SAR Clutter Characterization for ATR Performance Prediction

Period of Performance: 04/04/2000 - 01/04/2001

$1.61MM

Phase 2 SBIR

Recipient Firm

Alphatech, Inc.
6 New England Executive Park
Burlington, MA 01803
Principal Investigator

Research Topics

Abstract

The proposed effort will produce a software tool for prediciting the performance of SAR ATR algorithms in scenarios for which they have yet to be tested. The computations are based on a MAP estimation paradigm that is able to combine diverse sources of prior information such as DFAD, FFD, DTED, and other sensor modalities, with pre-computed performance results based on stock SAR image sets, to produce an initial estimate of performance. An adpatation mechanism is provided that allows this initial estimate to be updated using actual data from the scenario of interest as part of an on-the-fly adaptive performance characterization scheme. The output of the algorithm is a probability distribution for the expected ROC curve of the system.