Adaptive texture-preserving filtering of multitemporal ERS-1 SAR images

A novel approach is proposed to adaptive filtering of multi-temporal images. A transformation aimed at decorrelating data is defined on the set of images taken at different times. Adaptive filtering is performed on the new set of data which are then transformed back. Results on ERS-1 multi-temporal SAR images show effectiveness in reducing speckle, while the intrinsic texture, edges and point targets are better preserved than by filtering each observation separately.