<title>Performance modeling for automatic target recognition systems</title>

Abstract

This paper explores using linear regression and Artificial Neural Networks to model the performance of an ATh algorithm based on a given set of data. Here, a probability of detection response surface as a function of relevant parameters (depression angle of a tank, age of breast cancer patient, etc.) is simulated. It is then shown that this surface can be approximated using either linear regression or an artificial neural network with good results. These regression surfaces can provide valuable information to the ATh developer/customer in terms of trying to predict ATR performance in untested areas. The application of this ATR performance modeling methodology becomes clear when we consider applying it to a common problem, such as air-to-ground target detection, where the changing parameters of the target can give a good set of data points from which to build the response curve.

Predicting an upper bound on SAR ATR performance

“…The estimated performance metrics included the probabilities of detection and false alarms, in addition to the confusion matrix for four classes (three targets and clutter). Catlin et al [8] presented a method for estimating the function relating the probability of correct identification to aspect and depression angles using SAR data. Each point on the function was determined experimentally by passing a number of real images to an ATR system.…”