基于整体变分和变换域分析的复合去噪模型  

A Hybrid Denoising Model Basing On Total Variation and Analysis in Transform Domain

作　　者：仵冀颖[1] 阮秋琦[1] 

机构地区：[1]北京交通大学信息科学研究所,北京100044

出　　处：《信号处理》2008年第2期277-280,共4页Journal of Signal Processing

基　　金：国家973项目(2004CB318005);国家自然科学基金(60472033);教育部博士点基金(20030004023)

摘　　要：本文在泛函空间理论基础上提出了一种整体变分与小波阈值萎缩复合的图像去噪模型。复合模型在小空间规整化约束下实现整体变分优化去噪,保持图像边缘特征减弱值阶跃现象;复合模型处理图像中不存在Gibbs现象且可应用于大噪声图像恢复缺失信息。DCT变换是一种表征图像纹理特征的变换域处理方式,本文最后将DCT变换引入复合模型,得到保持纹理特征的复合去噪模型。理论和实验证明了本文提出的模型在图像去噪中的有效性。A novel hybrid denoising model which combines Total Variation and wavelet shrinkage basing on function space is proposed in this paper. The hybrid model denoises by minimizing Total Variation under the regularization constraint in small function space; it preserves edge feature and weakens staireasing. There is no Gibbs phenomenon in the denoised image and the novel model can restore lost information. DCT is used to characterize texture;it is introdueed into the hybrid model in this paper, and get another hybrid denoising model which preserves texture better. Both theoretical analysis and experiments have verified the validity of the new model.

关 键 词：图像去噪 整体变分模型(TV) 小波阈值萎缩 DCT变换 特征保持 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置] TV131.61[自动化与计算机技术—控制科学与工程]