luminance map. For example when looking at
Fig.4 one can see that the maximum luminance value is between
6 and 7 cd/m2, so the value can be set to 7. If anyone wants to
determine exactly the maximum value of luminance it can be
done just by calculating the maximum value smaller than 7. For
luminance uniformity calculation this is not really relevant, the
important thing being that the threshold value is bigger than the
maximum luminance on the luminance map generated based on
the calculation surface.
After analyzing Fig. 4 the threshold value was set at 7 value
which seems to be bigger than the maximum luminance value.
According to (1), to calculate U0 first Lmin and Lav have to be
determined. Applying the proposed method, the results are
centralized in Table I. It can be noticed that Lmin is much smaller
than it should be. This is due to the vegetation which
overshadows the margin area near the border of the carriageway.
Fig. 4. Luminance map (cd/m2) for the night scene of
Independence Boulevard calculation surface 1 To better understand the luminance differences between the
directly lit carriageway in comparison to the shadowed area, a
After having the matrix of the luminance map, one can luminance profile was extracted (Fig. 6) following the direction
separate the values that are bigger than a threshold value. If the of the line drawn in Fig. 5. Table II centralizes the luminance
values in 8 points along the considered luminance profile (vegetation) and not the lighting system itself. Excluding any
(x represents the MATLAB R2020 notation for the element area from the calculation surface brings with it measurements
number in the matrix and y represents the luminance value in error, but this error will be much smaller than that generated
that particular point). by the shadows made by the vegetation on the road.
TABLE I. PHOTOMETRIC RESULTS FOR CALCULATION SURFACE 1 TABLE II. MEASUREMENT POINTS VALUES CENTRALIZATION
Photometric Results Element Number Y
Crt. Point
Nr. Lav Lmin x [cd/m2]
2 2 U0
[cd/m ] [cd/m ] 1. 17 0.659
1. 0.556 0.112 0.2
2. 245 0.226
3. 407 0.506
4. 875 0.583
5. 1,272 0.542
6. 1,404 0.146
7. 1,554 0.398
8. 1,643 0.294
Using this new calculation surface, the entire process is
repeated. Fig. 7 shows an overview of the area considered in the
second iteration, and Fig. 8 the corresponding luminance map.
A higher luminance uniformity can be observed, the luminance
values being closer to the ones expected. A solution to reduce
the impact generated by the vegetation is cutting the branches as
often as possible, a task that falls under the city hall
responsibility.

Fig. 5. Luminance map (cd/m2) for the night scene of


Independence boulevard, calculation surface 1, where A-B is
the direction of the luminance profile
Results show that from point 1 to point 2 the luminance
drops with more than 60% and from point 5 to point 6 with
almost 75%. This means that for a driver passing near the border
the visibility is totally compromised, a possible obstacle in that
area would not be observed. This is the results of too low values
of minimum luminance as result of the vegetation presence Fig. 7. Calculation surface 2
influence. Comparing the results given by Tables II and III, one notice
that luminance uniformity rise, from a value of U0=0.2 to
U0=0.398, which is almost equal to the recommended minimal
value of 0.4 [1]. By removing the shadowed area from the
calculation surface, not only the uniformity is better but the
average luminance value improves as well.

Fig. 6. Luminance profile on the direction A-B

Considering the very large luminance fluctuations


Fig. 8. Luminance map (cd/m2) for the night scene of
between two measuring points at a very short distance, it is
Independence Boulevard calculation surface 2
clear that these differences are due to external factors
TABLE III. PHOTOMETRIC RESULTS BASED ON FIELD The proposed method can be used for both evaluating the
MEASUREMENTS FOR CALCULATION SURFACE 2 existing street lighting systems and more detailed assessment at
Photometric Results the reception of the works done by contractors. This is
Crt.
Nr. Lav Lmin particularly important now, when LED lamps are used in
U0
[cd/m2] [cd/m2] modernization programs on a large scale.
1. 0.619 0.246 0.398
REFERENCES
III. CONCLUSIONS [1] Woon-Jo Byun et. al., “Design of Lighting Control System Considering
This paper proposes a determining method of the real Lighting Uniformity and Discomfort Glare for Indoor Space”,
luminance uniformity on a surface by using luminance International Conference on Platform Technology and Service (PlatCon),
29-31 January 2018, Jeju, Korea
measurements done with a calibrated DSLR camera. Few
[2] BS EN 13201-4:2016, “Road lighting - Part 4: Methods of measuring
conclusions can be drawn from the present study. lighting performance”
First, the proposed method is very easy to be implemented. [3] Cătălin-Daniel Gălățanu et. al., ”Imaging Measurements for Public
The most important advantage of this method consists in the Lighting Predictive Maintenance”, 978-1-7281-0101-9/19, The 11th
very big number of measurement points used for the average International Symposium on Advanced Topics in Electrical Engineering,
luminance calculation. This number is limited only by the March 28-30, 2019, Bucharest, Romania
camera resolution (in our case is 6000x4000), which is much [4] Alexandru Viorel Rusu, Dorin Dumitru Lucache, Gheorghe Livint,
higher than the one used by project designers for surfaces “Assessment of Conflict Area Parameters Measurements by Using
Photographic Methods”, SIELMEN 2019, 12th International Conference
calculation. Another advantage is the possibility to place the and Exhibition on Electromechanical and Energy Systems Chișinău,
observer (camera) in any desired position. Republica Moldova, October 10 – 11, 2019
Second, even if in the present paper it was used for a [5] BS EN 13201-2:2015, “Road lighting - Part 2: Performance
carriageway, the method can be used for determining the real Requirements”, BSI, 978-0-580-80625-4
luminance uniformity of any surface, such as sports fields. [6] NP 062: 2002, “Standard for the design of road and pedestrian lighting
systems”
Third, the study case shows the influence of vegetation on
[7] “Manual for Street Lighting Facilities and Planning”, April 2014
the lighting efficiency. In shadowed areas, the luminance value
dropped with up to 75%, causing potential danger for car [8] ANSI/IESNA RP8–14, “Roadway Lighting”
accident when passing through that area. It results the [9] C.D. Galatanu, “Luminance measurements for light pollution
assessment”, International Conference on Electromechanical and Power
importance of the landscape influence on the lighting quality and Systems (SIELMEN), 2017, Pages: 456 – 461
the necessity to avoid the interreference between landscape and
road lighting.

You might also like