Content uploaded by Norifumi Kawabata
Author content
All content in this area was uploaded by Norifumi Kawabata on Dec 25, 2019
Content may be subject to copyright.
Multi-view 3D CG Image Quality Assessment for Contrast Enhancement
Including S-CIELAB Color Space in case the Background Region is Gray Scale
Norifumi Kawabata1and Masaru Miyao2
1 2 Graduate School of Information Science, Nagoya University
Furo-cho, Chikusa-ku, Nagoya-shi, Aichi 464–8603, Japan
E-mail : {1norifumi, 2miyao}@nagoya-u.jp
Abstract: In this paper, we experimented the subjective eval-
uation for 3D CG image including the gray scale region with
8 viewpoints parallax barrier method, and we analyzed this
result statistically. Next, we measured about the relation be-
tween the luminance change and the color difference by using
S-CIELAB color space and CIEDE2000. As a result, we ob-
tained knowledge about the relation among the coded image
quality, the contrast enhancement, and gray scale.
Keywords— Contrast Enhancement, H.265/HEVC, Gray Scale, S-
CIELAB Color Space, CIEDE2000
1. Introduction
Thus far, there were studies for image quality assessment
based on S-CIELAB color space including the spatial fre-
quency characteristics of human vision system [1], [2]. We
also carried out the quality assessment of multi-view 3D im-
age [3], multi-view 3D image for contrast enhancement [4],
and multi-view 3D image for contrast enhancement of the ob-
ject region [5] based on S-CIELAB color space. As contents
creators, we consider that they need not only to consider the
coded degradation but also not stand out the background. We
consider that it is possible for these to achieve by using gray
scale in the background. In the 3D images including the lu-
minance that affected to both the stereoscopic effect and the
evaluation result, it is not clear that assessors are able to ac-
cept how degradation or luminance by whether background
region is gray scale or not. Therefore, we need to verify about
these problems. In this study, first, we carried out the qual-
ity evaluation experiment of 3D CG image encoded and de-
coded by H.265/HEVC for contrast enhancement in case the
background region is gray scale, and then, we analyzed these
results and classified for the luminance change and the coded
image quality (Exp. 1). Next, we measured the luminance L∗
objectively by transforming to S-CIELAB color space (Exp.
2). Finally, we measured the color difference objectively by
using CIEDE2000 (Exp. 3), and compare among results.
2. Image quality evaluation experiment
2.1 3D CG images used in this study
In this study, we used 3D CG contents “Museum,” “Wonder
World” provided by NICT [6] for free of charge as shown in
Fig. 1 (a)-(f) in order to carry out the image quality evalua-
tion experiment. As a generation procedure, we constructed
CG cameras by the number of viewpoints on the virtual space
generated the HD quality still image by 8 viewpoints. And
then, we carried out the region division for the object and the
background in still images by 8 viewpoints, and processed
Figure 1. 3D CG images used in this study
gray scale transformation to the background region image.
After that, we composed each regions again. Next, we carried
out the contrast enhancement for generated images by carry-
ing out the processing of Adaptive Histogram Equalization
(AHE). On the other hand, we prepared 60 patterns of im-
age sequences including the Quantization Parameter (QP =
0(ref ),20,30,40,and 51) by H.265/HEVC, the Luminance
change parameter (Lum = 0(ref ),0.25,0.5,1,2,4), and 3D
CG contents.
2.2 S-CIELAB color space and CIEDE2000
Figure 2 shows the flowchart for the process from loading an
image including gray scale background to transforming the S-
CIELAB color space and CIEDE2000 [1]. We have explained
the S-CIELAB color space and CIEDE2000 in Appendices A
and B, therefore, please see the explanation of Appendices
A and B in detail [1]. We show briefly from the following.
Finally, L∗, a∗, and b∗are shown in Eq. (1)–(3).
