Method for adjusting the brightness of a digital image utilizing belief values

• US 7,609,908 B2
• Filed: 04/30/2003
• Issued: 10/27/2009
• Est. Priority Date: 04/30/2003
• Status: Expired due to Fees

• Current Assignee: Monument Peak Ventures, LLC (Dominion Harbor Enterprises, LLC)
• Original Assignee: Eastman Kodak Company
• Inventors: Dupin, Michael W., Luo, Jiebo, Gallagher, Andrew C.
• Primary Examiner(s): Werner; Brian P
• Assistant Examiner(s): DULANEY, KATHLEEN YUAN
• Application Number:
  US10/426,592
• Publication Number:
  US 20040218832A1
• Time in Patent Office: 2,372 Days
• Field of Search: 382/274, 382/118, 382/103, 382/167, 345/58, 348/222.1
• US Class Current: 382/274
• CPC Class Codes: H04N 1/6027 Correction or control of colour gradation or colour contrast H04N1/6058 takes precedence
• CPC Class Codes: H04N 1/628 Memory colours, e.g. skin or sky

Abstract

A method of processing a digital image, including the steps of: assigning belief values to pixels in the digital image, wherein the belief values are proportional to a belief that the pixels are associated with human skin; calculating a brightness adjustment amount based on a statistic of the pixels with belief values exceeding a threshold and the assigned belief values of the pixels; and applying the brightness adjustment amount to the digital image to produce a processed digital image having improved overall brightness.

Claims

1. A method of processing a digital image, comprising the steps of:

a) assigning belief values according to instructions stored in a computer readable medium to pixels in the digital image, the belief values are subject to;

(i) detecting a probability of both human skin color pixels and non-skin color pixels in the digital image and (ii) computing a geometric feature for spatially contiguous regions of skin-colored pixels, wherein the belief values are proportional to a belief that the skin color and non-skin color pixels are associated with human skin;

b) ascertaining an image dependant adaptive skin probability threshold using both the human skin color pixels and the non-skin color pixels having desired gradient values;

c) calculating a brightness adjustment amount based on a statistic of the pixels with belief values exceeding the skin probability threshold and the assigned belief values of the pixels; and

d) applying the brightness adjustment amount to the digital image to produce a processed digital image having improved overall brightness.

2. The method claimed in claim 1, wherein the step of assigning the belief values includes the steps of:

a1) assigning skin probability values to pixels in the digital image, wherein the skin probability values are proportional to the probability that the pixels are associated with human skin color;

a2) extracting spatially contiguous regions of the pixels with skin probability values exceeding the threshold; and

a3) producing a main subject skin belief map comprising belief values indicating a belief in each extracted skin-colored region being part of a main human subject, wherein the main subject skin belief map indicates skin-colored regions, in terms of location and the assigned belief value, proportional to the belief that the skin-colored region is associated with the main human subject, having skin pixels, and located in the digital image.

3. The method claimed in claim 2, wherein the step of producing the main subject skin belief map includes the steps of:

a3a) computing at least one geometric feature for each spatially contiguous skin-colored region, selected from the group consisting of a centrality feature, a borderness feature, a compactness feature, and an aspect ratio feature;

a3b) computing the main subject skin belief value for each skin-colored region based on the geometric feature and/or an average skin color probability of the region; and

a3c) assigning all pixels of each spatially contiguous skin-colored region the same main subject skin belief value.

4. The method claimed in claim 1, wherein the step of calculating a brightness adjustment amount includes the step of:

c1) calculating a belief weighted average for brightness of the detected skin-colored pixels by a function of the assigned belief values.

5. The method claimed in claim 4, wherein the function is a linear function such that Brightness Adjustment=B*(belief value*average brightness−

A), where A and B are predetermined constants.

6. The method claimed in claim 4 wherein the function is a nonlinear function such that Brightness Adjustment=B*[f(belief value)*average brightness−

A], where A and B are predetermined constants and f(.) is a nonlinear function.

7. The method claimed in claim 1, wherein an initial scene balance algorithm is applied to the digital image before assigning the probability values to the skin-colored pixels.

8. The method claimed in claim 7, wherein the initial scene balance algorithm is the Evans gray world algorithm.

9. The method claimed in claim 7, wherein the initial scene balance algorithm is the LATD brightness scene balance algorithm.

