Method and apparatus for correction of color shifts due to illuminant changes

• US 5,771,311 A
• Filed: 12/27/1995
• Issued: 06/23/1998
• Est. Priority Date: 05/17/1995
• Status: Expired due to Fees

• Current Assignee: TOYO Ink Manufacturing Company Limited (artience Co. Ltd.)
• Original Assignee: TOYO Ink Manufacturing Company Limited (artience Co. Ltd.)
• Inventors: Arai, Yoshifumi
• Primary Examiner(s): Boudreau, Leo
• Assistant Examiner(s): Tran, Phuoc
• Application Number:
  US08/579,549
• Time in Patent Office: 909 Days
• Field of Search: 382/156, 382/157, 382/158, 382/162, 382/167, 382/191, 382/274, 358/518, 358/523, 364/526
• US Class Current: 382/162
• CPC Class Codes: H04N 1/6086 by scene illuminant, i.e. conditions at the time of picture capture, e.g. flash, optical filter used, evening, cloud, daylight, artificial lighting, white point measurement, colour temperature

Abstract

Color separation values such as CMYK values from an input device are transformed into characteristic parameter values by a first transforming device having a neural network which has undergone learning in advance in such a manner as to output characteristic parameter values obtained by multivariate analysis of spectral distributions which correspond to the color separation values and are illuminant-independent. These characteristic parameter values are subjected to a linear transformation by a second transforming device, the linear transformation having a constraint that calorimetric values under a specified illuminant are to be equal or another similar constraint, by using mean vectors and eigenvectors (principal component vectors) of predetermined spectral reflectances stored in a storage device. The transformed characteristic parameter values are transformed into color separation values of a target color reproduction device by a third transforming device having a neural network which has undergone learning in advance.

Claims

1. A color reproduction method for effecting color matching for a plurality of different color reproduction devices for reproducing the color of an original color image, comprising the steps of:

setting an intermediate color representation system which is represented by at least three characteristic parameter values obtained by multivariate analysis of a spectral reflectance or a spectral transmittance which is not dependent on an illuminant;

learning a first neural network such that, when at least three color information values peculiar to each of said color reproduction devices are inputted, said at least three characteristic parameter values of said intermediate color representation system corresponding to said inputted color information values are outputted, and learning a second neural network such that, when said at least three characteristic parameter values of said intermediate color representation system are inputted, color information values corresponding to said inputted characteristic parameter values are outputted;

transforming said at least three color information values peculiar to a first color reproduction device into said at least three characteristic parameter values of said intermediate color representation system by using said first neural network which has been learned;

transforming said at least three characteristic parameter values under a constraint concerning colorimetric values of colors reproduced by said first color reproduction device and a second color reproduction device different therefrom; and

transforming said at least three characteristic parameter values transformed under the constraint, into color information values of said second color reproduction device by using said second neural network which has been learned, so as to effect color matching for said plurality of different color reproduction devices.

2. A color reproduction method according to claim 1, wherein said at least three color information values are values for representing three predetermined primary colors based on an additive mixture.

3. A color reproduction method according to claim 2, wherein said additive mixture uses R, G, and B values.

4. A color reproduction method according to claim 1, wherein said at least three color information values are values for representing each of predetermined colors based on a subtractive mixture.

5. A color reproduction method according to claim 4, wherein said subtractive mixture uses C, M, and Y values or C, M, Y, and K values.

6. A color reproduction method according to claim 1, wherein principal component analysis is used as said multivariate analysis, and a color system which is represented by principal component values of the spectral reflectance obtained by said principal component analysis is used as said intermediate color representation system.

7. A color reproduction method according to claim 1, wherein each of said neural networks is a multilayered feedforward neural network with three to five layers having a learning function.

8. A color reproduction method according to claim 1, wherein said constraint has a condition that colorimetric values of a subject and those of a reproduced color under a designated illuminant are equal, or that a mean color difference value between a subject color and a reproduced color measured under a plurality of designated illuminants is minimized.

9. A color reproduction method according to claim 1, wherein any one of an illuminant A, an illuminant B, an illuminant C, an illuminant D, and illuminants F1 to F12 which are specified by CIE (Commission Internationale de l'"'"'Eclairage) or an illuminant having a spectral distribution equivalent thereto is used as said illuminant used in the transformation into said color information values for said second color reproduction device.

10. The method of claim 1, wherein the step of transforming said at least three transformed parameter values is performed by using eigenvectors and a mean vector of said spectral reflectance or said spectral transmittance which are obtained by said multivariate analysis.

11. A color imaging apparatus for effecting color matching for a plurality of different color reproduction devices for reproducing the color of an original color image, said color imaging apparatus comprising:

input means for converting at least three color information values into electrical signals;

storage means for storing eigenvectors and mean vectors of spectral reflectances of reproduced colors by a first color reproduction device and a second color reproduction device which are obtained by performing the multivariate analysis of spectral reflectances in advance, connection weights of two neural networks which have been learned in advance, a spectral distribution of at least one illuminant, and color matching functions;

first transforming means adapted to transform the electrical signals from said input means into electrical signals corresponding to characteristic parameter values;

second transforming means for transforming the electrical signals from said first transforming means by a linear transformation having a condition that colorimetric values of a subject color and those of a reproduced color under a designated illuminant are equal, or that a mean color difference value between a subject color and a reproduced color measured under a plurality of designated illuminants is minimized;

third transforming means adapted to transform characteristic parameter values of said second color reproduction device outputted from said second transforming means into electrical signals corresponding to color information values of said second color reproduction device; and

output means for outputting as output signals the electrical signals outputted from said third transforming means.

12. The apparatus of claim 11, wherein said first transforming means comprises a multilayered feedforward neural network with a learning function which has learned the transformation of said at least three color information values into principal components of a spectral reflectance of a reproduced color corresponding to a set of color separation values of said at least three color information values.

13. The apparatus of claim 11, wherein said characteristic parameter values are obtained by the multivariate analysis of spectral reflectances or spectral transmittances of a reproduced color corresponding to said at least three color information values.

14. The apparatus of claim 11, wherein said second transforming means is a means for transforming by using said eigenvectors and mean vectors of said spectral reflectances of said reproduced colors by said first color reproduction device and said second color reproduction device which are stored in advance in said storage means.

15. The apparatus of claim 11, wherein said third transforming means comprises a multilayered feedforward neural network with a learning function which has learned the transformation of principal components of a spectral reflectance of a reproduced color into color information values.