A simple, efficient filter for multitemporal SAR images
It has been recently shown how multitemporal data may be used to improve the radiometric resolution of Synthetic Aperture Radar (SAR) images. Yet even the optimal speckle reduction schemes may result in blurred images in featured zones. Structure detection is a prior requirement but complicates further the numerical treatments. Alternatively, it is shown that satisfactory results can be obtained using a simple adaptation of Lee's sigma filter, as illustrated by comparison with theoretically based methods. The proposed methodology combines temporal filtering with a subwindowing approach and appears to be a good compromise that smoothes efficiently the speckle in relatively homogeneous areas while carefully preserving the sharpness of the edges. It may be of particular interest for those concerned with the variations in time of backscatter, especially in an agricultural environment.