Iterated multi-channel filtering of SAR images

Optimised methods for filtering sequences of SAR intensity image are known, but are expensive in computer time and memory. This paper shows how filtering can be carried out sequentially. Under the assumption that the current image is uncorrelated with the rest of the data sequence, all previously filtered images can be updated without redoing the whole calculation. The only new quantity needing to be calculated is the local mean in the latest image (using spatially adaptive methods). Although applicable to any data sequence, this approach is most naturally applied to multitemporal data, and has the major advantage that archived data can be used to provide a cheap platform on which to base filtering of later data. The algorithm is illustrated using ERS PRI data