Object recognition system

• US 20040013286A1
• Filed: 07/22/2002
• Published: 01/22/2004
• Est. Priority Date: 07/22/2002
• Status: Active Grant

• Current Assignee: Mitsubishi Electric Research Laboratories, Inc. (Mitsubishi Electric Corporation)
• Original Assignee: Mitsubishi Electric Research Laboratories, Inc. (Mitsubishi Electric Corporation)
• Inventors: Jones, Michael J., Viola, Paul A.
• Granted Patent: US 7,031,499 B2
• Time in Patent Office: Days
• US Class Current: 382/118
• CPC Class Codes: G06V 40/168 Feature extraction; Face representation

Abstract

A method recognizes an object in an image. Gallery images include identified objects, and a probe image includes an unidentified object. The probe image is paired with each gallery image to generate pairs of images. First filters are applied to the probe image of each pair to obtain a first feature value for each application of each filter. Second filters are applied to the each gallery image of each pair to obtain a second feature value for each application of each filter. The feature values are summed for each application of each filter, and for each application, a score is set to an acceptance weight if the sum is greater than a predetermined threshold, and otherwise setting the score to a rejection weight. The scores are summed for all of the application, and the probe image is identified with any gallery image when the score is greater than zero.

Claims

1. A method for recognizing an object in an image, comprising:

providing a plurality of gallery images, each gallery image including an identified object;

pairing each gallery image with a probe image of an unidentified object to generate a plurality of pairs of images;

applying a first set of filters to the probe image of each pair to obtain a first feature value for each application of each filter, and applying a second set of filters to the each gallery image of each pair to obtain a second feature value for each application of each filter;

summing, for each application of each filter, the first and second feature values;

setting, for each application, a score to an acceptance weight if the sum is greater than a predetermined threshold, and otherwise setting the score to a rejection weight;

summing the scores for all of the application; and

identifying the probe image with any gallery image when the score is greater than zero.

2. The method of claim 1 wherein the object is a face.

3. The method of claim 1 further comprising:

normalizing the images with respect to size and orientation.

4. The method of claim 1 wherein different views are used for the identified object and the unidentified object.

5. The method of claim 1 wherein the filters are template filters.

6. The method of claim 1 wherein the filters are non-linear functions.

7. The method of claim 1 wherein the filters are local functions applied to a portion of the images.

8. The method of claim 1 wherein multiple sets of gallery images are provided, each set using a different view of the objects.

9. A system for recognizing an object in an image, comprising:

a plurality of gallery images, each gallery image including an identified object;

a probe image of an unidentified object, the probe image to be paired with each gallery image to generate a plurality of pairs of images;

means for applying a first set of filters to the probe image of each pair to obtain a first feature value for each application of each filter;

means for applying a second set of filters to the each gallery image of each pair to obtain a second feature value for each application of each filter;

means for summing, for each application of each filter, the first and second feature values;

means for setting, for each application, a score to an acceptance weight if the sum is greater than a predetermined threshold, and otherwise setting the score to a rejection weight;

means for summing the scores for all of the application; and

means for identifying the probe image with any gallery image when the score is greater than zero.