Detecting and Tracking Small Moving Target by using Gabor/LTG Image Stabilization Method and Saliency Analysis

Period of Performance: 03/18/2005 - 12/18/2005

$100K

Phase 1 SBIR

Recipient Firm

Compusensor Technology Corp.
352-C Christopher Avenue
Gaithersburg, MD 20879
Principal Investigator

Abstract

Detecting and tracking small moving targets is a great challenge for aircraft. Passive Ground Moving Target Indicator (PGMTI) algorithms must carry out the simultaneous tasks of robust platform motion compensation, false alarm suppression, and accurate target tracking, all in real-time. CompuSensor propose using its Lie Transformation Group method of Gabor-type place token matching as the basis of image stabilization, using saliency analysis method for false alarm suppression, small moving target detection and tracking. 1) Compared with image feature correspondence method, the neurobiology inspired Gabor-type place token matching is more robust to small size perturbation factors in a dynamic scene, and through Lie transformation group method, capable of rapid and accurate motion compensation. 2) The saliency analysis based method provides a natural and generic method for false alarm suppression and small target detection. 3) A robust algorithm for tracking moving targets resulted from combination of the two methods. The proposed method is suitable for real-time operation.