Adaptive edge preserving regularized image restoration

Abstract

In this work, an adaptive iterative regularized image restoration scheme is proposed. For better approximation, a regularization function and an anisotropic regularization operator are suggested instead of constant regularization parameter and isotropic regularization operator. The parameters of the regularization function are chosen from the noise distribution in the image and the regularization operator is based on the edge directions. Experimental results on different images show the efficacy of the proposed approach under different blur and noise condition over its counterparts.