Color gamut of SOCS and its comparison to Pointer's gamut

Transcription

1 Color gamut of SOCS ad its compariso to Poiter's gamut Masao Iui*, Tomotaka Hirokaw, oshihiko Azum, ad Johji Tajim* * Tokyo Polytechic Uiversity 58 Iiyama, Atsugi, Kaagawa , Japa ** Nagoya City Uiversity amaohata, Mizuho-cho, Mizuho-ku, Nagoya , Japa Abstract The Stadard object color spectra (SOCS) database was developed for proper evaluatio of color-sesor quality of image iput devices. The SOCS has bee published i a ISO techical report, ISO TR 666. A CD-ROM is attached to the ISO-TR. The CD-ROM cotais more tha fifty thousads spectra of objects, such as photographic materials, offset prits, computer color prits, pait (ot for art), paits (oil paits, water colors), textiles, flowers ad leaves, outdoor scees, ad huma ski. Color gamut of SOCS for stadard illumiat D65 was calculated ad discussed. Color gamut efficiecy of SOCS, which is ratios of the area ad the volume of color gamut of SOCS to these withi optimal color loci, is almost the same as the performace of photographic color reversal film. Also, color gamut of SOCS for illumiat C was calculated ad compared to Poiter's gamut. Itroductio The Stadard object color spectra (SOCS) database was developed for proper evaluatio of color-sesor quality of image iput devices [][2]. The SOCS has bee published i a Japaese Idustrial Stadard Techical Report (JIS-TR) i 998[]. Spectral reflectace or trasmittace data of 49,672 objects were stored i the CD attached to JIS-TR, which were classified accordig to object categories. After the publicatio, data for other categories were added. Though they are very valuable data eve i the origial form, the umber of data is very large, ad their distributio is biased toward some categories. Though may data are eeded to statistically evaluate the quality of color image iput devices, careful data arragemet is ecessary for meaigful use, ad a smaller data set for easily evaluatig the quality of color image iput devices was desired. Oly several hudreds of represetative spectral data for use istead of the full data set were selected as typical sets ad differece sets[4] ad was the pricipal part of ISO techical report, ISO TR 666[5]. A CD-ROM is attached to ISO-TR. The CD-ROM cotais more tha fifty thousads spectra of objects, such as photographic materials (trasparecies ad reflectio prits), offset prits, computer color prits (dye sublimatio priter, ik-jet priter, electrostatic priter), pait (ot for art), paits (oil paits, water colors), textiles (sythetic dyes ad plat dyes), flowers ad leaves, outdoor scees, ad huma ski (bare orth Asia ski, foudatio-applied orth Asia ski, bare south Asia ski, foudatio-applied south Asia ski, bare Caucasia ski, ad bare Negroid ski). The umber of collected data is summarized i Table [4]. Each data is -dimesioal spectral reflectace or trasmittace. It is very iterestig ad useful to kow gamut of SOCS i a color space. We have obtaied the gamut of SOCS ad discussed, ad compared it with the Poiter's gamut [6]. Also, a useful measure to evaluate the performace of colorat set for color reproductio system, color gamut efficiecy [7], is applied to the gamut of SOCS. Table Number of spectral data collected for SOCS Category Number of Number sub-categories of colors Photographic materials (Trasparecies/ 8 2,4 Reflectio prits) Graphic prits (Offset/Gravure),624 Computer color prits 2 7,856 Pait (ot for art) 6 Paits (for art) Textiles 6 2,82 Flowers 48 Leaves 92 Huma faces 8,57 Kriov data 7 Total 5,6

