SAR Target Recognition with Data Fusion

This paper presents an approach for synthetic aperture radar (SAR) target recognition with data fusion. The data of multi-aspect images of a target are fused by principal component analysis (PCA) or discrete wavelet transform (DWT) after preprocessing. Wavelet domain PCA is used to extract feature vectors from the fused data. Support vector machine (SVM) is applied to classify the extracted feature vectors. Experiments are implemented with three military targets in MSTAR database for analyzing the effects on recognition rate of targets caused by different number of images and aspect intervals in different fusion algorithms. The experimental results demonstrate the higher recognition rate of the proposed method than that of the method without data fusion. Therefore, the proposed method can be applied in SAR image target recognition effectively and advance recognition rate of targets significantly.