Iteratively reweighted second-order regularization based multiplicative noise removal model

作者：Wang Xudong^*; Feng Xiangchu; Zhang Xuande

来源：Journal of Xidian University, 2014, 41(2): 130-136.

DOI：10.3969/j.issn.1001-2400.2014.02.022

• Hessian matrices
• Model and algorithms
• Multiplicative noise
• Multiplicative noise removals
• Numerical experiments
• Primal dual algorithms
• Regularization models
• Second order derivatives

In order to remove the multiplicative noise in an image, an iteratively reweighted second order derivatives (Frobenius norm of Hessian matrix) regularization model is proposed under the assumption that the multiplicative noise follows a Gamma distribution, which is an extension of the iteratively reweighted total variation model. A primal-dual algorithm for iteratively reweighted Frobenius norm of the Hessian matrix regularization model is designed. Numerical experiments show that the proposed model and algorithm can remove noise effectively while preserving details, restraining the staircase effect and avoiding edge blurring.

• 出版日期2014
• 单位宁夏大学; 西安电子科技大学; 广西大学