A novel method for estimation of the target rotation angle in SAR image

The rotation angle estimation is useful method to extract the feature of target for the application of SAR-ATR, which would improve the efficiency of the recognition. A novel algorithm based on corner point estimation (CP) is proposed for the target rotation angle estimation in the SAR image, which extracts the feature of corner point information to achieve high precision. In the presented CP method, the pre-processing of suppressing the speckle is needed, and it is necessary to separate target from backgrounds. In this paper, the segmentation by the pulse coupled neural network is used. Then the corner feature based on the Harris algorithm has been extracted. With the corner points, the target rotation has been estimation by the least squares method. The results of experiments by the presented method (CP) show the validity and accuracy of the algorithm.