Showing papers on "Markov random field published in 1985"

Classification of textures using Gaussian Markov random fields

Abstract: The problem of texture classification arises in several disciplines such as remote sensing, computer vision, and image analysis. In this paper we present two feature extraction methods for the classification of textures using two-dimensional (2-D) Markov random field (MRF) models. It is assumed that the given M × M texture is generated by a Gaussian MRF model. In the first method, the least square (LS) estimates of model parameters are used as features. In the second method, using the notion of sufficient statistics, it is shown that the sample correlations over a symmetric window including the origin are optimal features for classification. Simple minimum distance classifiers using these two feature sets yield good classification accuracies for a seven class problem.

Optimal Bayesian Estimators For Image Segmentation and Surface Reconstruction

Abstract: A very fruitful approach to the solution of image segmentation and surface reconstruction tasks is their formulation as estimation problems via the use of Markov random field models and Bayes theory. However, the Maximuma Posteriori (MAP) estimate, which is the one most frequently used, is suboptimal in these cases. We show that for segmentation problems the optimal Bayesian estimator is the maximizer of the posterior marginals, while for reconstruction tasks, the threshold posterior mean has the best possible performance. We present efficient distributed algorithms for approximating these estimates in the general case. Based on these results, we develop a maximum likelihood that leads to a parameter-free distributed algorithm for restoring piecewise constant images. To illustrate these ideas, the reconstruction of binary patterns is discussed in detail.

Markov Processes on Infinite Dimensional Spaces, Markov Fields and Markov Cosurfaces

Abstract: We review work on Dirichlet forms and symmetric Markov processes on infinite dimensional spaces. Especially we consider the connections with the construction of homogeneous generalized Markov random fields. We also discuss a non commutative extension to the case where the state space is a group. The extension involves a stochastic calculus for group valued mappings defined on hypersurfaces of codimension 1.

Adaptive hierarchical algorithm for accurate image segmentation

Abstract: A conceptually new algorithm is presented for segmenting textured images into regions in each of which data is modelled as one of C 2-D Markov Random Field (MRF). The algorithm is designed to operate in real time when implemented on new parallel architecture. Gaussian MRF is used to model textures in visible light images of outdoor and indoor scenes. Image segmentation is realized as a maximum likelihood estimation. To simplify the segmentation algorithm, the image is partitioned into disjoint square windows in each of which there will be one or atmost two different texture regions. In any given window the segmentation algorithm is hierarchical and uses a pyramid-like structure. This paper is an extension to material introduced in [1,2] and concentrates on exploring the segmentation accuracy of the algorithm and addressing more fully the question of how the algorithm can operate in adaptive modes when the parameters of the texture field are partially or totally unknown.

Segmentation of multilevel images using Gibbs distribution

Abstract: An algorithm for multilevel image segmentation is presented. The main feature is that it is based on repetition of a binary segmentation algorithm, in a fashion similar to the binary expansion of an integer number. The binary segmentation considered assumes a Markov Random Field model for the original scene, and an additive i.i.d, noise signal. Simulation rasults are presented, and compared with other filtering algorithms, such as median filtering.

Multivariate Extensions of Maximum Entropy Methods

Abstract: Since almost all of the papers presented at these workshops deal with the single-channel, or one-dimensional, maximum entropy spectral estimation problem, it seemed appropriate that someone should review the interesting work being reported on the analogous multivariate problem. This survey is intended to acquaint the reader with this area of research.

The Implications of Boltzmann-Type Machines for SAR (Synthetic Aperture Radar) Data Processing: A Preliminary Survey,

Abstract: : This document proposes that Markov random field models (MRFs) be used as a framework within which to construct models of synthetic aperture radar (SAR) images. Its author clarifies the relationship between this class of models and the Boltzmann machine (BM) of artificial intelligence. He then generalizes the BM training procedure and use it to train MRF models. Using this techniques he investigate the ability of a simple MRF texture model to learn a texture by maximizing a relative entropy objective function. It is found that the marriage of MRF models with the BM training procedure is fruitful. (Author)

Modelling of interplant competition

Abstract: A pure plant community with a simple structure of the arrangement of individuals is considered. The stochastic model represented by a Markov random field (MRF), describes a productivity property of the plant community. The parameters of the model may be interpreted as a strength of the competition between plants. The problem of the interplant competition modelling is to choose the probability structure and parameters of the model and to simulate a Markov random field. The synthesized image patterns may be applied to a study of real data.