2 : Retinex (Regular Paper) 17 5, 2012 9 (JBE Vol. 17, No. 5, September 2012) http://dx.doi.org/10.5909/jbe.2012.17.5.851 ISSN 1226-7953(Print) Retinex a), b), c) Color Improvement of Retinex Image Using the Maximum Color Difference Signal Table Jae-Won Lee a), Jee-Hoon Jung b), and Sung-Hoon Hong c) Retinex. Retinex RGB. YCbCr Retinex,... Abstract Retinex algorithm enhances the contrast of image through visibility improvement. However, the conventional Retinex methods may produces color distortions due to error of hue representation and over-saturation since the methods work in RGB color space. In this paper, we propose a new Retinex algorithm with color correction, which improves contrast by using MSR(Multi-Scale Retinex) working in YCbCr color space and adaptively compensates the color saturation based on the maximum color difference table. Our algorithm maps the color difference signals to the correct gamut to prevent over-saturation phenomenon by considering the correlation between luminance and hue dependent saturation. Simulations results show that the proposed method gives better color improvement compared to the conventional methods. Keyword : Retinex, MSRCR, color enhancement, over-saturation a) (Division of electronic computer Engineering, Chonnam National Univ.) b) LG (LG Innotek Ansan Institute) c) () School of Electronics & Computer Engineering, Chonnam National Univ. (Information & Telecommunication Research Institute) Corresponding Author : (Sung-Hoon Hong) E-mail: hootybaby@hanmail.net Tel: +82-62-530-0820, +82-62-530-1758 Fax: +82-62-530-1809 Manuscript received July 16, 2012 Revised September 17, 2012 Accepted September 17, 2012
(JBE Vol. 17, No. 5, September 2012) I. (Resolution) Dynamic Range. (Pixel) (Luminance) (Chroma).. Dynamic Range. Dynamic Range Dynamic Range [1][2]. Retinex Retinex contrast [3]~[6]. Retinex Jobson, SSR(Single-Scale Retinex) [4][5]. SSR C/S (center/surround model) (color constancy). SSR contrast (halo effect). MSR(Multi-Scale Retinex) SSR, SSR [6]. MSR RGB RGB. MSR Rahman MSR RGB MSRCR(Multi-Scale Retinex with Color Restoration) [7]. RGB RGB. Numan Retinex [8]. MSR RGB RGB MSRCR,. Hanumantharaju MSRCR RGB [9]. RGB RGB, RGB RGB. RGB Retinex Lee YCbCr Retinex, Cb Cr [10]. Cb Cr, RGB, Cb Cr. Retinex.
2 : Retinex.. 2 Retinex, 3. 4, 5.. Gaussian SSR, SSR Gaussian, 1. SSR MSR. MSR MSRCR(Multi-Scale Retinex with Color Restoration) [7].. Retinex Jobson (1) SSR (Single-Scale Retinex) [4][5]. log log R, G, B, SSR, Convolution. Gaussian Center/Surround. SSR Gaussian. MSR(Multi-Scale Retinex), MSR (2) Gaussian SSR, [6]. MSRCR, MSR. (4). log 와 gain, S,. MSRCR MSR. Numan (5) RGB RGB [8]. MSR, Gaussian R, G, B max, Retinex
(JBE Vol. 17, No. 5, September 2012). RGB MSRCR. Hanumantharaju (6)-(9) Reti- nex [9]., gain offset. (7). (8) (9). log MSR Retinex. MSRCR RGB, RGB RGB.. Retinex MSRCR RGB MSR. MSR,. MSRCR MSRCR gain gain. YCbCr YIQ MSR, MSR. Lee YCbCr (10) Retinex [10]., MSR YCbCr. Cb, Cr Y Retinex. Cb Cr. YCbCr MSR (Cb, Cr), (Cb, Cr) LUT(Look-Up Table) (over-saturation)
