Quaternion-based weighted nuclear norm minimization for color image restoration

作者：

Highlights：

• We develop the WNNM method to quaternion domain for color image restoration. With pure quaternion representation, more interconnected information among RGB channels can be preserved.

• To better recover the color image in the quaternion domain, we extend the 2-dimension blurring matrix to quaternion domain and design a quaternion blurring operator.

• We solve the proposed model by the ADMM method and the theoretical analysis of the uniqueness of the solution is presented.

• Experiments show that the proposed method outperforms some state-of-the-art methods, including the CNN-based method.

摘要

•We develop the WNNM method to quaternion domain for color image restoration. With pure quaternion representation, more interconnected information among RGB channels can be preserved.•To better recover the color image in the quaternion domain, we extend the 2-dimension blurring matrix to quaternion domain and design a quaternion blurring operator.•We solve the proposed model by the ADMM method and the theoretical analysis of the uniqueness of the solution is presented.•Experiments show that the proposed method outperforms some state-of-the-art methods, including the CNN-based method.

论文关键词：Quaternion representation,Color image restoration,Weighted nuclear norm,Variational method,Low-rank matrix analysis

论文评审过程：Received 13 January 2021, Revised 20 January 2022, Accepted 21 March 2022, Available online 23 March 2022, Version of Record 29 March 2022.

论文官网地址：https://doi.org/10.1016/j.patcog.2022.108665