An automatic target recognition (ATR) scheme in colored Gaussian noise based on parametric model and hybrid algorithm

Abstract

This paper firstly presents a concise and physically relevant parametric model for use in automatic target recognition (ATR), data compression and scattering studies. Then, the extraction algorithm of high-resolution target feature vectors for ATR in the presence of colored Gaussian noise is developed, which is not restricted by the low SNR and Rayleigh resolution limit. Therefore, an eight-stage ATR scheme based on the hybrid algorithm is proposed. The experimental results show its practicability in the application.