基于加权核范数最小化的泊松噪声图像恢复  

Poisson Image Denoising Based on Weighted Nuclear Norm Minimization

作　　者：郑子君[1] 吴恒[1] 牛善洲[1] 黄进红[1] 

机构地区：[1]赣南师范大学数学与计算机科学学院,江西赣州341000

出　　处：《赣南师范大学学报》2017年第6期99-103,共5页Journal of Gannan Normal University

基　　金：校级研究生创新基金项目(YCX15B002)

摘　　要：本文提出了一种基于加权核范数最小化的泊松噪声图像恢复方法,通过Anscombe变换把泊松型数据转化为Gaussian型噪声数据.然后,用加权核范数最小化方法对变化后的Gaussian型噪声数据进行去噪.实验结果表明,我们的方法是有效的.In this paper, a Poisson image denoising method based on weighted nuclear norm minimization is proposed, and the Poisson-type data is transformed into Gaussian data by Anscombe transform. In this paper, the nonlocal self-similarity of images is considered, and the Gaussian noise is denoising by using weighted nuclear norm minimization. Numerical results are reported show that the proposed method is effective.

关 键 词：Poission噪声 Anscombe变换 Gaussian噪声 加权核范数 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]