(12) Ulllted States Patent (10) Patent N0.: US 7,605,943 B2 Berns et a]. (45) Date of Patent: *Oct. 20, 2009

Transcription

1 US B2 (12) Ulllted States Patent (10) Patent N0.: US 7,605,943 B2 Berns et a]. (45) Date of Patent: *ct. 20, 2009 (54) PRDUCTIN F A CLR CNVERSIN (58) Field of Classi?cation Search /1.9, PRFILE UTILIZING CMPARISN F 358/3.23, 518; 382/ 167 SPECTRAL REFLECTANCE DATA AND See application?le for complete search history. CMPARATIVE CLRS (56) References Cited (75) Inventors: Roy S. Berns, Pittsford, NY (US); US. PATENT DCUMENTS Lawrence A. Taplin, Rochester, NY. (Us); YoshifumiAraL Nagano_ken (JP); 5,619,348 A * 4/1997 Tokai /518 Michinao sawa, Nagano-ken (JP) (Continued) (73) Assignee: Seiko Epson Corporation, Tokyo (JP) FREIGN PATENT DCUMENTS ( * ) Notice: Subject to any disclaimer, the term of this patent is extended or adjusted under 35 U.S.C. 154(b) by 592 days. This patent 15 Subject I a terminal (115- claimer. EP /2001 (Continued) THER PUBLICATINS Abstract of Japanese Patent Publication No , Pub. Date: Mar. 12, 1986, Patent Abstracts of Japan. (21) Appl. N0.: 10/570,420 (Continued) _ Primary ExamineriMark K Zimmerman (22) PCT Flled: NV Assistant ExamineriJavier J Ramos (86) _ PCT No.. PCT/US2004/ (74) Attorney, Agent, Gencarella LLP or FirmiMartine Penilla & 371 (0X1), (57) ABSTRACT (2), (4) Date? Mar A method and apparatus for producing a color conversion pro?le for use in a color inkjet printer, a color laser printer, or (87) PCT Pub- NJ W02005/ the like. In one embodiment, a spectral printing model con verter is used to sample ink amount data to spectral re?ec PCT Pub- Date? May 12a 2005 tance Rsmp(7»). Using this spectral re?ectance Rsmp(7»), _ colorimetric values CV1, CV2 under two different viewing (65) P1101 Publlcatlon Data conditions are calculated, and an evaluation index El that Us 2006/ A1 D _ mcludes an mdex of color dlfference between the colonmet ec ric values CV1, CV2 is calculated. 1 another embodiment, an Related US Application Data evaluation index El that includes an index of color difference between sample color represented by sample ink amount data (60) Provisional application No. 60/517,236,?led on Nov. and reference color obtained from a reference color patch is 3, calculated. Sample ink amount data having the best evalua tion index E1 is then selected as a representative sample. (51) Int. Cl. Pro?les are created from the plurality of selected representa G06F 15/00 ( ) tive samples. (52) US. Cl /1.9; 358/3.23; 358/518; 382/167 6 Claims, 16 Drawing Sheets F100 F120 g, c _ 122 Evaluation Index Generator 5 M p?mivlgiilidel 6 a v ' er Rsmpu) cal?lll? llgl dex N E K o :: (Egg/mg?. r lllalgulator (gill, CV2 Ehmc") g G -> o srgb Pm?le [- 152 sum srgb rm f- 140 Pro?le Generator High Evalua?on Sample Data "mum r120 mu Pro?le r142 cumvmomso srgb CMYKG r150 L a b CMYKG I C M1-YfZl?-1IG1 Gamut Mapplng Processor (D50)

