Learning a Dilated Residual Network for SAR Image Despeckling
This addresses the problem of speckle noise in synthetic aperture radar images for remote sensing applications, representing an incremental improvement with a novel deep learning method.
The authors tackled SAR image despeckling by proposing a dilated residual network (SAR-DRN) that learns a non-linear mapping from noisy to clean images, achieving superior performance over state-of-the-art methods, particularly for strong speckle noise.
In this paper, to break the limit of the traditional linear models for synthetic aperture radar (SAR) image despeckling, we propose a novel deep learning approach by learning a non-linear end-to-end mapping between the noisy and clean SAR images with a dilated residual network (SAR-DRN). SAR-DRN is based on dilated convolutions, which can both enlarge the receptive field and maintain the filter size and layer depth with a lightweight structure. In addition, skip connections and residual learning strategy are added to the despeckling model to maintain the image details and reduce the vanishing gradient problem. Compared with the traditional despeckling methods, the proposed method shows superior performance over the state-of-the-art methods on both quantitative and visual assessments, especially for strong speckle noise.