Dynamic features are important aspects of the ocean. However the dynamic information is lost in most conventional Synthetic Aperture Radar (SAR) image processing methods, because they treat the image as an instantaneous state of the observed area. In fact, we can obtain dynamic features of the ocean from sequential sub-aperture images, because we know that the different parts of the azimuthal aperture correspond to different imaging instances. A key step for retrieving the dynamic features from sequential images is image-matching. However, the heavy noise characteristic of sub-aperture SAR images renders the traditional image-matching methods ineffective. In this paper we propose an image matching method based on improved phase correlation to
deal with the heavy noise problem of SAR sub-aperture images. Experimental results show that the improved image-matching method presents an accuracy of 0.15 pixel and noise robustness. The analysis indicates that the improved algorithm is competent for obtaining dynamic information from the medium resolution airborne SAR images or high resolution spaceborne SAR images with 0.15-0.3 m/s estimation precision under most SNR conditions. The improved algorithm was used on an airborne SAR data to retrieve the movement velocity. The retrieved velocity ranged from 0.05-0.5 m/s, which seems to be reasonable value for the ocean current velocity.