Adaptable eye artifact identification and correction system
View Patent ↗An adaptable eye artifact identification and correction method is disclosed. Eye artifacts are identified and classified based on color, severity, shape, eye location, and cause. Based on this classification, an eye artifact correction algorithm is selected from a series of eye artifact correction techniques. For minor artifacts, simple color correction techniques are deployed to restore the iris color and to drive the pupil to once again appear black. For severe eye artifacts face detection and metadata analysis are utilized to search the user's image collection for recent images of the subject without the eye artifact condition. Once located, these images provide eye color and shape information for use to replace the pixels expressing the eye artifact condition. The non-artifact eye images are used to provide the appropriate eye color and shape to correct the eye artifact condition for more severe eye artifacts.
1. A method for identifying and correcting eye artifacts in digital photography comprising:
providing one or more digital images to a computer system;
initiating image analysis via an image processing software accessible through a user interface of the computer system;
identifying an eye artifact type and a subject eye type the one or more digital images;
determining eye artifact severity of the identified eye artifact;
determining an appropriate correction modality considering available subject eye type correction models, wherein the available subject eye type correction models comprise generic models selected and modified with other images having comparable size, color, shape, type, emotional state, and event window;
identifying related images within an event window of a subject without eye artifacts or with less severe eye artifacts using a combination of face detection and clustering algorithms;
selecting a correction modality and/or a related image and correcting the eye artifact;
outputting one or more corrected digital images having a corrected eye artifact, the corrected eye artifact being incorporated into the one or more digital images via application of the correction modality and/or the related image; and
obtaining a user approval.
2. The method of claim 1 wherein the eye artifact type comprises a camera related lighting induced artifact type, an ambient light induced artifact type, an eyewear related artifact type, or a physical condition related artifact type.
3. The method of claim 2 wherein if the artifact type is determined to be a physical condition related artifact type, then the user is alerted and verification is requested to proceed.
4. The method of claim 1 wherein the subject eye type comprises a human eye type or an animal eye type.
5. The method of claim 4 wherein the human eye type comprises classifications that are selected from the group consisting of race, gender, age, eye color, skin color, eyewear, or facial type.
6. The method of claim 4 wherein the animal eye type can be classified according to pupil type.
7. The method of claim 1 wherein determining eye artifact severity comprises determining whether the eye artifact is correctable or requires replacement.
8. The method of claim 1 wherein the available subject eye type correction models are based on eye artifact severity, subject eye type, and eye artifact type.
9. The method of claim 8 wherein the available subject eye type correction models rescale and reposition eyes to compensate for optical distortion due to eyeglasses.
10. The method of claim 8 wherein the available subject eye type correction models re-colorize eyes based on predicted or assigned color.
11. The method of claim 10 wherein the predicted or assigned color is determined using skin tone, hair color, metadata, a social network comment, or user input.
12. The method of claim 8 wherein the available subject eye type correction models re-colorize eyes based on color obtained from other images identified using face detection or image tagging within an event window.
13. The method of claim 8 wherein the available subject eye type correction models are used to correct existing eye images with non-artifact related eye images.
14. The method of claim 13 wherein the non-artifact related eye images are selected using appropriate candidate eye images that are rescaled and edited.
15. The method of claim 8 wherein the available subject eye type correction models comprise generic models selected and modified with predictions from metadata for size, color, shape, type, and emotional state.
16. The method of claim 1 wherein obtaining the user approval comprises accepting, editing, or selecting an alternative correction modality.
17. The method of claim 1 wherein the method is performed at a digital photo kiosk.