2 Calculatio of color gamut As metioed above, 5,6 spectra are stored i SOCS CD. A umber of colors have bee deemed iappropriate for the highest chroma of the segmets ad removed i this study, as same as Aex C i Ref. 5, as show i the followig. all colors i files from kriov to kriov8 (Kriov) #9 i pa_o file (Oil paits) #58, #59, #6 i pa_a file (Water colors) These are iterpolated data, metallic colors, ad colors geerated with fluorescet colorats. These 74 spectra are ot used, therefore 52,987 spectra are used for calculatig color gamut i this study. At first tristimulus values,, ad Z are calculated from SOCS spectra. k Z k k k R( x( R( y( R( z( 78 8 y( where, is spectral power of illumiat D65 or C, R( is spectral reflectace or trasmittace of SOCS spectrum data, ad x(, y( ad z( are color matchig fuctios. Next, colorimetric values CIE 976 L*b* are calculated from the tristimulus values. The chroma C* ad hue agle h are calculated from ad b*. L a b * * * Z,, Z Thus colorimetric values L*b* ad C* for all SOCS spectra are calculated for illumiats D65 ad C. C* 2 2 ( a * + b * ) ( ) / 2 h arcta b * a * The lightess L* was divided at itervals of 5 betwee 5 ad to 95, so the umber of lightess steps were 9. Ad Z Z hue agle h was divided at itervals degrees betwee degree ad 5 degrees, so the umber of hue agle step was 6. Therefore the color space is divided ito 684 segmets. The rage of lightess is 5 ad the rage of hue agle is degree for each segmet. For example, the segmet for chroma 2 ad hue agle degrees meas the segmet of lightess ragig from 7.5 to 22.5 ad of agle ragig from 25 degrees to 5 degrees. The poits correspodig to SOCS data exist i three dimesioal L*b* color space. The space is divided ito the segmet of lightess ad hue agle. The maximum values of chroma C* i the segmet were searched. Color gamut is a area withi the loci coectig the highest chroma poits. Color gamut of SOCS Chroma C* ad hue agle h of SOCS spectra were calculated for CIE stadard illumiat D65. The highest chromas i the 684 segmets were searched. Noe of poits was cotaied i 4 segmets. There were o poits i 4 segmets. Number of highest chromas for each category is show i Table 2. The category of the largest umber of highest chroma is "Computer color prits." More tha a half of highest chromas are cotaied i "Computer color prits." Highest ad lowest lightess are 98. ad.62 respectively. Categories, file ames ad idetificatio umbers of these lightesses are paits/pa_g.it/ ad pait/pait.it/4, respectively. The highest chroma poits at each L* level ca form a polygo whose vertices correspod to the poits. The area iside the polygo represets a cross sectio of SOCS color gamut for each L* level. The color gamut of SOCS is show i Fig.. The theoretical limit of object color, optimal color, calculated by a fast algorithm computig optimal color [8] is also show i Fig.. The fast algorithm uses immediate replicatio of the visual spectrum i the ifrared rage to create a exteded spectrum. Table 2 Number of highest chroma for each category Category Number of highest chroma Computer color prits 79 Graphic prits Textiles 68 Photographic materials 55 Paits (for art) Pait (ot for art) 9 Total 64

3 2 b* L*5 2 b* L* 2 b* L*5 2 b* L* b* L*25 2 b* L* 2 b* L*5 2 b* L* b* L*45 2 b* L*5 2 b* L*55 2 b* L* b* L*65 2 b* L*7 2 b* L*75 2 b* L* b* L*85 2 b* L*9 2 b* L* Fig. Color gamut of SOCS (solid lie) ad optimal color locus (thi lie) for stadard illumiat D65

4 I Fig., the color gamut of SOCS for L*2 shows higher chroma tha the optimal color at hue agle 4 degrees. It appears as if the color gamut of SOCS exceeds the theoretical limit. The color gamut of SOCS shows the highest chroma at the segmet, i which the rage of lightess is 5 ad the rage of hue agle is degrees. O the other had the optimal color loci are calculated precisely for the give lightess, ot a maximum value of chroma of optimal colors i the segmet. The relatioship of these poits i a cross sectio of hue agle 4 degrees are show i Fig. 2. It is uderstood that the highest chroma of SOCS i the segmet for L*2 ad hue agle h4 degrees is ot higher tha optimal color from the schematic explaatio To calculate a measure to evaluate the performace of colorat set for color reproductio system, color gamut efficiecy [7], area of color gamut of SOCS ad area withi optimal color are calculated ad show i Fig.. The area color gamut efficiecy, CGE A, is defied by dividig the area of color gamut of SOCS, A SOCS, by the area withi the correspodig optimal color locus, A optimal, ad the expressig the result as a percetage, as i the followig equatio [7]. CGE Similarly, volume color gamut efficiecy, CGE V, is defied by the followig equatio. CGE A A SOCS A optimal V V SOCS V optimal [%] [%] where V SOCS is the volume of color gamut of SOCS, ad V optimal is the volume withi the optimal color loci. Color gamut efficiecy of SOCS is show i Fig. 4. Maximum area color gamut efficiecy is about 45%. I the Ref. 8, the maximum area color gamut efficiecies of the hypothetical dye sets lie i the rage of 6% to 7%, while the maximum area color gamut efficiecy of the real dye set for a photographic color reversal film barely exceeds 4%. The volume color gamut efficiecy of SOCS is 5%. This percetage is almost the same as the performace of the real dye set for a photographic color reversal film. Compariso to Poiter's gamut Also the chroma C* ad hue agle h for illumiat C were calculated for compariso to the Poiter's gamut o real surface colors previously reported [6]. Color gamut of SOCS, Poiter's gamut ad optimal colors are show i Fig. 5. There are o figures for lightess L*5 ad 5, because there were o such figures i Poiter s literature. It was foud that gamut of SOCS is slightly wider tha those of Poiter's i the low lightess regio ad both are almost same area at the middle lightess. I the high lightess regio, gamut of SOCS is slightly wider tha those of Poiter's i the blue regio, but slightly arrower i red ad gree regios. Fig. 2 Cross sectio of hue agle 4 degrees at L*2 Fig. Areas of color gamut of SOCS ad withi optimal color locus Fig. 4 Color gamut efficiecy of SOCS, the ratio of the area or volume of color gamut of SOCS to the area or volume withi optimal color loci

5 2 b* L*5 2 b* L*2 2 b* L*25 2 b* L* b* L*5 2 b* L*4 2 b* L*45 2 b* L* b* L*55 2 b* L*6 2 b* L*65 2 b* L* b* L*75 2 b* L*8 2 b* L*85 2 b* L* Fig. 5 Color gamut of SOCS (solid lie), optimal color locus (thi lie), ad Poiter s gamut (broke lie) for illumiat C

