Methods and apparatus for automated, multi-level red eye correction

1. A non-transitory machine readable medium that includes instructions that, when executed by the machine, cause the machine to perform operations, comprising:

computing a set of feature scores, wherein each feature score corresponds to a different eye feature and indicates the likelihood that the eye feature is present in a red eye candidate zone within an image;

computing a red eye candidate score for the red eye candidate zone based on the set of feature scores;

determining that the red eye candidate score exceeds a detection threshold value indicating that the red eye candidate zone corresponds to an actual red eye;

applying a partial level of correction to the actual red eye when the red eye candidate score is less than a full correction threshold, or

applying a full level of correction to the actual red eye when the red eye candidate score is greater than or equal to the full correction threshold.

2. The non-transitory machine readable medium that contains instructions of claim 1 , wherein the partial level of correction corrects a chroma value associated with pixels forming the actual red eye in the digital image.

3. The non-transitory machine readable medium that contains instructions of claim 1 , wherein the full level of correction corrects a chroma value and a luminance value associated with pixels forming the actual red eye in the digital image.

4. The non-transitory machine readable medium that contains instructions of claim 1 , wherein the full level of correction creates a highlight in the actual red eye in the digital image.

5. The non-transitory machine readable medium that contains instructions of claim 1 , wherein the full level of correction corrects a ring around the actual red eye using a diminishing correction ratio.

6. The non-transitory machine readable medium that contains instructions of claim 3 , wherein the chroma value is corrected toward a brown coloration.

7. The non-transitory machine readable medium that contains instructions of claim 1 , which when executed by the machine perform the further operations, comprising: detecting a single red eye candidate above the preset threshold and applying the level of correction to a pair of eyes.

8. The non-transitory machine readable medium that contains instructions of claim 1 , wherein the machine readable medium is contained within a processing unit and the processing unit receives the digital image from a one time use digital camera.

9. A computer-implemented method for red-eye correction, comprising one or more computers implementing in the following steps:

computing a set of feature scores, wherein each feature score corresponds to a different eye feature and indicates the likelihood that the eye feature is present in a red eye candidate zone within an image;

computing a red eye candidate score for the red eye candidate zone based on the set of feature scores;

determining that the red eye candidate score exceeds a detection threshold value indicating that the red eye candidate zone corresponds to an actual red eye; and

applying a partial level of correction to the actual red eye when the red eye candidate score is less than a full correction threshold; or

applying a full level of correction to the actual red eye when the red eye candidate score is greater than or equal to the full correction threshold.

10. The method of claim 9 , wherein the partial level of correction compensates pixels in an identified red eye candidate zone rather than compensating an entire pupil, iris, and white portion of the eye.

11. A computer system, comprising:

a memory; and

a processing unit configured to:

compute a set of feature scores, wherein each feature score corresponds to a different eye feature and indicates the likelihood that the eye feature is present in a red eye candidate zone;

compute a red eye candidate score for the red eye candidate zone based on the set of feature scores;

determine that the red eye candidate score exceeds a detection threshold value indicating that the red eye candidate zone corresponds to an actual red eye; and

apply a partial level of correction to the actual red eye when the red eye candidate score is less than a full correction threshold; or

apply a full level of correction to the actual red eye when the red eye candidate score is greater than or equal to the full correction threshold.

12. The computer system of claim 11 , wherein the memory stores instructions that, when executed by the processing unit, configure the processing unit to compute the set of feature scores, compute the red eye candidate score for the red eye candidate zone, determine that the detection threshold value is exceeded and apply the partial level of correction or the full level of correction.