Blind image deblurring with Gaussian curvature of the image surface

作者：

Highlights：

• A novel regularization based on Gaussian curvature (GC) of the image surface for blind image deblurring is proposed.

• Minimizing the proposed regularization is able to preserve sharp edges and remove detrimental structures for blur kernel estimation.

• An effective optimization scheme based on Gaussian curvature filter (GCF) and half-quadratic splitting method is provided.

• Effectiveness of the proposed method is evaluated on synthetic datasets and real-world blurry images.

摘要

•A novel regularization based on Gaussian curvature (GC) of the image surface for blind image deblurring is proposed.•Minimizing the proposed regularization is able to preserve sharp edges and remove detrimental structures for blur kernel estimation.•An effective optimization scheme based on Gaussian curvature filter (GCF) and half-quadratic splitting method is provided.•Effectiveness of the proposed method is evaluated on synthetic datasets and real-world blurry images.

论文关键词：Blind image deblurring,Gaussian curvature regularization,Gaussian curvature filter

论文评审过程：Received 12 March 2021, Revised 3 September 2021, Accepted 1 October 2021, Available online 13 October 2021, Version of Record 21 October 2021.

论文官网地址：https://doi.org/10.1016/j.image.2021.116531