Fractal/Wavelet-Based Region-of-Interest Classification

Period of Performance: 07/11/1997 - 04/07/1998

$98.9K

Phase 1 STTR

Recipient Firm

Information System Laboratories
6767 Old Madison Pike Suite 180
Huntsville, AL 35806
Principal Investigator
Firm POC

Research Institution

University of Alabama, Huntsville
301 Sparkman Drive
Huntsville, AL 35899
Institution POC

Abstract

This Phase I STTR effort will establish the feasibility of employing wavelet transform techniques, fractal theory techniques, and standard multivariant filter techniques to determine target regions of interest (ROI) in tactical imagery. This effort will be accomplished through the application of wavelet transform techniques for edge and feature detection capabilities as well as generating reduced matched filters. The effort will also be accomplished by the generation of fractal filters for clutter characterization and the use of fractal dimensions for distinguishing man-made from natural objects for subsequent target and clutter ROI determination. The combinations of generalized filters with wavelet and fractal filtering methods will be explored to improve on the characterizing ability of single wavelet, fractal, or standard matched filtered. Information Systems Laboratories (ISL) proposes to define algorithms that successfully incorporate these concepts in standard and non-standard tactical image processing and classification techniques.