Optimum morphological filtering to remove speckle noise from SAR images

Speckle together with usual additive noises cause severe degradation of Synthetic Aperature Radar (SAR) images. Spatial averaging is the commonly used technique for removing speckle noised However, this technique reduces image resolution appreciably and as a result the image is blurred. Morphological closings and openings offer a better way to reduce the speckle noise without blurring the image. In our earlier work we have shown² that the order in which these two morphological operations are performed is crucial in the restoration of SAR images. Although a close-open operation with two dimensional (2-D) structuring elements does not blur the image as the spatial averaging does, the fractal features in the image and the fine details of the region boundaries suffer significantly in this process. In past we used2 the min-of-closings and max-of-opening operators introduced by Serra³ and further studied by Maragos^{4, 5} to reduce this problem. We will call these operators as median operators^{4, 5} in the subsequent discussion. This technique preserves much more information in the SAR images than either spatial averaging or close-opening with 2-D structuring elements. This approach, however, has limitation of the number of directions of the linear probing elements thus resulting in a loss of several fractal features. Thin but long features which can not be probed by straight lines also suffer in this filtering process. In this paper we have introduced new operators to remove dark or bright spots which can not fit inside the boundary of a convex 2-D structuring element. Any region that can not fit inside the boundary is preserved. A multiscale filtering process is required to remove noise spots of different sizes. While using sampled images for processing at higher scales, a preprocessing is required before the sampling to retain important image features that may be lost in sampling. Finally the paper presents an algorithm that ensures that no distortion is introduced in the final image as a result of intermediate sampling and reconstruction steps. We have used this algorithm to filter the noise in SAR images obtained at different wavelengths. The present technique is remarkably more successful in restoring complex image details than either spatial averaging or morphological filtering using median operators.