A Two-Stage Method for Joint Denoising and Compression of SAR Images

Synthetic Aperture Radar (SAR) images are usually corrupted by speckle noise, which introduces additional challenges to lossy image compression algorithms. Traditional optical image compression techniques fail to take this scenario into account, resulting in wastage of additional bits for storing undesired speckles and reconstructed images with unpleasant noise. Inspired by these observations, we optimize image compression algorithms to incorporate denoising capabilities, enabling joint denoising and compression of SAR images. Specifically, a novel two-stage training method is designed to transform the raw speckled images into noise-free bitstreams. We further devise a multi-objective optimization strategy to guide the model learning for different objectives in each stage. Experimental results demonstrate that the proposed method outperforms existing methods in terms of rate-distortion performance when evaluated on both synthetic and real-world SAR datasets.