Automatic registration of SAR and optics image based on multi-features on suburban areas

Image registration, the process of estimating the optimal correspondence relation between two or more images taken at different time, from different viewpoint or by different sensors, is essential for a variety of applications in remote sensing, computer vision, pattern recognition and medicine, etc. It is a prerequisite for accomplishing high level tasks such as sensor fusion, surface reconstruction, change detection, and object recognition. In order to realize the automatic registration between Synthesize Aperture Radar(SAR) and Optical images, in this paper, an automatic registration algorithm based on different features is proposed. This method is proposed according to characteristic of image of remote sensing. Patches, edge and point features are used in our paper to registry. The evaluation of the image registration accuracy proves that the stratified registration algorithm based on multi-features proposed in this paper work well in automatic registration of SAR image and optical image with pixel accuracy.