May. 22 2007 2:00PM

May. 22 2007 3:30PM

Identifying objects of interest in a background of noisy data and similar shapes is a problem faced in myriad situations. A challenge for the visual discrimination skills of human operators is a processing of images from the following areas:
• undersea mine hunting using side-scan sonar
• warhead and decoy recognition in anti-ballistic
missile radar-based detection systems
• improvised explosive device (IED) classification
in ground penetrating radar (GPR)
• tumor masses in computer aided tomography (CAT)
Feature Based Pattern Recognition (FBPR), in which a family of unitary image transforms associated to non-commutative groups is generated and applied to sensor images, is a novel approach for attacking such problems.