The optimal gradient matching method: application to X-SAR and Magellan stereo images

The authors propose a new method for improving terrain reconstruction from SAR stereo images. The method is based on preprocessing the original images by employing an optimal filter which suppresses speckle noise and enhances relief structures that are afterwards used by the matching process. In tests on X-SAR images of a rugged terrain in the Austrian Alps and Magellan images of Venus' surface, the preprocessing technique delivered better matching results than the conventional use of original SAR images.