A Practical Solution for SAR Despeckling With Adversarial Learning Generated Speckled-to-Speckled Images

In this letter, we aim to address a synthetic aperture radar (SAR) despeckling problem with the necessity of neither clean (speckle-free) SAR images nor independent speckled image pairs from same scene, and practical solution for (PSD) is proposed. First, an adversarial learning framework designed generate speckled-to-speckled (S2S) scene in situation where only single are available. Then, S2S employed train modified Nested-UNet model using Noise2Noise (N2N) strategy. Moreover, iterative version PSD method (PSDi) also presented. Experiments conducted on both real data demonstrate superiority proposed methods compared several state-of-the-art methods. The results show that our can reach good tradeoff between feature preservation speckle suppression.