ABSTRACT
This chapter provides the reader with some of the recent scientific advances in automatic target recognition (ATR) from the perspectives of dynamic modelling, algorithmic advances and sensor modelling. Classical ATR approaches detect possible targets using various image processing procedures and then extract features to be fed into a pattern classifier for target recognition. The chapter presents a brief overview of neural networks, discussing the neural network structure and some associated mathematical models for some of the basic element of a neural network. It discusses artificial intelligence for an ATR system from the perspective of designing and implementing neural networks. An efficient target tracker is necessary for the success of an ATR system. Bayesian statistics is a well-evolved and elegant way to tackle the ATR problem mathematically instead of depending on training data and heuristics that are inherently not robust to variations in the scene.
