Regularized Image Restoration Based on Edge Restoration and Artifacts Removing
基于边缘恢复和伪像消除的正则化图像复原

For many reasons, the artifacts can not be avoided in restored images such as Gibbs effect, grain noise and edge ring. Therefore, a new regularized approach for image restoration is proposed based on edge restoration and artifacts removing. Two new regularized constraint items are constructed in the approach with preserving classical smoothing regularized constraint item. Firstly, the degraded image is divided into edge region, texture region and flat region. Then, the increasing local variance of edge region and decreasing local variance of flat region after image restoration are used respectively to construct regularized constraint item as edge restoration constraint and artifacts removing constraint. Experimental results show the proposed approach is better than classical method in image restoration effect with two above-mentioned additional regularized constraint items.