Block-based Collaborative Filtering (BCF) for SAR image denoising

This paper introduces a novel technique for the suppression noise and sidelobes in synthetic aperture radar (SAR) imagery. In the proposed Block-based Collaborative Filtering (BCF) technique, multiple SAR images of the same scene are formed using randomly reduced apertures. The resulting SAR images are then blocked-based processed in a collaborative fashion. More specifically, collocated blocks of these images are vectorized to form columns of a collective matrix. This collective matrix is then principally decomposed into a sum of a signal part and a noise part. The signal part is employed to reconstruct a noise-reduced SAR image. The key principle behind this technique is that noise pattern is more varying, whereas backscatter signals from physical objects are more stable and consistent across many instances of randomly reduced aperture SAR images, and thus, can be extracted using principal component analysis. This paper demonstrates the results of the proposed technique using both simulation and real radar data from the U.S. Army Research Laboratory´s (ARL) low-frequency SAR radar in forward-looking mode.