Nonlinear diffusion methods based on robust statistics for noise removal

作     者：贾迪野 黄凤岗 苏菡 

作者机构：School of Computer Science and TechnologyHarbin Engineering University 

出 版 物：《Journal of Harbin Institute of Technology(New Series)》 (哈尔滨工业大学学报（英文版）)

年 卷 期：2007年第14卷第3期

页      面：440-444页

核心收录：

学科分类：12[管理学] 083002[工学-环境工程] 1204[管理学-公共管理] 120402[管理学-社会医学与卫生事业管理(可授管理学、医学学位)] 0830[工学-环境科学与工程（可授工学、理学、农学学位）] 08[工学] 0837[工学-安全科学与工程] 

主　　题：Bayesian regularization M-estimation nonlinear diffusion bilateral filter 

摘      要：A novel smoothness term of Bayesian regularization framework based on M-estimation of robust statistics is proposed, and from this term a class of fourth-order nonlinear diffusion methods is proposed. These methods attempt to approximate an observed image with a piecewise linear image, which looks more natural than piecewise constant image used to approximate an observed image by P-M model. It is known that M-estimators and W-estimators are essentially equivalent and solve the same minimization problem. Then, we propose PL bilateral filter from equivalent W-estimator. This new model is designed for piecewise linear image filtering, which is more effective than normal bilateral filter.