10. The method claimed in claim 7, wherein the initial scene balance algorithm includes the Kwon color scene balance algorithm.

11. The method claimed in claim 7 for deriving the statistic of the detected skin-colored pixels, in the digital image, comprising the steps of:

c1) for each pixel in the digital image, calculating a probability that the pixel is a skin-colored pixel;

c2) determining the skin color probability threshold based on the statistic derived from the digital image; and

c3) designating any pixels having a probability greater than the skin color probability threshold as skin-colored pixels.

12. The method claimed in claim 11, wherein calculation of the digital image, expressed in RGB coordinates, and the probability that the pixel is a skin-colored pixel includes the steps of:

c1a) converting the digital image from RGB to Lst coordinates; and

c1b) determining the skin color probability using a predetermined skin probability density function expressed in Lst coordinates developed from a population of images containing skin and non skin-colored pixels.

13. The method claimed in claim 11, wherein the step of determining the skin color probability threshold includes the steps of:

c2a) a forming a skin probability map using skin color probability values of all the pixels in the image;

c2b) computing gradient values from the skin probability map;

c2c) selecting pixels having gradient values greater than a predetermined value;

c2d) forming a histogram of skin probability values using only the selected pixels from the previous step;

c2e) locating a valley point in the histogram; and

c2f) determining the skin color probability threshold as the skin probability value at the located valley point of the histogram.

14. The method claimed in claim 1, further comprising the step of counting the number of detected skin-colored pixels, and performing the brightness adjustment only if a minimum number of skin-colored pixels are detected.

15. The method claimed in claim 1, further comprising the step of computing the maximum belief value of detected skin-colored pixels, and performing the brightness adjustment only if there is at least one region of skin-colored pixels that has belief values greater than a pre-determined threshold.

16. The method claimed in claim 1, wherein the digital image is obtained by digitization of a picture originally captured on film, by direct capture in a digital camera, or by scanning of a picture on printed media.

17. The method claimed in claim 1, further includes printing the processed digital image having improved overall brightness on photographic paper.

19. A computer storage medium having instructions stored therein for causing a computer to perform the method of claim 7.

18. A method of processing a digital image, comprising the steps of:

a) assigning a probability value to skin-colored pixels and non-skin-colored pixels in the digital image according to instructions stored in a computer readable medium;

b) producing a skin probability map for the image according to the probability value for each of the skin-colored pixels and non-skin-colored pixels in the digital image;

c) determining gradient values for both the skin and non-skin colored pixels in the digital image;

d) determining a variable skin probability threshold for the digital image according to selected gradient values of both skin and non-skin colored pixels in the digital image;

e) extracting spatially contiguous regions of skin-colored pixels;

f) producing a main subject skin belief map of belief values, wherein the belief values are calculated from the probability values of both the skin colored pixels and the non-skin colored pixels, the belief values indicate a belief in each extracted skin-colored region being pan of a main human subject, wherein the main subject skin belief map indicates skin-colored regions, in terms of location and the belief value, proportional to the belief that the skin-colored region is associated with the main human subject, having skin pixels, and located in the image;

g) calculating a brightness adjustment amount based on a statistic of the detected skin-colored pixels and the belief values of such pixels; and

h) applying the brightness adjustment amount to the digital image to produce a processed digital image having improved overall brightness.

20. A computer storage medium having instructions stored therein for causing a computer to perform the method of claim 18.

21. A method of processing a digital image, comprising the steps of:

a) assigning belief values according to instructions stored in a computer readable medium to pixels in the digital image, the belief values are subject to;

(i) detecting a probability of human skin color pixels and non-skin color pixels for each pixel in the digital image and (ii) computing a geometric feature for spatially contiguous regions of skin-colored pixels, wherein the belief values are proportional to a belief that the human skin color pixels and non-skin color pixels are associated with human skin;

b) calculating a brightness adjustment amount based on a statistic of the pixels with belief values exceeding an adaptive skin probability threshold and the assigned belief values of the pixels;

c) applying the brightness adjustment amount to the digital image to produce a processed digital image having improved overall brightness; and

d) customizing the adaptive skin probability threshold for the digital image by using an image-dependent adaptive thresholding algorithm dependent on the gradient of both the skin color pixels and the non-skin color pixels in the digital image.