Sequential discrimination with multi-features to remove false ROIs in SAR ATR

Because the prescreening usually adopts an anomaly detection approach to highlight the vehicle-targets in SAR image, many false regions of interest (ROIs) are produced to reduce hardly the efficiency of the ATR. A new method based on sequential discrimination with multi-features is proposed to remove these false ROIs in this paper. By quantitatively analyzing the redundancy, robustness and separability of the candidates, the optimal features are selected to form an orderly vector, which is compared with the vector of threshold to finish a sequential discrimination. The performance of the above algorithm is validated by the X band MSTAR data, and is compared with the quadratic distance discriminating (QDD) method.