Red-eye detection and correction

1. A processing method comprising:

receiving a digital image into memory of an image processing apparatus;

for each of at least a preponderance of pixels of the digital image or a reduced pixel resolution digital image therefrom, determining a probability that the pixel is in a red-eye as a function of a color of the pixel;

for each of a plurality of the pixels, determining a correction to apply to the pixel which is a function of the determined probability (P) that the pixel is within a red-eye and a color of the pixel;

applying a correction to the digital image or reduced pixel resolution image based on the determined corrections; and

outputting the corrected digital image in a format suitable for viewing;

wherein the correction of the pixels includes replacing a red component value of pixels having a probability (P) of being in a red-eye of greater than zero with a corrected red component value; and

wherein the corrected red component value is derived from the expression:

R corrected=(1− P ^(1/degree))* R original+ P ^(1/degree)*red Reduced,

where Roriginal is the red component value of the pixel, redReduced is an average of the green and blue component values of the pixel and degree is from about 1 to about 4.

2. The method of claim 1 wherein the probability of the pixel being in a red-eye is scaled over the digital image.

3. The method of claim 1 , wherein the probability of the pixel being in a red-eye has a value of from 0 to a maximum value.

4. The method of claim 1 , wherein the determining of the probability includes determining a redness probability which is a function of red, green, and blue components of the pixel.

5. The method of claim 4 , wherein the probability that the pixel is in a red-eye is a function of the extent to which the pixel is more red than it is green and blue.

6. The method of claim 1 , wherein the determining of the probability that the pixel is within a red-eye includes assigning a redness/luminance probability which is a function of the extent to which the pixel is more red than it is green and blue and also the extent to which the pixel is more red than bright.

7. The method of claim 6 , wherein the assignment of the redness/luminance value for the pixel includes subtracting the mean of a pixel's green and blue component values from its red component value to determine a redness value and further reducing the redness value by subtracting a weighted sum of red, blue and green components of the pixel.

8. The method of claim 1 , wherein a plurality of the pixels in the image are assigned a probability of the pixel being in a red-eye of between 0 and 1.

9. The method of claim 1 , further comprising:

determining the correction to apply to pixels in the digital image which, for each of the pixels, is a function of the assigned probability that the corresponding pixel in the reduced pixel resolution image is within a red-eye and the color of the pixel in the digital image.

10. A processing method comprising:

receiving a digital image into computer memory;

using a computer, determining a probability for each pixel of the entire image that the pixel is in a red-eye as a function of a color of the pixel;

thereafter, based on the determined probabilities, identifying any regions in the image of contiguous pixels which satisfy at least one test for a red-eye, the at least one test relating to one of the group consisting of a size of the region, a shape of the region, and an extent of overlap with a region comprising pixels having at least a threshold determined probability of being in a red-eye; and

for each of the plurality of pixels in an identified region, determining the correction to apply to the pixel which is a function of the determined probability that the pixel is within a red-eye and the color of the pixel;

applying a correction to pixels of the digital image based on the determined correction; and

wherein the correction of the pixels includes replacing a red component value of the pixels having a probability (P) of being in a red-eye of greater than zero with a corrected red component value; and

wherein the corrected red component value is derived from the expression:

R corrected=(1− P ^(1/degree))* R original+ P ^(1/degree)*redReduced,

where Roriginal is the red component value of the pixel, redReduced is an average of the green and blue component values of the pixel and degree is from about 1 to about 4.

11. The method of claim 10 , wherein the identification of regions includes identifying pixels which meet a spatial characteristic, the spatial characteristic being related to a degree of circularity of a contiguous region in which the pixel is located; and

identifying any regions within the image comprising contiguous pixels which are identified as meeting the spatial characteristic.

12. The method of claim 10 , wherein the method includes determining probability values for a plurality of sets of pixels, each set of pixels lying on a circle which is centered on the pixel and having a different radius.

13. The method of claim 10 , wherein the pixels in regions which do not satisfy the at least one test are assigned a probability that the pixel is in a red-eye of 0.

14. The method of claim 10 , wherein the at least one test for a red-eye evaluates a size of the region in relation to a size of the digital image.

15. The method of claim 10 , wherein the at least one test for a red-eye evaluates at least one of an extent to which the region is compact and an extent to which the region has equal height and width.

