Blind image deconvolution via an adaptive weighted TV regularization

Article type: Research Article

Authors: Xu, Chenguang^b | Zhang, Chao^b | Ma, Mingxi^a | Zhang, Jun^{a; b; *}

Affiliations: [a] College of Science, Nanchang Institute of Technology, Nanchang, Jiangxi, China | [b] Jiangxi Province Key Laboratory of Water Information Cooperative Sensing and Intelligent Processing, Nanchang Institute of Technology, Nanchang, Jiangxi, China

Correspondence: [*] Corresponding author. Jun Zhang. E-mail: junzhang0805@126.com.

Abstract:  Blind image deconvolution has attracted growing attention in image processing and computer vision. The total variation (TV) regularization can effectively preserve image edges. However, due to lack of self-adaptability, it does not perform very well on restoring images with complex structures. In this paper, we propose a new blind image deconvolution model using an adaptive weighted TV regularization. This model can better handle local features of image. Numerically, we design an effective alternating direction method of multipliers (ADMM) to solve this non-smooth model. Experimental results illustrate the superiority of the proposed method compared with other related blind deconvolution methods.

Keywords: Blind deconvolution, Total variation regularization, Adaptive weighted matrix, ADMM

DOI: 10.3233/JIFS-223828

Journal: Journal of Intelligent & Fuzzy Systems, vol. 44, no. 4, pp. 6497-6511, 2023