Method of classifying red-eye objects using feature extraction and classifiers
View Patent ↗Disclosed are methods, devices, and computer program products for red-eye detection in an image. In one example embodiment, a method for detecting red-eye objects in an image includes several acts. First, a set of candidate red-eye objects identified in the image is received. Then, features are extracted from the candidate red-eye objects and, with a plurality of classifiers, a false red-eye object is eliminated from the set of candidate red-eye objects based on the extracted features. First and second ones of the plurality of classifiers are optimized for classifying objects in a first range of sizes using first and second ones of the extracted features, respectively. Furthermore, third and fourth ones of the plurality of classifiers are also optimized for classifying objects using the first and second ones of the extracted features, respectively, but for objects in a second range of sizes.
1. A method for red-eye detection in an image, the method comprising the acts of:
receiving a set of candidate red-eye objects identified in the image;
extracting features from the candidate red-eye objects; and
eliminating a false red-eye object from the set of candidate red-eye objects with a plurality of classifiers based on the extracted features,
wherein first and second ones of the plurality of classifiers are optimized for classifying objects in a first range of sizes using first and second ones of the extracted features, respectively, and
wherein third and fourth ones of the plurality of classifiers are optimized for classifying objects in a second range of sizes using the first and second ones of the extracted features, respectively.
2. The method of claim 1 , wherein at least one of the first through fourth classifiers is a neural network classifier.
3. The method of claim 1 , wherein at least one of the plurality of classifiers is a simple classifier.
4. The method of claim 1 , wherein the extracted features include features corresponding to a plurality of regions associated with each of the candidate red-eye objects.
5. The method of claim 4 , wherein the extracted features include mean and standard deviation values of features in each region.
6. A system for red-eye detection in an image, the system comprising:
an object size discriminator configured to receive a group of candidate red-eye objects and to generate therefrom a first set of objects having a first range of sizes and a second set of objects having a second range of sizes;
a first plurality of feature extractors and corresponding trained classifiers configured to eliminate false red-eye objects from the first set of objects;
a second plurality of feature extractors and corresponding trained classifiers configured to eliminate false red-eye objects from the second set of objects;
a simple classifier configured to eliminate false red-eye objects from the objects remaining in the first and second sets of objects after elimination by the trained classifiers;
a filter configured to eliminate false red-eye objects from the objects remaining after elimination of objects by the simple classifier.
7. The system of claim 6 ,
wherein the first plurality of feature extractors and trained classifiers are arranged in a cascaded configuration,
wherein the second plurality of feature extractors and trained classifiers are arranged in a cascaded configuration, and
wherein the first and second cascaded pluralities of feature extractors and trained classifiers are arranged in a parallel configuration with respect to each other.
8. One or more non-transitory computer-readable media having computer-readable instructions thereon which, when executed, implement a method for processing an image to detect red-eye effects, the method comprising the acts of:
receiving a set of candidate red-eye objects identified in the image;
extracting features from the candidate red-eye objects; and
eliminating a false red-eye object from the set of candidate red-eye objects with a plurality of classifiers based on the extracted features,
wherein first and second ones of the plurality of classifiers are trained for classifying objects in a first range of sizes using first and second ones of the extracted features, respectively, and
wherein third and fourth ones of the plurality of classifiers are trained for classifying objects in a second range of sizes using the first and second ones of the extracted features, respectively.
9. An image processing device comprising:
a processor; and
the one or more non-transitory computer-readable media as recited in claim 8 , the computer-readable instructions thereon being configured to be executed by the processor.
10. The image processing device as recited in claim 9 , wherein the image processing device comprises a printer, a scanner, a digital camera, a digital camcorder, a portable photo viewer, or some combination thereof.