16. The method of claim 10 , wherein the determining of the probability to the pixels, the identifying of regions in the image of contiguous pixels which satisfy at least one test for a red-eye, and the determining a correction to apply to the pixels is performed automatically.

17. The processing method of claim 10 , further comprising outputting the corrected digital image in a format suitable for viewing.

18. A computer system for correction of images comprising:

a computer processing module including:

a detection module which, for each of at least a preponderance of pixels in an optionally reduced pixel resolution digital image, determines a probability that a pixel is within a red-eye as a function of a color of the pixel, the probability having a value which is variable from a minimum value to maximum value;

a module which determines whether a region in which a pixel is located meets at least one test for being a red-eye based on the determined probabilities of the pixels;

a correction module which determines a correction to apply to the pixels in regions that meet the at least one test, the correction being a function of the determined probability and of a color of the pixel;

memory, which stores the digital image that is processed by the modules;

wherein the correction of the pixels includes replacing a red component value of the pixels having a probability (P) of being in a red-eye of greater than zero with a corrected red component value; and

wherein the corrected red component value is derived from the expression:

R corrected=(1− P ^(1/degree))* R original+ P ^(1/degree)*red Reduced,

where Roriginal is the red component value of the pixel, red Reduced is an average of the green and blue component values of the pixel and degree is from about 1 to about 4.

19. The system of claim 18 , wherein the detection module further comprises a component which assigns a circularity probability to each of a plurality of the pixels.

20. The system of claim 18 , further comprising:

a user interface which allows a user to view the corrected image and accept or decline the correction.

21. A processing method comprising:

storing a digital image or a reduced pixel resolution digital image therefrom in memory;

using a computer, for each of at least a preponderance of pixels of the digital image or the reduced resolution digital image, automatically determining a probability of the pixel being in a red-eye as a function of a color of the pixel;

based on the determined probabilities, identifying any regions in the digital image or the reduced pixel resolution digital image of contiguous pixels which satisfy at least one test for a red-eye, the at least one test relating to one of the group consisting of a size of the region, a shape of the region, and an extent of overlap with a region comprising pixels having at least a threshold probability of being in a red-eye;

for each of the plurality of pixels in an identified region, automatically determining the correction to apply to the pixel which is a function of the assigned probability that the pixel is within a red-eye and the color of the pixel;

applying a correction to pixels of the digital image based on the determined correction; and

wherein the correction of the pixels includes replacing a red component value of the pixels having a probability (P) of being in a red-eye of greater than zero with a corrected red component value; and

wherein the corrected red component value is derived from the expression:

R corrected=(1− P ^(1/degree))* R original+ P ^(1/degree)*redReduced,

where Roriginal is the red component value of the pixel, redReduced is an average of the green and blue component values of the pixel and degree is from about 1 to about 4.

22. The processing method of claim 21 , further comprising outputting the corrected digital image in a format suitable for viewing.

23. A processing method comprising:

storing an optionally reduced pixel resolution digital image into memory;

using a computer, for each of at least a preponderance of pixels of a digital image, assigning a probability to the pixel of the pixel being in a red-eye as a function of a color of the pixel, the probabilities varying between maximum and minimum values;

generating a probability map of the assigned probabilities for the entire image;

identifying regions of the image which have at least a minimum probability of being a red-eye, based on the assigned probabilities;

computing a circularity probability value of the pixels in the identified regions which is related to the degree of circularity of a region in which the pixel is located;

applying one or more tests to the regions to eliminate false positive regions; and

for each of the plurality of pixels in at least the regions remaining after elimination of the false positive regions, determining a correction to apply to the pixel which is a function of the assigned probability that the pixel is within a red-eye and a color of the pixel;

applying a correction to pixels of the digital image based on the determined correction; and

wherein the correction of the pixels includes replacing a red component value of the pixels having a probability (P) of being in a red-eye of greater than zero with a corrected red component value; and

wherein the corrected red component value is derived from the expression:

R corrected=(1− P ^(1/degree))* R original+ P ^(1/degree)*redReduced,

where Roriginal is the red component value of the pixel, redReduced is an average of the green and blue component values of the pixel and degree is from about 1 to about 4.

24. The processing method of claim 23 , further comprising outputting the corrected digital image in a format suitable for viewing.