马尔可夫随机场在低信噪比图像恢复中的应用  

Low SNR Image Restoration Based on Markov Random Field(MRF)

作　　者：张鹏[1] 张桂林[1] 

机构地区：[1]华中科技大学图像识别与人工智能研究所,武汉430074

出　　处：《计算机与数字工程》2006年第1期41-43,共3页Computer & Digital Engineering

摘　　要：提出一种基于马尔可夫随机场的二值图像恢复算法。该算法在迭代计算中对ISING模型中耦合系数J动态修改,是一种求解最大后验概率(MAP)的随机松弛算法,该算法兼顾条件迭代(ICM)算法计算量少和模拟退火(SA)算法全局收敛的优点。利用该方法恢复被加性高斯噪声污染的低信噪比图像,取得良好的实验结果。This document provides a method for binary image restoration based on Markov Random Field （MRF）. During the iterative calculation, the author changes the J parameter dynamically, which means the strength of the interaction in ISING model. Then a new stochastic relaxation method which is to find the Maximum a Postefiofi （MAP） is provided . The method holds both low - computational costs in Iterated Conditional Mode （ICM） and global optimum in simulated annealing （SA）. Image restoration experiment is done with low SNR images, which is added with Gaussian noise. Experimental result is ideal.

关 键 词：图象复原 高斯噪声 MARKOV随机场 最大后验概率 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]