2 : Retinex. 1. LUT(Look-Up Table). 1 YCbCr (Saturation) (Hue), (gamut).,, 1. YCbCr Fig 1. YCbCr Color coordinates.. YCbCr (over-saturation) (Y) (Hue) (Saturation). 2 (a) (0 360 ) (0255) (0~255) 0( ) 255()., 2 (a). 2 (b) (c) (a) (Cb, Cr). Cb Cr (b) (c).. LUT(Look Up Table), LUT (a) (b) (c) 2.. (a), (b) Cr, (c) Cb Fig 2. maximum range of chrominance signal. (a) maximum range of saturation, (b) maximum range of Cr, (c) maximum range of Cb
(JBE Vol. 17, No. 5, September 2012). LUT. RGB 0 255 (Hue), 0 255, (Hue) 2563., RGB,, Cb, Cr (Cb_Max, Cr_Max) LUT. RGB (Y) (Cb, Cr), (H) (S) (11) (12) [11]. 3. Fig 3. Calculation of Weight factor for change of Y cos min LUT 2,. tan- 1(Cr/Cb) Cb Cr. 2. LUT Retinex Retinex.. 3 LUT (Hue), Retinex Y_in Y_out. 3 A B. (Y_out/Y_in) D. (C_max_out/C_max_in) C. (Hue) Cb Cr.. (Cb, Cr) (13). min max max max max
2 : Retinex Cb_max_in Cr_max_in LUT (Cb_in, Cr_in) (Y_in), Cb_max_out Cr_max_out LUT (Cb_in, Cr_in) (Y_out). 4. Fig 4 Block diagram of Proposed algorithm 4 LUT Reti- nex. (Y_in) MSR (Y_out). LUT, Gain (13) α β. β (14), (15). min max max IV.. LUT 2563, LUT 1 360, 256. LUT 360 256 byte Cr_Max, Cb_Max 256 256 byte Hue. MSRCR. 5 (a), (b) +60. 195 255. (b). (c) MSRCR,. (d). LUT Retinex Retinex MSRCR Numan, Hanumantharaju Lee. MSRCR α β, Hanumantharaju G(gain=2.25) b(offset=-30)
(JBE Vol. 17, No. 5, September 2012) (a) (b) (c) (d) 5. (a), (b), (C) MSRCR, (d) Fig 5. result of saturation enhancement (a) original image, (b) intensity changed image, (c) MSRCR method, (d) Proposed method G(gain=2.0) b(offset=0). 6. MSRCR R, G, B (a) (b) MSRCR (c) Numan s (d) Hanumantharaju `s (e) Lee s (f) 6. Retinex Fig 6. result of Retinex algorithm in dark image
2 : Retinex. Numan MSRCR. Hanumantharaju MSRCR R, G, B. Lee., MSR. 7 contrast. (a) (b) MSRCR (c) Numan s (d) Hanumantharaju `s (e) Lee s (f) 7. Contrast Retinex Fig 7. result of Retinex algorithm in high contrast image (a) (b) MSRCR (c) Numan s (d) Hanumantharaju `s (e) Lee s (f) 8. Retinex Fig 8. result of Retinex algorithm in back-light image
(JBE Vol. 17, No. 5, September 2012). contrast,. 8 Retinex. Numan. Numan. MSRCR, Hanumantharaju Lee.,. 9..... Contrast Visual Contrast Measure. VCM. 1. Block. 2. Block. 3. VCM ( 4-1)., Block, Block. Contrast. 4.1 40X40, 8 VCM., VCM. (a) (b) MSRCR (c) Numan s (d) Hanumantharaju `s (e) Lee s (f) 9. Retinex Fig 9. result of Retinex algorithm in back-light image
2 : Retinex 1. VCM Table 1. result of VCM comparison MSRCR Hanumantharaju s Numan s Lee s 6 25.52 61.46 59.38 58.85 61.98 61.46 7 57.81 55.73 50.52 53.65 53.51 61.46 8 48.96 94.79 94.79 88.02 95.31 95.83 9 29.17 72.40 75.52 79.17 79.71 75.52 40.37 71.09 70.05 69.92 72.63 73.57 Contrast..,, [1]. (17) RGB. min R, G, B Red, Green, Blue. 2 (17). 2.... RGB Retinex.. (H),. (18). 2. Table 2. result of saturation comparison MSRCR Hanumantharaju s Numan s Lee s 6 87.41 60.88 82.89 78.68 80.63 91.73 7 135.59 49.68 73.72 85.21 86.93 134.00 8 118.87 115.89 143.58 132.23 132.19 123.69 9 88.00 84.19 107.59 83.31 86.41 139.52 107.47 77.66 101.94 94.86 96.54 122.24