6 Summary Color gamut of SOCS were calculated ad compared to Poiter's gamut. Cosequetly there is o great differece betwee the gamut of SOCS ad Poiter s gamut. Also, color gamut efficiecies were calculated ad compared to these of photographic color reversal film. Maximum area color gamut efficiecy of SOCS is about 45% ad is slightly greater tha oe of the film, ad both of volume color gamut efficiecy is almost the same. The authors would be pleased to hear from ayoe who is able to offer spectral data of objects color ad measure calibratio color patches with the colorimeter, which is used to measure the spectral data. Refereces [] Johji Tajima, et al., Developmet ad Stadardizatio of a Spectral Characteristics Data Base for Evaluatig Color Reproductio i Image Iput Devices, Proc.SPIE, Vol.49, pp.42-5 (998) [2] Johji Tajima, Masao Iui ad oshihiko Azuma, Developmet of stadard object colour spectra database for colour reproductio evaluatio (SOCS), J.Soc.Photogr.Sci. Techol.Japa, 62(2), 26-5 (999) [] JIS TR 2:998, Stadard object colour spectra database for colour reproductio evaluatio (SOCS) [4] Johji Tajima, Hideaki Haeishi, Nobutoshi Ojima, ad Masato Tsukada, Represetative Data Selectio for Stadard Object Colour Spectra Database (SOCS), th Color Imagig Coferece, pp.55-6 (22) [5] ISO TR 666, Graphic techology - Stadard object colour spectra database for colour reproductio evaluatio [6] M.R.Poiter, The gamut of real surface colours, Color Res. Appl., 5(), (98) [7] Masao Iui, The quatificatio of color gamut efficiecy, J.Soc.Photogr.Sci.Techol.Japa, 6(6), (2) [8] Masao Iui, A fast algorithm computig optimal colors, J.Soc.Photogr.Sci.Techol.Japa, 57(6), (994) Biography Masao Iui received his M.Eg. from Chiba Uiversity i 97 ad joied the Uiversity staff there i the same year. I 986, he joied Koica Corporatio, where he advaced to the positio of Chief Research Associate. I 99, he received his Ph.D. from Chiba Uiversity, ad i 998, Dr. Iui took a professorship at Tokyo Polytechic Uiversity. ad his special iterests iclude image aalysis, evaluatio, desig, ad processig. He is a member of the IS&T, The Royal Photographic Society, ad The Society of Photographic Sciece ad Techology of Japa. Tomotaka Hirokawa is a studut of Tokyo Polytechic Uiversity. His research iterests iclude aalysis ad modelig of color reproductio of hardcopy prit. oshihiko Azuma received a BS degree i physical egieerig i 979 ad a MS degree i iformatio processig egieerig i 98, both from Tokyo Istitute of Techology. After graduatio, he joied DaiNippo Pritig Compay where he egaged i developmet of digital prepress systems. He has bee a assistat professor at Tokyo Polytechic Uiversity sice 995. His research iterests iclude color reproductio model i halftoe image, cross-media color matchig ad color imagig based o characteristics of huma color perceptio. Johji Tajima graduated from the Faculty of Sciece, the Uiversity of Tokyo, i 97 ad received a doctorate i 99. From 97 to 2, he was a research member of NEC Corporatio. He is a professor of Graduate School of Natural Scieces, Nagoya City Uiversity, from 2. He has bee egaged i research o image processig ad patter aalysis, especially color image processig ad D visio. Prof. Tajima is a member of the IEICE, IIEEJ, the Iformatio Processig Society of Japa (IPSJ), ad IEEE Computer Society. He is a fellow of IAPR (Itl. Assoc. for Patter Recogitio).

7 Color gamut of SOCS ad its compariso to Poiter's gamut Masao Iui*, Tomotaka Hirokaw, oshihiko Azum, ad Johji Tajim* * Tokyo Polytechic Uiversity 58 Iiyama, Atsugi, Kaagawa , Japa ** Nagoya City Uiversity amaohata, Mizuho-cho, Mizuho-ku, Nagoya , Japa Abstract The Stadard object color spectra (SOCS) database was developed for proper evaluatio of color-sesor quality of image iput devices. The SOCS has bee published i a ISO techical report, ISO TR 666. A CD-ROM is attached to the ISO- TR. The CD-ROM cotais more tha fifty thousads spectra of objects, such as photographic materials, offset prits, computer color prits, pait (ot for art), paits (oil paits, water colors), textiles, flowers ad leaves, outdoor scees, ad huma ski. Color gamut of SOCS for stadard illumiat D65 was calculated ad discussed. Color gamut efficiecy of SOCS, which is ratios of the area ad the volume of color gamut of SOCS to these withi optimal color loci, is almost the same as the performace of photographic color reversal film. Also, color gamut of SOCS for illumiat C was calculated ad compared to Poiter's gamut. Keywords SOCS, color gamut, optimal color, color gamut efficiecy, Poiter s gamut