一种改进的马尔可夫随机场图象恢复与分割模型

在现有的马尔可夫随机场图象恢复与分割模型中,图象场能量最低组态被看成是原始景物的一种最优估计。但在图象灰度值发生变化的边界上,能量最低组态不对应于原始景物,从而造成恢复(或分割)误差。本文对这类模型作了改进,利用改进的模型给出了一种引入边界信息的松弛算法,并给出了应用该算法对低信噪比图象进行恢复处理的计算机模拟结果。

A MODIFIED VERSION OF MARKOV RANDOM FIELD MODEL FOR IMAGE RESTORATION AND SEGMENTATION

The current Markov random field models for image restoration and segmentation are discussed. A configuration of the image field is regarded as an optimal estimate of the original scene when its energy is the lowest. However, the lowest energy configuration does not correspond to the scene on the edges, which results in errors of restoration or segmentation. Improvements of the model are made and a relaxation algorithm based on the improved model is presented using edge information obtained by a coarse-to-fine procedure. Some examples are also presented on the application of the algorithm to restoration of noisy images.