2 US 7,605,943 B2 Page 2 US. PATENT DCUMENTS 5,907,495 A * 5/1999 Snyder et al /6 5,929,906 A 7/1999 Arai et al. 6,542,634 B1* 4/2003 hga /167 7,053,910 B2 5/2006 Newman 7,298,513 B2 11/2007 Namikata 2003/ A1* 5/2003 Berns et al / / A1* 11/2003 Takahashi et al.. 356/ / A1* 12/2003 Van Bael / / A1* 5/2005 Berns et al.. 358/ / A1* 5/2005 Berns et al /1.9 FREIGN PATENT DCUMENTS JP /1986 JP /1999 JP /2001 JP /2002 JP /2003 JP /2003 JP /2003 W0 W0 02/ /2002 THER PUBLICATINS Y. Arai et al., Device-and-illuminant independent color reproduc tion using principal component analysis and neural networks, Proc. SPIE, 1995, SPIE vol. 2414, pp Y. Arai et al., A neural network method for correcting the color shifts due to illuminant changes, Proc. ICNIP, 1996, vol. 1, pp Y. Arai et al., Color Correction Method Based on the Spectral Re?ectance Estimation using aneural Network, Proc. Fourth Color Imaging Conf. Color Science, Systems and Applications, 1996, pp Y Arai et al., A Color Conversion Method Using a Neural Network for a Cross-Media Color Appearance Matching, Proc. ofaic Color 97, 1997, vol. II, pp Y Chen et al., A Multi-Ink Color-Separation Algorithm Maximizing Color Constancy, IS&T/SID 11th Color Imaging Conf, 2003, pp M.R. Rosen et al., Spectral Redundancy in a 6-Ink Ink-Jet Printer, Proc. IS&T s 2003 PICS Conf, 2003, pp Abstract of Japanese Patent Publication No , Pub. Date: 24 Sep. 1999, Patent Abstracts of Japan. Abstract of Japanese Patent Publication No , Pub. Date: 16 Nov. 2001, Patent Abstracts of Japan. Abstract of Japanese Patent Publication No , Pub. Date: 06 Dec. 2002, Patent Abstracts of Japan. Abstract of Japanese Patent Publication No , Pub. Date: 23 May 2003, Patent Abstracts of Japan. Abstract of Japanese Patent Publication No , Pub. Date: 23 May 2003, Patent Abstracts of Japan. Abstract of Japanese Patent Publication No. JP , Pub. Date: ct. 24, 2003, Patent Abstracts of Japan. * cited by examiner

3

4 US. Patent 0a. 20, 2009 Sheet 2 0f 16 US 7,605,943 B2 Fig.2 L First Embodiment Start 7 v r 5 Prepare spectral printing model (fonivard model converter) r52 Prepare a large number of virtual samples in ink space Calculate CIELAB values of virtual samples using the spectral printing model 0 r 54 Divide the CIELAB space into a plurality of cells and sort all the samples into the different cells 0 r 55 De?ne evaluation index : Color lnconstancy Index + a l r56 Calculate evaluation index for each sample and select the best sample in each cell r57 Create a preliminary ink pro?le for?ner cells using non-niform interpolation based on selected samples for coarse cells 0 r 58 Create a?nal ink pro?le by uniform interpolation based on previous lookup table Gamut map to produce printer lookup table KS9 End

5 _ US. Patent 0a. 20, 2009 Sheet 3 0f 16 US 7,605,943 B2 I I I F I l I I l a - F I.D o I v 00 U ) l l I I I l L I I 8 -_ o 0 0* u- 0 (D (0 l\ (D ID 1' ('0 N \- ' r 3- o I l l I l I I l I 2 o _ 'F on A C v d) l l l l '- I I 8 LL Q o 01 co I- 10 o I0 <r n N \- 0 I a. l I I F l I I I I 8 * o u A it, ('0 0') I I l I I l 8 '_ 0 01' LL N o oo (0 <r N I}! <lr It ) 09' \-

6 US. Patent 0a. 20, 2009 Sheet 4 0f 16 US 7,605,943 B2 Fig.4 Step S6 (Selection of best sample) f $10 Select sample ink amount data in the cell /_ S11 Convert sample ink amount to spectral re?ectance Foniuard model A2 \ /_ S15 Calculate XYZ for D50 Calculate XYZ for F11 r- 813 l f- 516 Chromatic adaptation Chromatic adaptation transform to D65 transform to D65 /- S14 4, f 517 Calculate CIELAB : Calculate CIELAB : CV1=(L*a*b*)o50->oss CV2=(L*a*b*)F11 >D6s Calculate Evaluiation Index N0 S19 Done? Yes /- $20 Select the best sample in the cell Return I

7 US. Patent ct. 20, 2009 Sheet 5 0f 16 US 7,605,943 B2 Fig.5(A) Non-uniform interpolation 0.5»e Fig.5(B) Before non-uniform interpolation Fig.5(C) After non-uniform interpolation 0 o.n _ _- IR" o 0 0 o 0o 0 0 i0 AU 0 o o o 0 o o oo o m.b & ow _ l5-0 l0.... U. I I. :mm I. I... D. 5.

