Method and apparatus for generating color image mask

• US 5,659,490 A
• Filed: 06/06/1995
• Issued: 08/19/1997
• Est. Priority Date: 06/23/1994
• Status: Expired due to Fees

• Current Assignee: Dainippon Screen Manufacturing Company Limited (SCREEN Holdings Co., Ltd.)
• Original Assignee: Dainippon Screen Manufacturing Company Limited (SCREEN Holdings Co., Ltd.)
• Inventors: Imamura, Atsushi
• Primary Examiner(s): Cosimano, Edward R.
• Application Number:
  US08/466,598
• Time in Patent Office: 805 Days
• Field of Search: 358/500, 358/518, 364/526, 382/164, 382/168, 382/272, 395/129
• US Class Current: 358/500
• CPC Class Codes: G06T 11/001 Texturing; Colouring; Generation of texture or colour inpainting G06T5/77

Abstract

A subject color and a non-subject color of a mask is specified, and a first color vector representing the subject color and a second color vector representing the non-subject color are determined. A third color vector is set to be linearly independent of the first and second color vectors. First through third coefficients are then determined for the first through third color vectors. Bit map data representing the distribution of the first coefficient is binarized to give binary mask data representing a mask. A mask area and a non-mask area are distinguishably displayed in the color image. Accordingly, the present invention will produce a mask area whose color is close to the subject color while excluding the non-subject color.

Claims

1. A method of generating mask data representing a desired color area in a color image, said method comprising the steps of:

(a) specifying n pieces of subject colors and m pieces of non-subject colors for defining said desired color area, where m and n are integers;

(b) obtaining n pieces of first color vectors for said n pieces of subject color in a color space;

(c) obtaining m pieces of second color vectors for said m pieces of non-subject colors in said color space;

(d) forming n×

m combinations of said first and second color vectors, and obtaining a third color vector linearly independent of the first and second color vectors with respect to each of said n×

m combinations to form n×

m sets of bases composed of said first through third color vectors;

(e) expressing color of each pixel included in said color image by a linear combination of said first through third color vectors with respect to each of said n×

m sets of bases, to thereby obtain first through third coefficients for said first through third color vectors with respect to each of said n×

m sets of bases; and

(f) generating said mask data as a function of said first coefficient obtained with respect to each of said n×

m sets of bases.

2. A method in accordance with claim 1, wherein said step (d) comprises the steps of:

transforming each of said mxn combinations of said first and second color vectors to fourth and fifth color vectors in a hue/saturation/brightness space; and

selecting a vector representing white in said color space as said third color vector when saturation components of said fourth and fifth color vectors are more than a predetermined value; and

when a saturation component of at least one of said fourth and fifth color vectors is less than said predetermined value, rotating a hue component of one said fourth and fifth color vectors, whose saturation component is the greater, by a predetermined angle in said hue/saturation/brightness space to generate a sixth color vector and transforming said sixth color vector into said color space to generate said third color vector.

3. A method in accordance with claim 2, wherein each of said m and n is equal to one.

4. A method in accordance with claim 3, wherein said step (f) comprise the step of:

binarizing said first coefficient with a threshold value to generate binary mask data.

5. A method in accordance with claim 2, further comprising the steps of:

(g) displaying a mask area represented by said mask data distinguishably from a non-mask area in said color image on a display device; and

(i) executing the processing of said steps (a) through (g) on said mask area in said color image while specifying n'"'"' pieces of subject colors and m'"'"' pieces of non-subject colors at said step (a), to thereby generate second mask data representing a second mask area within said mask area.

6. A method in accordance with claim 5, wherein each of said m, m'"'"', n and n'"'"' is equal to one.

7. A method in accordance with claim 2, wherein said step (f) comprises the step of:

determining a value of said mask data for each pixel in said color image by the steps of;

(1) determining a maximum of n pieces of first coefficients with respect to each pixel in said color image, said n pieces of first coefficients being obtained with respect to n pieces of bases which include an identical one of said m pieces of second color vectors, to thereby obtain m pieces of said maximums for said m pieces of second color vectors;

(2) determining a minimum of said m pieces of said maximums with respect to each pixel in said color image; and

(3) determining said value of said mask data for each pixel in said color image as a function of said minimum of said m pieces of said maximums.

