Object classification method utilizing wavelet signatures of a monocular video image

• US 20060088219A1
• Filed: 10/26/2004
• Published: 04/27/2006
• Est. Priority Date: 10/26/2004
• Status: Abandoned Application

• Current Assignee: Stephen J. Kiselewich, Yan Zhang
• Original Assignee: Stephen J. Kiselewich, Yan Zhang
• Inventors: Zhang, Yan, Kiselewich, Stephen J.
• Application Number:
  US10/973,584
• Publication Number:
  US 20060088219A1
• Time in Patent Office: Days
• US Class Current: 382/224
• CPC Class Codes: B60R 21/01538 for image processing, e.g. cameras or sensor arrays
• CPC Class Codes: G06V 10/446 using Haar-like filters, e.g. using integral image techniques
• CPC Class Codes: G06V 10/52 Scale-space analysis, e.g. wavelet analysis multi-scale boundary representations G06V10/42
• CPC Class Codes: G06V 40/103 Static body considered as a whole, e.g. static pedestrian or occupant recognition

Abstract

A stream of images including an area occupied by at least one object are processed to extract wavelet coefficients, and the extracted coefficients are represented as wavelet signatures that are less susceptible to misclassification due to noise and extraneous object features. Representing the wavelet coefficients as wavelet signatures involves sorting the coefficients by magnitude, setting a coefficient threshold based on the distribution of coefficient magnitudes, truncating coefficients whose magnitude is less than the threshold, and quantizing the remaining coefficients.

Claims

1. A method of object classification, comprising the steps of:

receiving images of an area occupied by at least one object;

extracting wavelet coefficients from the images;

truncating and quantizing said wavelet coefficients to form wavelet signatures; and

classifying the object based on specified characteristics of said wavelet signatures.

2. The method of claim 1, including the steps of:

truncating wavelet coefficients having a magnitude that is less than a threshold; and

quantizing wavelet coefficients having a magnitude that is at least as great as said threshold.

3. The method of claim 1, including the steps of:

assigning a zero value to wavelet coefficients having a magnitude that is less than a threshold;

assigning a predefined positive value to wavelet coefficients that are positive and have a magnitude that is at least as great as said threshold; and

assigning a predefined negative value to wavelet coefficients that are negative and have a magnitude that is at least as great as said threshold.

4. The method of claim 2, including the step of:

determining said threshold based on said wavelet coefficients and their magnitudes.

5. The method of claim 4, including the steps of:

sorting said wavelet coefficients by magnitude;

selecting a group of highest magnitude wavelet coefficients;

identifying a lowest magnitude wavelet coefficient of said group; and

setting said threshold equal to a magnitude the identified wavelet coefficient.