8 US. Patent 0a. 20, 2009 Sheet 6 0f 16 US 7,605,943 B2 mamas; 5:60 lllv SEA 5:33: $ ".1 *w 5:88am mm

9 US. Patent 0a. 20, 2009 Sheet 7 0f 16 US 7,605,943 B2 Fig CMYKG ---- CMYKcm a wuzocmtmu samples Comparison of A E00 color differences between original target and reproduction by CM-YKcm and CMYKG printers

10 US. Patent 0a. 20, 2009 Sheet 8 0f 16 US 7,605,943 B2 E wvcczo In EEmz 1LT: _ I--_---- vmwwmwmwmvmwmwov mvownmvm w? N moeemm B EEEQES 52:: 5:32.65 B ouzzzo Ea E :88am: E5.90 w 60 N (0 L0 <1 0 ) N 1- xapu Aoumsuoom.IIQ

11

12 US. Patent 0a. 20, 2009 Sheet 10 0f 16 US 7,605,943 B2 Fig. 1 0 Second Embodiment Start 1 f- S21 Prepare spectra! printing model (forward model converter), f 522 Prepare a plurality of reference colors (comparative colors) \ I S23 Measure spectral re?ectance of reference colors and calculate their ClELAB values Prepare a plurality of virtual samples for each of the reference colors I S24 [- S25 Calculate ClELAB values of virtual samples using the spectral printing model De?ne evaluation index : Metameric Index + Cl Calculate evaluation index for each sample and select the best sample for each reference color J I 528 Create a lookup table based on selected samples for each reference color

13

14 US. Patent 0a. 20, 2009 Sheet 12 0f 16 US 7,605,943 B2 Fig.12 Third Embodiment Start J- 531 Prepare spectral printing model (fonrvard model converter) Divide CIELAB space into a plurality of cells De?ne evaluation index: Color lnconstancy index + a y [-534 Set initial sample ink amount for a cell of interest 7 /-S35 Calculate evaluation index for the i 0 sample r- 837 Modify sample ink amount Meet the criteria? $36 No Yes S38 Create a lookup table based on approved samples r839

15

16 US. Patent 0a. 20, 2009 Sheet 14 0f 16 US 7,605,943 B2 Fig.14(A) Cell division in Cellular Yule-Nielsen Spectral Neugebauer Model dmax ' dc=dm=0 R R 30 F g.14(b) lnk area coverage fc(d) fc(d) 1.0 / l/! 0 dmax. Ink amount lros R13 R23 R33 fm, dm Fig. 1 4(0) Calculation of Rsmp( A)

17 US. Patent ct. 20, 2009 Sheet 15 of 16 US 7,605,943 B2 Fig.15 Selected digital counts and area coverages of every ink for Cellular Yule-Nielsen Spectral Neugebauer Model Digital counts of ink amount Cyan Magenta Yellow Black Green range