8. A method in accordance with claim 3, wherein said step (3) comprises the step of:

binarizing said minimum with a threshold value to generate binary mask data.

9. A method in accordance with claim 8, wherein said step of binarizing comprises the steps of:

preparing a histogram showing frequency of said minimum in said color image; and

determining a local minimum of said histogram as said threshold value.

10. A method in accordance with claim 2, wherein said step (a) comprises the steps of:

(1) obtaining statistical color distribution of said color image;

(2) determining a plurality of representative colors of said color image from said statistical color distribution, and displaying said plurality of representative colors on a display device; and

(3) selecting said n pieces of subject colors and said m pieces of non-subject colors from said plurality of representative colors displayed on said display device.

11. A method in accordance with claim 10, wherein said step (1) comprises the step of:

dividing said color space into a plurality of divisional spaces; and

obtaining a frequency of each of said plurality of divisional spaces within said color images as said statistical color distribution; and

wherein

said step (2) comprises the steps of;

displaying said plurality of representative colors corresponding to at least part of said plurality of divisional spaces according to said frequency.

12. A method in accordance with claim 11, wherein said step of dividing said color space comprises the steps of:

obtaining three histogram for three color components of said color space concerning said color image; and

finding local minimums for said three histograms, and setting said local minimums as borders of said plurality of divisional spaces.

13. A method in accordance with claim 2, wherein said step (a) comprises the steps of:

displaying a color selection image for selecting arbitrary color components on a display device; and

specifying said n pieces of subject colors and said m pieces of non-subject colors while using said color selection image.

14. A method in accordance with claim 13, further comprising, between said steps (d) and (e), the steps of:

(1) selecting at least one of said color selection image and a pixel-skipped image of said color image as a subject image;

(2) expressing color of each pixel included in said subject image by a linear combination of said first through third color vectors with respect to each of said n×

m sets of bases, to thereby obtain first through third coefficients for said first through third color vectors with respect to each of said n×

m sets of bases;

(3) generating preliminary mask data as a function of said first coefficient obtained with respect to each of said n×

m sets of bases in said step (2) for each pixel included in said subject image; and

(4) displaying a preliminary mask area represented by said preliminary mask data distinguishably from a preliminary non-mask area in said subject image on said display device.

15. A method in accordance with claim 14, further comprising the step of:

repeatedly executing the processing of said steps (a) through (d) and (1) through (4) in response to a demand from a user while re-selecting said n pieces of subject colors and said m pieces of non-subject colors at said step (a).

16. A method in accordance with claim 15, wherein each of said m and n is equal to one.

17. An apparatus for generating mask data representing a desired color area in a color image, said apparatus comprising:

means for specifying n pieces of subject colors and m pieces of non-subject colors for defining said desired color area, where m and n are integers;

means for obtaining n pieces of first color vectors for said n pieces of subject color in a color space;

means for obtaining m pieces of second color vectors for said m pieces of non-subject colors in said color space;

means for forming n×

m combinations of said first and second color vectors, and obtaining a third color vector linearly independent of the first and second color vectors with respect to each of said n×

m combinations to form n×

m sets of bases from said first through third color vectors;

means for expressing color of each pixel included in said color image by a linear combination of said first through third color vectors with respect to each of said n×

m sets of bases, to thereby obtain first through third coefficients for said first through third color vectors with respect to each of said n×

m sets of bases;

means for generating said mask data as a function of said first coefficient obtained with respect to each of said n×

m sets of bases.

18. An apparatus in accordance with claim 17, wherein each of said m and n is equal to one.

19. An apparatus in accordance with claim 17, wherein said means for specifying colors comprises:

means for obtaining statistical color distribution of said color image;

means for determining a plurality of representative colors of said color image from said statistical color distribution, and displaying said plurality of representative colors on a display device; and

means for selecting said n pieces of subject colors and said m pieces of non-subject colors from said plurality of representative colors displayed on said display device.

20. An apparatus in accordance with claim 17, wherein said means for specifying colors comprises:

means for displaying a color selection image for selecting arbitrary color components on a display device; and

means for specifying said n pieces of subject colors and said m pieces of non-subject colors while using said color selection image.