Automatic SAR Image Registration Based on Neighborhood Entropy

Since the neighborhood entropy can indicate the information of the region and eliminate the effects of registration caused by the variety of image brightness. A synthetical approach of automatic registration for the synthetic aperture radar image is proposed. The SAR images are preprocessed, and then the neighborhood entropy of image is calculated. Finally, the image registration is accomplished by the use of the neighborhood entropy which is treated as image description. Furthermore, the optimization registration is proposed for enhancing the speed of the processing. The approach is validated according to several estimate results. Based on the MSTAR actual data, the experimental results show that the approach increases the precision, and it is robust and efficient.