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.