18

19 1 PRDUCTIN F A CLR CNVERSIN PRFILE UTILIZING CMPARISN F SPECTRAL REFLECTANCE DATA AND CMPARATIVE CLRS CRSS-REFERENCE T RELATED APPLICATINS Priority is claimed from Provisional Patent Application No. 60/517,236,?led Nov. 3, 2003, the disclosure Which is hereby incorporated herein by reference in its entirety. FIELD F THE INVENTIN The present invention relates to a method and apparatus for producing a color conversion pro?le for use in a color ink jet printer, a color laser printer, or the like, as Well as a printer using such a color conversion pro?le. BACKGRUND F THE INVENTIN In recent years, the use of color ink-jet printer and color laser printers has become Widespread. A color printer uses a color conversion lookup table to convert input color image data into corresponding amounts of inks of plural colors. In the conventional method for creating a color conversion lookup table, (i) a plurality of sample color patches are printed, (ii) colors of the color patches are measured to obtain their colorimetric values, and (ii) a lookup table that repre sents correspondence relationships between color patch colo rimetric values and the amounts of ink used to print the color patches is created. HoWever, colorimetric values are dependent upon viewing conditions during color measurement. Accordingly, Where a color conversion lookup table has been created so as to give a printout having good color reproduction under a certain view ing condition, it Will not necessarily be the case that printouts created using that color conversion lookup table Will afford good color reproduction under other viewing conditions. Accordingly, there has for some time existed a need for a color conversion lookup table Which enables good color reproduction under various viewing conditions. There has also been a need, When creating a reproduction of a painting or drawing for example, to create a color conversion lookup table that can reproduce as faithfully as possible the color appearance of the original. Such requirements are not limited just to color conversion lookup tables, but are generally desir able in all manner of pro?les used for color conversion. SUMMARY F THE INVENTIN Accordingly, the present invention provides a method and apparatus for producing a color conversion pro?le capable of achieving good color reproduction under various viewing conditions. The invention also provides a method and appa ratus for producing a color conversion pro?le for reproducing as faithfully as possible the color appearance of an original. More speci?cally, the invention provides a method of pro ducing a pro?le de?ning correspondence between colorimet ric value data and ink amount data representing a set of ink amounts of plural inks usable by a printer. The method com prises: (a) providing a spectral printing model converter con?gured to convert ink amount data to spectral re?ectance of a colorpatch to be printed according to the ink amount data; (b) providing a plurality of sample ink amount data each repre senting a set of ink amounts of plural inks; (c) converting each sample ink amount data into spectral re?ectance of a sample US 7,605,943 B patch to be printed With the ink amounts represented by the sample ink amount data using the spectral printing model converter; (d) calculating an evaluation index for each sample ink amount data, the evaluation index including a color dif ference index representing a color difference between a sample color Which is calculated from the spectral re?ectance and a comparative color Which is selected as a basis for comparison; (e) selecting plural sample ink amount data based on the evaluation index; and (f) producing a pro?le de?ning correspondence between colorimetric value data and ink amount data based on the selected plural sample ink amount data. The plural inks preferably include one or more spot color inks. According to this method, by calculating appropriate evaluation indices it is possible to produce a color pro?le that achieves good color reproduction under various viewing con ditions, as Well as to produce a color pro?le capable of repro ducing as faithfully as possible the color appearance of an original. The evaluation index may be one including a color incon stancy index, or one including a metamerism index. When an evaluation index that includes a color inconstancy index is used, it is possible to produce a color pro?le that achieves good color reproduction under various viewing conditions. n the other hand, When a metamerism index is used, it is possible to produce a color pro?le capable of reproducing as faithfully as possible the color appearance of an original. The metamerism index represents a color difference evaluation index CDI representing color difference between two colors. The color inconstancy index also represents a color difference evaluation index CDI representing the color difference of a single sample viewed under two viewing conditions. The present invention may take the form of any of a number of different embodiments, for example, a pro?le production method and pro?le production device, a printing method and printing apparatus using a pro?le, a computer program for realizing the functions of such a method or device, a storage medium having such a computer program recorded thereon, and a data signal including such a computer program and embodied in a carrier Wave. BRIEF DESCRIPTIN F THE DRAWINGS FIG. 1 is a block diagram showing the overall system arrangement in Embodiment 1 of the invention. FIG. 2 is a?owchart illustrating the process?ow in Embodiment 1. FIGS. 3(A)-3(C) are graphs showing sample colors in the CIELAB color space in a Working example. FIG. 4 is a?owchart showing in detail the procedure of Step S6 in FIG. 2. FIGS. 5(A)-5(C) illustrate non-uniform interpolation in Step S7. FIGS. 6(A) and 6(B) illustrate gamut mapping in Step S8. FIG. 7 is a graph showing color difference calculations in a Working example and a comparative example. FIG. 8 is a graph showing color inconstancy index calcu lations in a Working example and a comparative example. FIG. 9 is a block diagram showing system arrangement in Embodiment 2 of the invention. FIG. 10 is a?owchart illustrating the process?ow in Embodiment 2. FIG. 11 is a block diagram showing the overall system arrangement in Embodiment 3 of the invention. FIG. 12 is a?owchart illustrating process?ow in Embodi ment 4.

