Generative and probability models in image processing and computer vision

Abstract

This paper gives an analysis of the role of generative models in image processing and computer vision. Oriented and unoriented graphical models (Bayesian and Markov networks) are considered, along with the possibilities of using them in image processing, in particular, to solve problems of noise filtering, segmentation, and stereo vision. Probability programming is considered as a method of specifying arbitrary generative models that possess substantially larger expressive force than graphical models. It is shown that the main limitation of probability programming is associated with the problem of the output efficiency on arbitrary generative models, for the solution of which it is necessary to develop methods of automatic specialization of general output procedures.