SYSTEM AND METHOD FOR CHANGING HAIR COLOR IN DIGITAL IMAGES

• US 20120075331A1
• Filed: 09/24/2010
• Published: 03/29/2012
• Est. Priority Date: 09/24/2010
• Status: Active Grant

• Current Assignee: TAAZ Incorporated
• Original Assignee: TAAZ Incorporated
• Inventors: Mallick, Satya P., Kriegman, David, Qiao, Hui
• Granted Patent: US 8,884,980 B2
• Time in Patent Office: Days
• US Class Current: 345/594
• CPC Class Codes: A45D 44/005 for selecting or displaying personal cosmetic colours or hairstyle
• CPC Class Codes: G06T 11/001 Texturing; Colouring; Generation of texture or colour inpainting G06T5/77
• CPC Class Codes: G06T 11/60 Editing figures and text; Combining figures or text
• CPC Class Codes: G06T 2207/10024 Color image
• CPC Class Codes: G06T 2207/30196 Human being; Person
• CPC Class Codes: G06V 40/162 using pixel segmentation or colour matching
• CPC Class Codes: G06V 40/171 Local features and components; Facial parts eye characteristics G06V40/18; Occluding parts, e.g. glasses; Geometrical relationships

Abstract

A system and method for digital hair coloring is described. A starting image of a subject including a hair region is obtained. At least one foreground selection of the starting image is obtained, the foreground selection including a portion of the hair region. At least one background selection of the starting image excluding the hair region is obtained. A matte defining a calculated foreground area of the starting image is calculated based on the foreground selection and the background selection. A target color distribution based on a target hair color is obtained. A color image is generated by applying a color transformation based on the target color distribution to at least the calculated foreground area of the starting image. An output image is generated from the colored image and the starting image.

Claims

1. A computer-implemented method for digital hair coloring comprising:

obtaining a starting image of a subject including a hair region;

calculating a matte defining a calculated foreground area of said starting image;

obtaining a target color distribution, wherein said target color distribution is based on a target hair color;

generating a colored image by applying a color transformation based on said target color distribution to at least said calculated foreground area of said starting image; and

generating an output image from said colored image and said starting image.

2. The computer-implemented method of claim 1, further comprising:

obtaining at least one foreground selection of said starting image comprising a portion of said hair region; and

obtaining at least one background selection of said starting image excluding said hair region,

wherein said matte is calculated based on said foreground selection and said background selection.

3. The computer-implemented method of claim 1, further comprising:

obtaining a bounding box around said hair region of said starting image;

detecting a skin region of said starting image;

obtaining one or more foreground points within said hair region of said starting image;

wherein said matte is calculated based on said bounding box, said skin region, and said one or more foreground points.

4. The computer-implemented method of claim 1,

wherein said generating said colored image comprises applying said color transformation to said calculated foreground area, and

wherein said generating said output image comprises combining said calculated foreground area after applying said color transformation and a calculated background area.

5. The computer-implemented method of claim 1,

wherein said generating said colored image comprises applying said color transformation to said starting image, and

wherein said generating said output image comprises combining said starting image after applying said color transformation and a calculated background area.

6. The computer-implemented method of claim 1, wherein:

said generating said colored image comprises applying said color transformation to said starting image to generate said colored image, and

said generating said output image further comprises combining a foreground area of said colored image with a background area of said starting image using said matte.

7. The computer-implemented method of claim 1, wherein:

said calculating said matte comprises calculating an alpha mask for said starting image comprising alpha values in a range from a background alpha value to a foreground alpha value, wherein said background alpha value indicates a background pixel, said foreground alpha value indicates a foreground pixel, and an alpha value between said background alpha value and said foreground alpha value indicates a proportion of foreground color and background color of a pixel; and

said generating said output image comprises using said alpha values to determine a colored image contribution to said output image and a starting image contribution to said output image.

8. The computer-implemented method of claim 2, wherein said at least one foreground selection and said at least one background selection are obtained by image processing.

9. The computer-implemented method of claim 1, wherein said target color distribution comprises three histograms corresponding to three color channels of a selected color space.

10. The computer-implemented method of claim 9, wherein said selected color space is selected from RGB, CIE Lab, HSV, YIQ, YUV, YCbCr, CMYK, CIE XYZ, and HSI.

11. The computer-implemented method of claim 9, wherein said generating said output image comprises:

applying a nonlinear transformation to said starting image to map a starting color distribution of said starting image to said target color distribution, wherein said color transformation is independent of pixel location.

12. The computer-implemented method of claim 11, wherein said nonlinear transformation is histogram matching applied to one or more color channels.

13. The computer-implemented method of claim 1, wherein said target color distribution is approximated by a multivariate normal distribution with a mean vector and covariance matrix in a selected color space.

14. The computer-implemented method of claim 13, wherein target principal axes of said target color distribution are estimated using principal component analysis.

15. The computer-implemented method of claim 13, wherein said generating of said output image comprises:

transforming starting principal axes of color distribution of said starting image to said target principal ellipsoid axes of said target color distribution using an affine transformation.

16. The computer-implemented method of claim 15, wherein said affine transformation comprises rotation, scaling and translation of colors.

17. The computer-implemented method of claim 1, wherein said target color distribution is extracted from a photographic image of hair representative of said target hair color.

18. The computer-implemented method of claim 1, further comprising recommending a hair color product associated with said target hair color, wherein said output image comprises a simulation of a result of applying said hair color product to said subject.

19. The computer-implemented method of claim 18, further comprising:

obtaining at least one characteristic of said subject capable of affecting said result of applying said hair color product; and

modifying at least one of said target color distribution and said color transformation based on said at least one characteristic.

20. The computer-implemented method of claim 1, wherein said target color distribution is further based on at least one hair characteristic selected from a starting hair color, a hair grayness, a hair damage level, a hair texture, a hair thickness, and an ethnicity hair type.

21. The computer-implemented method of claim 20, wherein said at least one hair characteristic is obtained by extracting one or more of said at least one characteristic from said starting image.

22. The computer-implemented method of claim 1, further comprising using image decomposition to obtain a plurality of target color distributions, wherein each of said plurality of target color distributions is applied to a corresponding region of said starting image to achieve an effect in said output image.

24. The computer-readable medium of claim 22, wherein said computer-executable instructions further performs steps comprising providing a user interface configured to allow a user to:

select said starting image;

select said at least one foreground selection of said starting image and said at least one background selection of said starting image; and

25. select said target hair color. The computer-readable medium of claim 22, wherein said user interface is further configured to allow said user to select said photographic image of said hair comprising said target hair color, wherein said target color distribution is extracted from said photographic image.

27. The computer-implemented method of claim 25, wherein the selected target color distribution is extracted from said photographic image based on at least said hair region of said photographic image.

23. A computer-readable medium for digitally coloring hair encoded with computer-executable instructions that when executed by a processor of a computer performs steps comprising:

obtaining a starting image of a subject including a hair region;

calculating a matte defining a calculated foreground area of said starting image;

obtaining a target color distribution, wherein said target color distribution is based on a target hair color;

generating a colored image by applying a color transformation based on said target color distribution to at least said calculated foreground area of said starting image;

generating an output image from said colored image and said starting image; and

displaying said output image.

26. A computer-implemented method for custom digital hair styling comprising:

obtaining a photographic image of a subject including a hair region;

obtaining a selected target color distribution;

obtaining a hairstyle image comprising a color image and an alpha mask;

applying a color transformation based on said selected target color distribution to said hairstyle image;

fitting said hairstyle image to a head of said subject in said starting image; and

generating an output image comprising said starting image and said hairstyle image based on said alpha mask.