一种新的基于大尺寸信息的MRF先验模型  

A Novel MRF Prior Model Based on Large-Scale Image Knowledge

作　　者：占杰[1] 陈阳[1] 陈武凡[1] 

机构地区：[1]南方医科大学生物医学工程学院,广东广州510515

出　　处：《计算机工程与科学》2009年第1期53-57,共5页Computer Engineering & Science

基　　金：国家973计划资助项目(2003CB716102);国家自然科学基金资助项目(30730036)

摘　　要：基于马尔可夫随机场理论,Bayesian重建被认为是一种解决图像复原和重建中的病态问题的有效方法。通常,大部分先验模型中的信息都来自小邻域内像素灰度值的简单加权,因此仅能提供给正则化有限的信息。在研究大尺寸信息的过程中,本文提出一种新的非局部先验。在发射断层成像的相关实验表明,该MRF非邻域先验能比传统先验提供更为有效的正则化处理。Based on the Markov Random Fields （MRF） theory, Bayesian approaches have been accepted as effective solutions to overcoming the ill-posed problems of image restoration and reconstruction. Conventionally, the knowledge in most of the prior models comes from simply weighted differences between the pixel intensities within a small local neighborhood, so it can only provide limited prior information for regularization. Exploring the ways of incorporating more large-scale image knowledge into the MRF prior model, a novel nonlocal prior is put forward in this paper. Relevant experiments in the application of the emission tomography prove that the proposed MRF nonloeal prior is capable of imposing more effective regularization on the original reconstructions.

关 键 词：马尔可夫随机场(MRF) Bayesian重建 发射断层成像 非局部先验 

分 类 号：Q819[生物学—生物工程]