Automatic Target Recognition with Bayesian Networks

Abstract

This paper presents a general framework for automatic target recognition using a feature-based Bayesian inference approach. The approach works on selected features. Using a Bayesian network as a representation of the joint probability distribution of the data, the network structure can be either constructed by expert knowledge or learned from the training data. The conditional probability distributions of features given the targets are estimated statistically. Features which carry the most discriminant information are selected automatically when the network structure is learned from the training data. The performance on two fully polarimetric synthetic aperture radar (SAR) and inverse SAR image data sets is illustrated.