To deal with the conflict between speckle removal and edge preservation in the SAR despeckling operations, we propose an adaptive despeckling filter that can preserve edges based on noise identification and edge detection oriented to SAR images. In the filter, the template’s size and shape are adaptively changed according to the noise characteristics, and the edge position and direction in homogeneous and edge regions.
This way, noise in the SAR image is reduced, while edges are sharpened, and texture and fine structures preserved. RADARSAT-2 SAR HH images covering the North Arctic Pole obtained on August 25, 2010 are used in the experiment. The results show validation and practicability of the method.