SAR image target recognition based on NMF feature extraction and Bayesian decision fusion

In this paper, a new approach of synthetic aperture radar (SAR) image target recognition based on non-negative matrix factorization (NMF) feature extraction and Bayesian decision fusion is presented for recognizing ground vehicles in MSTAR database. First, feature vectors are extracted from image chips by NMF algorithm. Support vector machine (SVM) is used to classify the feature vectors. After multiple views of the same vehicle collected at different aspects are classified by SVM, the outputs are fused by Bayesian decision fusion algorithm and then the final classification decision is generated. We evaluate NMF algorithm and the Bayesian decision fusion approach. Experimental results indicate that there are significant target recognition performance benefits in the probability of correct classification when NMF algorithm is applied and three or more views are used for Bayesian decision fusion.