由于视觉信息处理中存在大量的不确定性,概率图模型在计算机视觉领域有着广泛的应用,备受广大学者的关注。许多视觉问题都可以通过建立概率图模型进行求解,随着高效求解算法的提出和发展,马尔可夫随机场在解决计算机视觉领域的大规模数据问题中具有很大的优势。首先简要地介绍了概率图模型的概念,然后对马尔可夫随机场模型的定义、特性和推导求解进行了分析和讨论,在此基础上,以马尔可夫随机场在视觉信息的应用为线索,对目前基于马尔可夫随机场的计算机视觉信息处理的主要技术进行了概述和比较研究。
Probabilistic graphical models (PGM) is widely applied in visual information processing for the intrinsic uncertainty in visual information, and followed by a group of researchers recently. PGM offers a number of advantages for resolving variety problems in visual information processing, in which Markov Random Field (MRF) can be used to model pixel level information processing based on the development of high efficiency inference algorithms. In this paper, we shortly introduced concepts of PGM, and gave detailed analysis and discussion on the definition, features and inference of MRF followed by typical examples of its application in computer vision.