L∗=
116 Y
Yn
1
3
−16 Y
Yn
>0.008856
903.3Y
Yn(otherwise)
(1)
a∗=
500 X
Xn
1
3
−Y
Yn
1
3X
Xn
,Y
Yn
>0.008856
500 ∗7.787 X
Xn+16
116−7.787 Y
Yn+16
116
(otherwise)
(2)
The 31st International Technical Conference on Circuits/Systems,
Computers and Communications (ITC-CSCC 2016)
579
Figure 2. Flowchart from transformation of S-CIELAB color
space to CIEDE2000
Figure 3. DSIS method
Table 1. EBU quality
scale
Score Quality
5 Imperceptible
4 Perceptible, but not annoying
3 Slightly Annoying
2 Annoying
1 VeryAnnoying
b∗=
200 Y
Yn
1
3
−Z
Zn
1
3Y
Yn
,Z
Zn
>0.008856
200 ∗7.787 Y
Yn+16
116−7.787 Z
Zn+16
116
(otherwise)
(3)
Finally, ∆E00 is shown in Eq. (4), (5).
∆E00 =L2+C2+H2+ (RTCH )(4)
L=∆L′
KL·SL
, C =∆C′
KC·SC
, H =∆H′
KH·SH
(5)
2.3 Experimental content and assessment method
In this experiment, we used “Newsight” 3D display developed
by Newsight Corporation [7].
As an assessment method of subjective quality evaluation
experiment, from Fig. 3, first, we displayed reference image
A for 10 seconds, and then mid-gray image G for 3 seconds.
Next, we displayed test condition image B for 10 seconds.
Subsequently, the assessor evaluated this cycle and inputted
the assessment value into the computer application, which re-
quired 10 seconds. This cycle was defined one set, we re-
peated for an assessor until finishing the last set (five sets). As
an experimental environment, we carried out the experiment
based on ITU-R BT.500 [8], [9]. From Table 1, the asses-
sor assigned evaluation scores according to five ranks (MOS
(MMO S , WM OS )). Here, we defined M OS = 4.5,3.5,and
2.5as “DL (Detective Limit),” “AL (Acceptability Limit),”
and “EL (Endurance Limit),” respectively. For the objec-
tive assessment, as shown in Fig. 2, first, we decomposed
the reference image and the coded image into R, G, and B
components. Next, we transformed these component images
to S-CIELAB color space, and we measured the luminance
of each image sequences and the color difference by using
CIEDE2000.
3. Experimental results and discussion
3.1 Result of Exp. 1
Figure 4 shows MOS in the vertical axis, QP in the hor-
izontal axis. The error bar is extended vertically from the
plot points in Fig. 4, which shows 95% confidence interval.
From experimental result, we can verify the same tendency
in L= 0.25,4.00. In L= 1.00,MM OS and WMO S shifted
before and after “AL” except for L=ref since they are more
than 3. On the other hand, when QP ≥40 is satisfied, we can
verify the decline of assessment value rapidly.
3.2 Statistical analysis by using Support Vector Machine
From results of subjective assessment, Table 2 and 3 show
SVM result by using SMO algorithm of Weka 3.6 [10]. “Pre-
cision,” “Recall,” and “F-Measure” values greater than 0.7 are
denoted in boldface font. Table 4 shows SVM correctly clas-
sified percentage. From Table 2, for Class (Lum), Precision
of “lum 2,” “lum ref,” Recall of “lum 0.25” are more than
0.7, however, correctly classified percentage in Class (Lum) is
less 40%. Therefore, we can judge as “not classified.” On the
other hand, from Table 3, for Class (QP), most of “QP ref,”
“QP 51” are boldface font because of more than 0.7, and cor-
rectly classified percentage in Class (QP) is also more than
0.7. Therefore, we can judge as “Classified.”
3.3 Result of Exp. 2 and 3
From Fig. 5, as a whole, in “Museum,” L∗shifted between 75
and 93. On the other hand, in “Wonder World,” L∗shifted be-
tween 24 and 32. Therefore, the luminance difference among
L∗in “Museum” is larger than those in “Wonder World.”