(JBE Vol. 17, No. 5, September 2012) 3. Table 3. result of color change comparison MSRCR Hanumantharaju s Numan s Lee s 6 17.79 9.43 3.54 4.97 0.9612 7 53.51 26.44 6.25 7.22 0.4689 8 20.12 11.37 2.58 7.02 0.9531 9 26.56 16.49 6.02 10.77 1.9177 29.50 15.93 4.60 7.50 1.08 cos R, G, B Red, Greed, Blue. 3 (18). 3. RGB RGB Retinex. Lee.. YCbCr, gamut LUT... YCbCr. Retinex. V. Retinex contrast. R,G,B,. Retinex. RGB [1] M. Y. Lee, C. H. Son, J. M. Kim, C. H. Lee and Y. H. Ha, Illumination-Level Adaptive Color Reproduction Method with Lightness Adaptation and Flare Compensation for Mobile Display, Journal of Imaging Science and Technology. Vol.51, No. 1, pp. 44 52, 2007. [2] B. Wandell, P. Catrysse, J. DiCarlo, D. Yang and A. E. Gamal, Multiple Capture Single Image with a CMOS Sensor, Chiba Conference on Multspectral Imaging, pp. 11-17, 1999. [3] E. H. Land, An alternative technique for the computation of the designator in the Retinex theory of color vision, Proc. Natl. Acad. Sci.
2 : Retinex U.S.A., vol. 83, 3076, 1986. [4] Z. Rahman, Properties of a center/surround Retinex: Part 1: Signal processing design, NASA Contractor Report 198194, pp. 13, 1995. [5] D. J. Jobson and G. A. Woodell, Properties of a center/surround Retinex: Part 2: Surround design, NASA Technical Memorandum 110188, pp. 15, 1995. [6] Z. Rahman, D. J. Jobson, and G. A. Woodell, Multiscale Retinex for color rendition and dynamic range compression, Proc. SPIE 2847, 183, 1996. [7] D. J. Jobson, Z. Rahman, and G. A. Woodell, A Multiscale Retinex for Bridging the Gap Between Color Images and the Human Observation of Scenes, IEEE Transactions on Image Processing, Vol. 6, No. 7, pp. 965-976, 1997 [8] Numan Unaldi, Praveen Sankaran, Vijayan K. Asari, and Zia-ur-Rahman, Image Enhancement for Improving face detection under Non-Uniform Lighting Conditions, Proceedings of 15th IEEE International Conference on Image Processing, pp.1332-1335, Oct., 2008 [9] Hanumantharaju M. C, Ravishnkar M., Rameshbabu D. R., and Ramachandran S, Color Image Enhancement using Multiscale Retinex with Modified Color Restoration Technique, 2nd Int. Conf. on EAIT, pp 93-97, 2011. [10] Sangwon Lee, Changyoung Song, Seongsoo Cho, Seongihl Kim, and Junegill Kang, Contrast Enhancement Based on Weight Mapping Retinex Algorithm Journal of The Institute of Electronics Engineers of Korea, Vol. 46, No.4, pp. 31-41, 2009 [11] R. C. Gonzalez and R. E. Woods, Digital Image Processing 3rd Ed., Pearson Prentice Hall, 2010. - 2004 3 ~ 2009 2 : - 2009 3 ~ 2012 2 : - 2012 3 ~ : - :,, - 2003 3 ~ 2010 2 : - 2010 3 ~ 2012 2 : - 2012 3 ~ : LG - :,, - 1988 2 : - 1991 2 : - 1999 2 : - 1991 ~ 2000 7 : LG DTV - 2007 ~ : - :,,