摘要
This paper aims to develop a fractional-order model and a primal-dual algorithm for image denoising, where a regularization parameter can be adjusted adaptively according to Morozov discrepancy principle at each iteration to ensure that the denoised image retains in a specific set. In the light of saddle-point theory, the convergence of our proposed algorithm is guaranteed. Simulations with comparisons are carried out to demonstrate the effectiveness of our proposed algorithm for image denoising.
- 出版日期2015-3-1
- 单位沈阳大学; 东北大学