Automatic red eye artifact reduction for images
View Patent ↗Systems and methods are provided for reducing eye coloration artifacts in an image. In the system and method, an eye is detected in the image and a pupil color for the eye in the image and a skin color of skin in the image associated with the eye are determined. At least one region of artifact coloration in the eye in the image is then identified based on the pupil color and the skin color, and a coloration of the region is modified to compensate for the artifact coloration.
1. A method for reducing eye coloration artifacts in an image, comprising:
detecting an eye in the image;
determining a pupil color for the eye in the image and a skin color of skin in the image adjacent to the eye;
identifying at least one region of artifact coloration in the eye in the image based on the pupil color and the skin color;
modifying a coloration of the region to compensate for the artifact coloration wherein the modifying includes forming an image mask specifying an amount of color correction for at least pixels associated with the region; and
adjusting a coloration of at least the region based on the image mask.
2. The method of claim 1 , wherein detecting the eye in the image comprises employing at least one among a face detection algorithm and an eye location algorithm.
3. The method of claim 1 , wherein determining the pupil color and the skin color further comprises:
selecting a bounding area of the image including the eye and the skin adjacent to the eye; and
sampling pixels in the bounding area associated with a pupil of the eye in the image and the skin.
4. The method of claim 1 , wherein determining the pupil color and the skin color further comprises:
selecting a bounding area of the image centered on the eye;
sampling pixels in a central area of the bounding area to determine the pupil color; and
sampling pixels along a peripheral area of the bounding are to determine the skin color.
5. The method of claim 1 , wherein identifying the region of artifact coloration further comprises:
segmenting pixels associated with the eye in the image into pupil and non-pupil pixels based at least on the pupil color and the skin color.
6. The method of claim 1 , wherein modifying the coloration of the region further comprises:
performing falsing classification on the identified region of artifact coloration to generate a falsing classifier; and
adjusting a coloration of at least the region based on at least the falsing classifier.
7. An image processing system, comprising:
a processing element that is arranged to receive an input image; and further arranged for: detecting an eye in the image;
determining a pupil color for the eye in the image and a skin color of skin in the image adjacent to the eye;
identifying at least one region of artifact coloration in the eye in the image based on the pupil color and the skin color;
modifying a coloration of the region to compensate for the artifact coloration, wherein the processing element modifies the coloration of the region based on forming an image mask specifying an amount of color correction for at least pixels associated with the region; and
adjusting a coloration of at least the region based on the image mask.
8. The system of claim 7 , wherein the processing element employs at least one among a face detection algorithm and an eye location algorithm for detecting the eye in the image.
9. The system of claim 7 , wherein the processing element determines the pupil color and the skin color based on selecting a bounding area of the image including the eye and the skin adjacent to the eye, and sampling pixels in the bounding area associated with a pupil of the eye in the image and the skin.
10. The system of claim 7 , wherein the processing element determines the pupil color and the skin color based on selecting a bounding area of the image centered on the eye, sampling pixels in a central area of the bounding area to determine the pupil color, and sampling pixels along a peripheral area of the bounding are to determine the skin color.
11. The system of claim 7 , wherein the processing element identifies the region of artifact coloration based on segmenting pixels associated with the eye in the image into pupil and non-pupil pixels based at least on the pupil color and the skin color.
12. The system of claim 7 , wherein the processing element modifies the coloration of the region based on performing falsing classification on the identified region of artifact coloration to generate a falsing classifier and adjusting a coloration of at least the region based on at least the falsing classifier.
13. A processor readable non-volatile storage medium that stores one or more components that, when executed on an image processing system, enable actions for reducing coloration artifacts in an image of an eye, comprising:
determining a pupil color for the eye in the image and a skin color of skin in the image associated with the eye;
identifying at least one region of artifact coloration in the eye in the image based on the pupil color and the skin-color;
modifying a coloration of the region to compensate for the artifact coloration, wherein the modifying includes forming an image mask specifying an amount of color correction for at least pixels associated with the region; and
adjusting a coloration of at least the region based on the image mask.
14. The processor readable non-volatile storage medium of claim 13 , wherein detecting the eye in the image comprises employing at least one among a face detection algorithm and an eye location algorithm.
15. The processor readable non-volatile storage medium of claim 13 , wherein determining the pupil color and the skin color further comprises:
selecting a bounding area of the image including the eye and the skin adjacent to the eye; and
sampling pixels in the bounding area associated with a pupil of the eye in the image and the skin.
16. The processor readable non-volatile storage medium of claim 13 , wherein determining the pupil color and the skin color further comprises:
selecting a bounding area of the image centered on the eye;
sampling pixels in a central area of the bounding area to determine the pupil color; and
sampling pixels along a peripheral area of the bounding are to determine the skin color.
17. The processor readable non-volatile storage medium of claim 13 , wherein identifying the region of artifact coloration further comprises:
segmenting pixels associated with the eye in the image into pupil and non-pupil pixels based at least on the pupil color and the skin color.
18. The processor readable non-volatile storage medium of claim 13 , wherein modifying the coloration of the region further comprises:
performing falsing classification on the identified region of artifact coloration to generate a falsing classifier; and
adjusting a coloration of at least the region based on at least the falsing classifier.