Abstract: This paper focuses on the registration of optical and synthetic aperture radar (SAR) images. Due to the completely different imaging characteristics between optical and SAR images and the influence of speckle noise in the latter, the difficulty of registration between them is increased. In addition, the size of SAR image data increases with increasing resolution, which also imposes higher requirement on the real-time properties of the registration algorithm. To solve these problems, a stepwise refinement method for optical and SAR images registration is presented. Firstly, the region of interest extracted in SAR images is used for coarse registration between the images, which can reduce the algorithm pressure efficiently. After that the stable structural features in the images is extracted to construct an exact transformation between optical and SAR images. According to the experimental result of measured data, the effectiveness of the algorithm can be proved.

Key words: registration of optical and synthetic aperture radar images, region of interest, template matching, Hausdorff distance