Color image inpainting based on low-rank quaternion matrix factorization

In this paper, a color image inpainting problem is modeled as low rank quaternion matrix optimization problem. A has smooth property inherently and all imaginary parts of the corresponding pure are nonnegative. Based on these characteristics, we introduce novel regularized model then propose factorization algorithm with convergence analysis. Numerical experiments demonstrate efficiency our proposed algorithm.