Automated method and apparatus for robust image object recognition and/or classification using multiple temporal views

• US 20080112593A1
• Filed: 10/30/2007
• Published: 05/15/2008
• Est. Priority Date: 11/03/2006
• Status: Abandoned Application

• Current Assignee: Keystream Corp.
• Original Assignee: Keystream Corp.
• Inventors: Cullen, Schuyler A., Ratner, Edward R.
• Application Number:
  US11/981,244
• Publication Number:
  US 20080112593A1
• Time in Patent Office: Days
• US Class Current: 382/103
• CPC Class Codes: G06F 18/254 of classification results, e.g. of results related to same input data
• CPC Class Codes: G06V 10/426 Graphical representations
• CPC Class Codes: G06V 10/62 relating to a temporal dimension, e.g. time-based feature extraction; Pattern tracking

Abstract

An automated method for classifying an object in a sequence of video frames. The object is tracked in multiple frames of the sequence of video frame, and feature descriptors are determined for the object for each of the multiple frames. Multiple classification scores are computed by matching said feature descriptors for the object for each of the multiple frames with feature descriptors for a candidate class in a classification database. Said multiple classification scores are aggregated to generate an estimated probability that the object is a member of the candidate class. Other embodiments, aspects and features are also disclosed.

Claims

1. An automated method for classifying an object in a sequence of video frames, the method comprising:

tracking the object in multiple frames of the sequence of video frames;

determining feature descriptors for the object for each of the multiple frames;

computing multiple classification scores by matching said feature descriptors for the object for each of the multiple frames with feature descriptors for a candidate class in a classification database; and

aggregating said multiple classification scores to generate an estimated probability that the object is a member of the candidate class.

2. The method of claim 1, wherein said aggregating comprises determining a highest classification score among the multiple classification scores.

3. The method of claim 1, wherein said aggregating comprises determining an average classification score from the multiple classification scores.

4. The method of claim 1, wherein said aggregating comprises determining a median classification score from the multiple classification scores.

5. The method of claim 1, wherein said aggregating comprises using a Bayesian inference to determine a combined probability.

6. The method of claim 1, wherein the object is tracked by partitioning of a temporal graph.

7. The method of claim 1, wherein the feature descriptors for the object are determined by applying scale invariant feature transforms.

8. The method of claim 1, wherein the classification scores are computed using a support vector machine engine.

9. The method of claim 1, wherein the object is tracked by partitioning of a temporal graph.

10. A computer apparatus configured to classify an object in a sequence of video frames, the apparatus comprising:

a processor for executing computer-readable program code;

memory for storing in an accessible manner computer-readable data;

computer-readable program code configured to track the object in multiple frames of the sequence of video frames;

computer-readable program code configured to determine feature descriptors for the object for each of the multiple frames;

computer-readable program code configured to calculate multiple classification scores by matching said feature descriptors for the object for each of the multiple frames with feature descriptors for a candidate class in a classification database; and

computer-readable program code configured to aggregate said multiple classification scores to generate an estimated probability that the object is a member of the candidate class.

11. The apparatus of claim 10, wherein said multiple classification scores are aggregated by determining a highest classification score among the multiple classification scores.

12. The apparatus of claim 10, wherein said multiple classification scores are aggregated by determining an average classification score from the multiple classification scores.

13. The apparatus of claim 10, wherein said multiple classification scores are aggregated by determining a median classification score from the multiple classification scores.

14. The apparatus of claim 10, wherein said multiple classification scores are aggregated by using a Bayesian inference to determine a combined probability.

15. The apparatus of claim 10, wherein the object is tracked by partitioning of a temporal graph.

16. The apparatus of claim 10, wherein the feature descriptors for the object are determined by applying scale invariant feature transforms.

17. The apparatus of claim 10, wherein the classification scores are computed using a support vector machine engine.

18. The apparatus of claim 10, wherein the object is tracked by partitioning of a temporal graph.