Postprocessing of Compressed Images via Sequential Denoising
This work addresses image quality degradation from compression artifacts for users of common compression techniques like JPEG, JPEG2000, and HEVC, representing an incremental improvement by adapting existing denoising methods to this specific problem.
The authors tackled compression-artifact reduction in images by proposing a postprocessing technique that frames it as an inverse problem using a Plug-and-Play Prior framework with ADMM, leading to sequential Gaussian denoising steps and linearization of compression processes, resulting in impressive gains in image quality for JPEG, JPEG2000, and HEVC compression methods.
In this work we propose a novel postprocessing technique for compression-artifact reduction. Our approach is based on posing this task as an inverse problem, with a regularization that leverages on existing state-of-the-art image denoising algorithms. We rely on the recently proposed Plug-and-Play Prior framework, suggesting the solution of general inverse problems via Alternating Direction Method of Multipliers (ADMM), leading to a sequence of Gaussian denoising steps. A key feature in our scheme is a linearization of the compression-decompression process, so as to get a formulation that can be optimized. In addition, we supply a thorough analysis of this linear approximation for several basic compression procedures. The proposed method is suitable for diverse compression techniques that rely on transform coding. Specifically, we demonstrate impressive gains in image quality for several leading compression methods - JPEG, JPEG2000, and HEVC.