An improved PCA based features for SAR ATR

In this paper, a SAR ATR method based on FD-PCA (Fourier Domain Principal Component Analysis) features is proposed. By using the Circular Sampling and FFT (Fast Fourier Transformation) technology, the improved method is insensitive to the azimuth rotation of target. MSTAR public dataset is used to compare the performance between FD-PCA and PCA method. By the comparison of the experimental results, FD-PCA features are more robust than the PCA features. The FD-PCA features extract method is proposed to use in the feature based SAR ATR system instead of PCA method.