Dynamic programming approach for context classification using the Markov random field

Haralick, Robert M.; Zhang, M.C.; Ehrich, Roger W.

doi:10.1109/icpr.1988.28466

Cited by 4 publications

(1 citation statement)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Markov random field modeling is a powerful approach to representation and analysis of images, A variety of such models have been proposed and explored for specific image processing problems ( [7], [8], 111, and [SI). In this paper we represent an image by the addition of a coarsely discretized array of pixels (pixel-classes) and an array of Gaussian random field to compensate for the coarse discretization.…”

Section: Introductionmentioning

confidence: 99%

Parameter estimation and applications of a class of Gaussian image models

Dattatreya¹,

Fang²

Proceedings of the IEEE Southwest Symposium on Image Analysis and Interpretation

View full text Add to dashboard Cite

This paper discusses variations of a model of images and develops algorithms for estimation of all the parameters from the raw image data. The model is suitable for some cases of (1) lossy image compression and realistic reconstruction, (2) texture synthesis and identification, (3) classification of remotely sensed data, and (4) analysis of medical images. Each pixel in the image is modeled as an element of a set of very few known intensity levels (henceforth called pixel-classes) plus an independent zero mean Gaussian random variable. Different statistical structures in the two dimensional lattice of pixel-classes lead to variations in the model. The image representation problem corresponds to estimation of the parameters of the discrete random field formed by the pixel classes, and the parameters of the additive Gaussian field. We discuss variations of the model and corresponding applications, and develop convergent estimators for all paramet ers .Markov random field modeling is a powerful approach to representation and analysis of images, A variety of such models have been proposed and explored for specific image processing problems ([7], [8], 111, and [SI). In this paper we represent an image by the addition of a coarsely discretized array of pixels (pixel-classes) and an array of Gaussian random field to compensate for the coarse discretization. The statistical dependency of the pixel-classes has a predominant influence on the content of the image. Different regions in the image can have different structures of pixel-class dependency. We will then split the image into tiles and assume uniform structure over each tile. The additive Gaussian signal may also have a dependent distribution over the two dimensional lattice. We assume the Gaussian signal distribution to be the same (stationary) over the entire image, even if the subregions of the image possess varying pixelclass dependency statistics.The remainder of this section discusses the variations of the model and the corresponding applications. Section I1 introduces the estimation problem and develops the estimator for the simplest of the model. Ba-18 0-8186-6250-6/94 $3.00 Q 1994 IEEE 1

show abstract

Section: Introductionmentioning

confidence: 99%