Automatic detection and correction of non-red eye flash defects

1. A method for detecting flash eye defects in an image, comprising using a processor in performing the following:

defining one or more luminous regions in said image, each region having at least one pixel having luminance above a luminance threshold value;

grouping neighboring selected pixels into said one or more luminous regions;

for each of said one or more luminous regions, applying a face filter for determining a probability that said region is located within a region of an image characteristic of a face;

for each region determined as having at least a threshold probability of being within a face, determining a further probability whether said region has a size, shape or color, or combinations thereof, characteristic of a pupil region of an eye region of a face;

for each region having at least a further threshold probability of having a size, shape or color, or combinations thereof, characteristic of a pupil region of an eye region of a face, determining whether said region corresponds to a flash eye defect including determining that said region does correspond to a candidate flash eye defect region upon determining that a flash was used when the image was acquired, or that said region does not correspond to a flash eye defect region upon determining that a flash was not used when the image was acquired; and

storing, displaying, transmitting, or printing said image or a corrected or further processed version of said image, or combinations thereof.

2. A method according to claim 1 , wherein said defining comprises selecting pixels of the image which have a luminance above a luminance threshold value and a redness above or below a redness threshold value.

3. The method according to claim 1 , further comprising correcting said flash eye defect.

4. The method according to claim 1 , further comprising applying a filter for adding pixels to a luminous region located with said luminous region and which have luminance below said luminance threshold value or a redness above or below a redness threshold value.

5. The method of claim 1 , wherein said shape comprises roundness, and the method further comprises calculating said roundness to determine if said region corresponds to an eye feature.

6. The method according to claim 1 , further comprising, for each luminous region, determining a corresponding aggregated region by: (i) determining a seed pixel for said aggregated region within a luminous region; and (ii) iteratively adding non-valley neighbouring pixels to said aggregated region until no non-valley neighboring pixels adjacent to said aggregated region remain.

7. The method according to claim 1 , further comprising calculating an intensity gradient of each luminous region.

8. A digital image processing device operable to detect flash eye defects in an image, and comprising a controller arranged to:

define one or more luminous regions in said image, each region having at least one pixel having luminance above a luminance threshold value;

group neighboring selected pixels into said one or more luminous regions;

for each of said one or more luminous regions, apply a face filter for determining a probability that said region is located within a region of an image characteristic of a face;

for each region determined as having at least a threshold probability of being within a face, determine a further probability whether said region has a size, shape or color, or combinations thereof, characteristic of a pupil region of an eye region of a face;

for each region having at least a further threshold probability of having a size, shape or color, or combinations thereof, characteristic of a pupil region of an eye region of a face, determine whether said region corresponds to a flash eye defect including determining that said region does correspond to a candidate flash eye defect region opon determining that a flash was used when the image was acquired, or that said region does not correspond to a flash eye defect region upon determining that a flash was not used when the image was acquired; and

store, display, transmit, or print said image or a corrected or further processed version of said image, or combinations thereof.

9. A device according to claim 8 , comprising: a digital camera or camera phone, a general purpose, portable or hand-held computer, a printer or a digital scanner, or combinations thereof.

10. One or more digital storage devices having executable program code embodied thereon for programming one or more processors to perform a method of correcting a flash eye defect in a digital image, the method comprising:

acquiring a digital image;

determining a luminance of pixels within the digital image;

selecting those pixels having a luminance above a certain threshold as candidate regions for correction of a flash eye defect;

defining one or more luminous regions in said image, including determining that each region has each of the following:

at least a threshold probability of having a size, shape or color, or combinations thereof, characteristic of a pupil region of an eye region of a face; and

at least one pixel having luminance above a luminance threshold value; and

determining that said region does correspond to a candidate flash eye defect region, upon determining that a flash was used when the image was acquired, or that said region does not correspond to a flash eye defect region upon determining that a flash was not used when the image was acquired;

filtering the selected pixels; and

correcting the white eye defect for non-filtered pixels among the selected pixels.

11. The one or more storage devices of claim 10 , wherein the filtering comprises geometrical filtering of pixels based on a size or shape or both of a selected pixel region.

12. The one or more storage devices of claim 10 , wherein the filtering comprises geometrical filtering of pixels based on a size of a selected pixel region being above a threshold size.

13. The one or more storage devices of claim 10 , wherein the correcting comprising calculating a roundness of a selected pixel region, and correcting the roundness if it does not exceed a certain threshold value of roundness.

14. The one or more storage devices of claim 10 , wherein the filtering comprises checking whether an average saturation of a selected pixel region exceeds a certain threshold saturation, and correcting the selected pixel region only if the threshold is exceeded.

15. The one or more storage devices of claim 10 , the method further comprising: (i) selecting a bright pixel as a seed pixel; and (ii) aggregating outwardly from the seed pixel to combine those pixels that are not valley points with the seed pixel as an aggregated region until a minimum number of non-valley neighbors are left or a threshold size is reached, or a combination thereof.

16. The one or more storage devices of claim 15 , the method further comprising smoothing the aggregated region.

17. The one or more storage devices of claim 15 , the method further comprising computing an intensity gradient for one or more candidate regions.

18. The one or more storage devices of claim 10 , the method further comprising filtering a candidate region that comprises merely a glint.

19. The one or more storage devices of claim 10 , the method further comprising detecting and correcting a red eye defect within the digital image.

20. The method of claim 1 , further comprising applying a glint filter that filters any candidate region that comprises merely a glint.

21. The device of claim 8 , wherein the controller is further arranged to apply a glint filter that filters any candidate region that comprises merely a glint.

22. The one or more storage devices of claim 10 , wherein the defining further including determining that each luminous region has a measure of redness below a redness threshold value determined as a ratio of a red component value of the at least one pixel to one or both of a blue component value and a green component value of the at least one pixel.

23. The method of claim 1 , further comprising determining a further probability that each luminous region has a measure of redness below a redness threshold value determined as a ratio of a red component value of the at least one pixel to one or both of a blue component value and a green component value of the at least one pixel.

24. The device of claim 8 , wherein the controller is further arranged to determining a further probability that each luminous region has a measure of redness below a redness threshold value determined as a ratio of a red component value of the at least one pixel to one or both of a blue component value and a green component value of the at least one pixel.