SAR target recognition based on Gabor filter and sub-block statistical feature

A method for SAR target recognition based on low-frequency sub-band image, Gabor filter bank and sub-block statistical feature is presented in the paper. The sub-band image extracted from the pre-processed SAR image is filtered by Gabor filter bank on different directions and scales. The each filtered sub-band image is divided into different sub-blocks and the statistical features derived from every sub-block are regarded as the target recognition feature, which can be used to recognize the targets with SVM. The proposed method is tested and validated on MSTAR dataset for 3-type SAR target recognition.