Deep Learning Methods For Synthetic Aperture Radar Image Despeckling: An Overview Of Trends And Perspectives

Synthetic aperture radar (SAR) images are affected by a spatially correlated and signal-dependent noise called speckle, which is very severe may hinder image exploitation. Despeckling an important task that aims to remove such so as improve the accuracy of all downstream processing tasks. The first despeckling methods date back 1970s, several model-based algorithms have been developed in years since. field has received growing attention, sparked availability powerful deep learning models yielded excellent performance for inverse problems processing. This article surveys literature on applied SAR despeckling, covering both supervised more recent self-supervised approaches. We provide critical analysis existing methods, with objective recognizing most promising research lines; identify factors limited success models; propose ways forward attempt fully exploit potential despeckling.