A Study on JPEG Artifact Removal by Four-Directional Difference-based ℓ0,1 norm Regularization
Iori Kobayashi, Ryo Matsuoka · 2022 IEEE 11th Global Conference on Consumer Electronics (GCCE) · 2022
This paper proposes an optimization method for JPEG artifact removal on clip-art images. We define four-directional difference-based $\ell_{0,1} -$norm regularization to robustly remove JPEG artifacts. Moreover, we formulate an optimization problem with the proposed regularization and a quantization constraint for the block-wise discrete cosine transform (DCT) coefficients and solve it by using the alternating direction method of multipliers (ADMM).