SAR Target Recognition with the Fusion of LDA and ICA

In this paper, an approach for synthetic aperture radar (SAR) target recognition with fusion of linear discriminant analysis (LDA) and independent component analysis (ICA) features is presented. We first employ LDA and ICA to extract feature vectors from SAR images. The extracted LDA and ICA features are then imported to two support vector machine (SVM) classifiers respectively. Ranking based decision fusion algorithm is used to fuse the results of two SVM classifiers and the final classification decision is achieved. Finally, we apply the method for various ground vehicles in MSTAR database to evaluate the recognition performance. Experimental results show the higher target recognition performance compared with the methods using LDA or ICA feature.