Speckle2Void: Deep Self-Supervised SAR Despeckling with Blind-Spot Convolutional Neural Networks
This addresses the problem of speckle noise in SAR images for remote sensing and scene analysis, offering a practical solution by eliminating the need for hard-to-acquire clean or multi-temporal data, though it is incremental as it builds on blind-spot denoising networks.
The paper tackles SAR image despeckling by proposing a self-supervised deep learning method that trains on noisy SAR images without clean data, achieving performance close to supervised methods on synthetic data and superior results on real data in quantitative and visual assessments.
Information extraction from synthetic aperture radar (SAR) images is heavily impaired by speckle noise, hence despeckling is a crucial preliminary step in scene analysis algorithms. The recent success of deep learning envisions a new generation of despeckling techniques that could outperform classical model-based methods. However, current deep learning approaches to despeckling require supervision for training, whereas clean SAR images are impossible to obtain. In the literature, this issue is tackled by resorting to either synthetically speckled optical images, which exhibit different properties with respect to true SAR images, or multi-temporal SAR images, which are difficult to acquire or fuse accurately. In this paper, inspired by recent works on blind-spot denoising networks, we propose a self-supervised Bayesian despeckling method. The proposed method is trained employing only noisy SAR images and can therefore learn features of real SAR images rather than synthetic data. Experiments show that the performance of the proposed approach is very close to the supervised training approach on synthetic data and superior on real data in both quantitative and visual assessments.