Gabor Jets for Clutter Rejection in Infrared Imagery

Report No. ARL-TR-3400
Authors: Mark Wellman and Nasser Nasrabadi
Date/Pages: December 2004; 19 pages
Abstract: The performance of an infrared imagery based (IR) automatic target detection algorithm suffers from several variables. The target size, shape, orientation and brightness play a role, as well as the cluttered backgrounds [1-5], which often exhibit greater intensity than the sensor noise or intensity of the targets to be detected. Once the detection process is finished and further decisions pending in the ATR system, one would like a simple technique to further reduce false alarms. The problem again, is one of recognition and past research has shown that joint spatial-frequency representations using Gabor schemes is beneficial in many automated, computer vision applications. We will present results using Gabor Jets together with a back propagation neural network for a detector post processing procedure, which could substantially reduce false alarms.
Distribution: Approved for public release
  Download Report ( 0.855 MBytes )
If you are visually impaired or need a physical copy of this report, please visit and contact DTIC.

Last Update / Reviewed: December 1, 2004