From Fig. 6, as a whole, the color difference ∆E00 in “Won-
der World” are larger than those in “Museum.”
580
Figure 4. Result of Exp. 1 (M OS)
Table 2. SVM (Lum)
3.4 Discussion
From experimental result of subjective assessment, in case
the luminance parameter is higher or lower, assessment value
tends to low, and in the case of nearly L= 1, assessment
value tends to high. From this, we consider that when the
Table 3. SVM (QP)
luminance change is more than a few, assessment values are
affected. From results of SVM, correctly classified percent-
age in “QP” is higher than that in “Lum.” From this, we con-
sider that it is easy for the assessor to perceive change of the
Quantization Parameter than that of the luminance.
581
Table 4. SVM (Correctly Classified Percentage)
Figure 5. Result of Exp. 2 (L∗)
4. Conclusion
From experimental results in this study, we obtained knowl-
edge that it is easy for the assessor to affect the coded image
quality than the luminance change.
Acknowledgment
This study was carried out based on the Grant-in-Aid for Sci-
entific Research (B) 24300046, and the research grant for
Ph.D. student at Nagoya University. This international con-
ference paper is included improvement based on contents of
our domestic conference proceedings [3], [11].
References
[1] G. M. Johnson, M. D. Fairchild, “A Top Down Description of S-
CIELAB and CIEDE2000,” Color Research & Application, vol. 28,
no. 6, pp. 425–435, June 2003.
Figure 6. Result of Exp. 3 (∆E00 )
[2] J. Bai, T. Nakaguchi, N. Tsumura, and Y. Miyake, “Evaluation of Im-
age Corrected by Retinex Method Based on S-CIELAB and Gazing
Information,” IEICE Trans. Fundamentals, vol. E89-A, no. 11, pp.
2955–2961, November 2006.
[3] N. Kawabata and M. Miyao, “3D CG Image Quality Evaluation In-
cluding S-CIELAB Color Space with 8 Viewpoints Glassless 3D,”
Proc. of the 2015 IEICE General Conference, A21-2, pp. 292, March
2015 (in Japanese).
[4] N. Kawabata and M. Miyao, “3D CG Image Quality Assessment
for the Luminance Change by Contrast Enhancement Including S-
CIELAB Color Space with 8 Viewpoints Parallax Barrier Method,”
Proc. of The 1st International Conference on Advanced Imaging (1st
ICAI2015), T107-01, pp. 632–635, June 2015.
[5] N. Kawabata and M. Miyao, “3D CG Image Quality Metrics for the
Contrast Enhancement of the Object Region Including S-CIELAB
Color Space with 8 Viewpoints Parallax Barrier Method,” IEICE Tech.
Rep., vol. 115, no. 48, IMQ2015-4, pp. 17–22, May 2015.
[6] http://www.nict.go.jp/, accessed May 25, 2016.
[7] Newsight Japan, http://newsightjapan.jp/,
accessed May 25, 2016.
[8] ITU-R BT.500-13,
https://www.itu.int/rec/R-REC-BT.500/en,
accessed May 25, 2016.
[9] N. Kawabata and M. Miyao, “3D CG Image Quality Metrics by Re-
gions with 8 Viewpoints Parallax Barrier Method,” IEICE Trans. on
Fundamentals, Vol. E98-A, No. 08, pp. 1696–1708, August 2015.
[10] Weka, http://www.cs.waikato.ac.nz/ml/weka/,
accessed May 25, 2016.
[11] N. Kawabata and M. Miyao, “3D CG Image Quality Metrics Including
Gray Scale Region with 8 Viewpoints Parallax Barrier Method,” The
2015 ITE Annual Conference, pp. 32B-2-1–32B-2-2, August 2015 (in
Japanese).
582