20 US 7,605,943 B2 3 FIGS. 13(A) and 13(B) illustrate the Spectral Neugebauer model. FIGS. 14(A)-14(C) illustrate the Cellular Yule-Nielsen Spectral Neugebauer model. FIG. 15 shows cell division grid coordinates in the Cellular 5 Yule-Nielsen Spectral Neugebauer model. FIG. 16 shows a cellular model. DETAILED DESCRIPTIN The embodiments of the invention Will be described in the following order. A. Embodiment 1 B. Embodiment 2 C. Embodiment 3 15 D. Example of Spectral Printing Model E. Variant Examples A. EMBDIMENT 1 FIG. 1 is a block diagram of a system in accordance With preferred embodiments of the present invention. The system comprises a spectral printing model converter 100, an evalu ation index generator 120, a selector 130, a pro?le generator 140, and a gamut mapping processor 160, although the sys tem could be comprised of other numbers and types of com ponents in other con?gurations. Spectral printing model con ver ter 100 converts ink amount data into spectral re?ectance Rsmp(7t) of a color patch to be printed according to the ink amount data. The term color patch herein is not limited to chromatic colors, but is used in a broad sense to include achromatic colors as Well. In this embodiment, assuming a color printer that can use six colors of ink, namely, cyan (C), magenta (M), yellow (Y), black (K), orange (), and green (G), spectral printing model converter 100 has as inputs ej ec tion amounts of these six inks. The spectral printing model Will be described in greater detail in Section D. Hereinafter, spectral printing model Will also be referred to as forward model. Evaluation index generator 120 includes a sample color calculator 122, a comparative color calculator 124, and an evaluation index calculator 126 although evaluation index generator 120 can incorporate other types and numbers of elements in other combinations. Sample color calculator 122 uses the spectral re?ectance Rsmp(7t) for a sample ink amount data to calculate a colorimetric value CV1 under a?rst view ing condition. In this embodiment, illuminant D50 is used as this?rst viewing condition. A color represented by colorimet ric value CV1 obtained under this?rst viewing condition is termed a sample color. Comparative color calculator 124 uses the spectral re?ectance Rsmp(7t) for a sample ink amount data to calculate a colorimetric value CV2 under a second viewing condition. In this embodiment, illuminant P11 is used as this second viewing condition. Hereinafter, a color represented by colorimetric value CV2 obtained under the second viewing condition is sometimes referred to as a com parative color. As Will be understood from the preceding description, in Embodiment 1 sample color calculator 122 and comparative color calculator 124 respectively calculate colorimetric val ues CV1, CV2 using the same spectral re?ectance Rsmp(7t), but under different viewing conditions. Using these colori metric values CV1, CV2, evaluation index calculator 126 calculates an evaluation index EIl for determining the quality of the sample ink amount data. Speci?c examples of the evaluation index EIl Will be described hereinbelow. Selector 130 selects sample ink amount data having a good evaluation index EIl. Pro?le generator 140 uses the selected sample ink amount data, together With colorimetric values (L*a*b* values) of colorpatches printed using the sample ink amount data, to create an ink pro?le 142. The ink pro?le 142 is in the form of a lookup table indicating correspondence relationships between colorimetric values (L*a*b* values) and CMYKG ink amounts. Ink pro?le may be referred to as output device pro?le as Well. Pro?le herein refers to a speci?c embodiment of a set of conversion rules for convert ing color space, and is used in a broad sense to include device pro?les and lookup tables of various kinds. Gamut mapping processor 160 uses the ink pro?le 142 and an input device pro?le 162, Which has been prepared in advance, to create a printer lookup table 180. As the input device pro?le 162 there may be used, for example, of a pro?le for converting the srgb color space to the L*a*b* color space. Printer lookup table 180 converts input color image data (eg srgb data) to ink amount data. FIG. 2 is a?owchart illustrating the process?ow in Embodiment 1. In Step S1, a spectral printing model is deter mined and a converter 100 is created. In one Working example, the Cellular Yule-Nielsen Spectral Neugebauer model is used as the spectral printing model. A detailed description is provided in Section D. In Step S2, a large number of virtual samples are prepared. Here, virtual sample refers to provisional ink amount data used in the pro?le creation process, and to a virtual color patch to be printed according to this ink amount data. Here inbelow, virtual samples are also referred to simply as samples. In the Working example, virtual samples (sample ink amount data) that combines amounts of all six inks Were prepared, setting ink amounts at eleven points at 10% inter vals Within the range 0-100%. As a result, 116 (:1,771,516 1,771,561) virtual samples Were prepared. 100% ink amount refers to the amount of ink providing solid coverage With a single ink. In Step S3, the sample ink amount data of the virtual samples is converted to spectral re?ectance Rsmp(7t) using spectral printing model converter 100, and colorimetric val ues L*a*b* in the CIELAB color system are calculated from this spectral re?ectance Rsmp(7t). In the Working example, colorimetric values Were calculated using the CIE illuminant D50 and CIE Standard bserver viewing condition. HereinbeloW, the color observed When a virtual sample is viewed under a certain viewing condition is termed sample color. FIGS. 3(A)-3(C) are graphs showing sample colors in the CIELAB color space in the Working example. In FIG. 3(A) the horizontal axis represents the a* axis of the CIELAB color space, and the vertical axis represents the b* axis. In FIGS. 3(B) and 3(C), the horizontal axis represents the a* axis and b*, and the vertical axis represents the L* axis. As Will be understood from the above description, the 1 16 sample colors are more densely concentrated Where lightness L* is low and distributed sparsely up to the area of high lightness L*. A more uniform distribution of sample colors may be achieved inter alia by setting sample ink amount data to?ner intervals in the range of relatively small ink amount and to coarser intervals in the range of relatively large ink amount. In Step S4, the color space of the colorimetric values (here, the CIELAB color space) is divided into plural cells, and the plural sample colors are sorted in relation to the cells. In the Working example, the CIELAB color space Was divided equally into 16><16><16 cells. In Step S5, an evaluation index EIl for use in selecting a good sample is de?ned. The evaluation index EIl used in Embodiment 1 is represented by the following Equation (1).

21

22

23

24

25

26

27