A Fractional Total Variational CNN Approach for SAR Image Despeckling

Abstract

Synthetic aperture radar (SAR) image despeckling is an essential problem in remote sensing technology, which has a strong influence on the performance of the following processing. We propose a new despeckled algorithm combining CNN and fractional-order total variation. Through constructing a CNN model and introducing the fractional-order total variation into loss function as the regularization term, the experimental results prove that our proposed method can avoid detail ambiguity and overly smooth caused by integral-order, and preserve rich texture and details information. Therefore, the high-quality despeckled images generated by our model will significantly improve the availability of SAR images.