zbMATH — the first resource for mathematics

Markov random field image models and their applications to computer vision. (English) Zbl 0665.68067
Proc. Int. Congr. Math., Berkeley/Calif. 1986, Vol. 2, 1496-1517 (1987).
[For the entire collection see Zbl 0657.00005.]
This paper accomplishes two tasks. First, it is a discussion of some basic principles concerning the application of Markov random field models to texture analysis, illustrated by computer experiments. Second, it contains original theoretical results in the parameter estimation problem, namely the estimation of parameters of a Markov random field from a single, large, sample. The used estimation method is proved to be consistent in the “large picture” limit, which is more appropriate than the usual “large sample size” limit.
Reviewer: R.Andonie

68T10 Pattern recognition, speech recognition
68U99 Computing methodologies and applications
Zbl 0657.00005