Abstract: For the SAR image with multiplicative noise, some edge detectors in optical image effectively does not apply in the SAR image. An improved edge detection algorithm for SAR image is presented. Three steps are involved. Firstly, a variety of evaluation index are used, which concluded that Lee filter can effectively remove the noise, and retain the gradient information to the maximum extent, suitable for SAR edge detection. And then put forward an improved Ratio detector, which is more conducive to image normalization. Finally, propose a new edge detector which combined the improved Ratio detector and Canny detector. The results indicate that new algorithm has a good performance in the edge detection and positioning accuracy, improves the edge continuity.

Key words: SAR image, edge detection, Lee filter, improved Ratio detector, Canny detector