zbMATH — the first resource for mathematics

Image analysis with partially ordered Markov models. (English) Zbl 1042.62611
Summary: Statistical approaches to image analysis, such as image restoration, segmentation, object classification, and reconstruction often require specification of a distributional model for the variability of the pixel intensities around the true image and a prior distributional model for the true image itself. Spatial dependence (i.e., nearby values tend to be more - or less - alike than those far apart) is often modeled by assuming a Markov random field (MRF) for the prior model and sometimes for the pixel-intensity model. When dealing with MRFs, there is typically an unwieldy normalizing constant that can cause inference to be either inefficient or computationally intensive. In this article, we propose a class of models that are a subset of the class of MRFs but whose members have probability distributions that can be written in closed form. This class, called the partially ordered Markov models (POMMs), contains as a special case the Markov mesh models (MMMs) and is seen to be an important subclass of graphical models used in the analysis of (Bayesian) networks. POMMs are used in experiments for both the forward problem of texture synthesis and the inverse problem of parameter estimation. Various images of textures are generated using POMMs and are seen not to exhibit any obvious directional patterns. Also, parameter estimates from maximum likelihood estimators are found using a real texture image, and the estimates are then used to generate a texture that is similar to the real data.

62M40 Random fields; image analysis
PDF BibTeX Cite
Full Text: DOI
[1] Abend, K.; Harley, T. J.; Kanal, L. N.: Classification of binary random patterns. IEEE trans. Inform. theory 11, 538-544 (1965) · Zbl 0129.11801
[2] Bartlett, M. S.; Besag, J. E.: Correlation properties of some nearest neighbor models. Bull. int. Statist. institute 43, 191-193 (1969)
[3] Besag, J. E.: On the correlation structure of some two-dimensonal stationary processes. Biometrika 59, 43-48 (1972) · Zbl 0246.62093
[4] Besag, J. E.: Spatial interaction and the statistical analysis of lattice systems. J. roy. Statist. soc. B 36, 192-236 (1974) · Zbl 0327.60067
[5] Besag, J. E.: On the statistical analysis of dirty pictures. J. roy. Statist. soc. B 48, 259-302 (1986) · Zbl 0609.62150
[6] Birkhoff, G., 1940. Lattice Theory. American Mathematical Society, Providence, RI. · Zbl 0063.00402
[7] Cannings, C.; Thompson, E. A.; Skolnick, M. H.: Probability functions on complex pedigrees. Adv. appl. Probab. 10, 26-61 (1978) · Zbl 0431.92019
[8] Chartrand, G.; Lesniak, L.: Graphs and digraphs. (1986) · Zbl 0666.05001
[9] Chellappa, R.; Kashyap, R. L.: Texture synthesis using 2-D noncausal autoregressive models. IEEE trans. Acoust., speech, signal process 33, 194-203 (1985)
[10] Clifford, P.: Markov random fields in statistics. Disorder in physical systems, 19-32 (1990) · Zbl 0736.62081
[11] Connors, R. W.; Harlow, C. A.: A theoretical comparison of texture algorithms. IEEE trans. Pattern anal. Mach intell. 2, 204-222 (1980) · Zbl 0432.68060
[12] Cressie, N.: Statistics for spatial data. (1993) · Zbl 0825.62477
[13] Davidson, J. L.; Cressie, N.: Markov pyramid models in image analysis. Image algebra and morphological image processing IV, society of photo-optical instrumentation engineers (SPIE) proc. 2030, 179-190 (1993)
[14] Davidson, J.; Cressie, N.; Hua, X.: Texture synthesis and pattern recognition for partially ordered Markov models. (1998)
[15] Devijver, P. A.: Image segmentation using causal Markov random field models. Pattern recognition. Lecture notes in computer science 301, 131-143 (1988)
[16] Enting, I. G.: Crystal growth models and Ising models: disorder points. J. phys. C 10, 1379-1388 (1977)
[17] Geman, S.; Geman, D.: Stochastic relaxation, Gibbs distributions and the Bayesian restoration of images. IEEE trans. Pattern anal. Mach intell. 6, 721-741 (1984) · Zbl 0573.62030
[18] Geyer, C. J.; Thompson, E. A.: Constrained Monte Carlo maximum likelihood for dependent data. J. roy. Statist. soc. B 54, 657-699 (1992)
[19] Goutsias, J. K.: Mutually compatible Gibbs random fields. IEEE trans. Inform. theory 35, 1233-1249 (1989) · Zbl 0698.60090
[20] Goutsias, J. K.: Unilateral approximation of Gibbs random field images. Comput. vis., graphics, image process: graphical models image process 53, 240-257 (1991) · Zbl 0799.60045
[21] Gray, A. J.; Kay, J. W.; Titterington, D. M.: An empirical study of the simulation of various models used for images. IEEE trans. Pattern anal. Mach intell. 16, 507-513 (1994)
[22] Hammersley, J. M.; Clifford, P.: Markov random fields on finite graphs and lattices. Unpublished manuscript (1971)
[23] Haslett, J.: Maximum likelihood discriminant analysis on the plane using a Markovian model of spatial context. Pattern recognition 18, 287-296 (1985) · Zbl 0575.62061
[24] Kiiveri, H.; Speed, T. P.; Carlin, J. B.: Recursive causal models. J. australian math. Soc. 36, 30-52 (1984) · Zbl 0551.62021
[25] Kong, A.; Cox, N.; Frigge, M.; Irwin, M.: Sequential imputation and multipoint linkage analysis. Genetic epidemiology 10, 483-488 (1993)
[26] Korezlioglu, H.; Loubaton, P.: Spectral factorization of wide sense stationary processes on Z2. J. multivariate anal. 19, 24-47 (1986) · Zbl 0607.60038
[27] Lacroix, V.: Pixel labeling in a second-order Markov mesh. Signal process 12, 59-82 (1987)
[28] Lauritzen, S. L.: Graphical models. (1996) · Zbl 0907.62001
[29] Lauritzen, S. L.; Dawid, A. P.; Larsen, B. N.; Leimer, H. G.: Independence properties of directed Markov fields. Networks 20, 491-505 (1990) · Zbl 0743.05065
[30] Lauritzen, S. L.; Spiegelhalter, D. J.: Local computations with probabilities on graphical structures and their application to expert systems. J. roy. Statist. soc. B 50, 157-224 (1988) · Zbl 0684.68106
[31] Liu, C. L.: Elements of discrete mathematics. (1985) · Zbl 0592.00005
[32] Liu, C. Y.; Cressie, N.: On the equivalence between symmetric Ising models and some partially ordered Markov models. Statistical laboratory preprint no. 95-9 (1995)
[33] Normand, S.; Tritchler, D.: Parameter updating in a Bayes network. J. amer. Statist. assoc. 87, 1109-1115 (1992)
[34] Pickard, D.: A curious binary lattice process. J. appl. Probab. 14, 717-731 (1977) · Zbl 0385.60083
[35] Pickard, D.: Unilateral Markov fields. Adv. appl. Probab. 12, 655-671 (1980) · Zbl 0439.60100
[36] Qian, W.; Titterington, D. M.: Pixel labelling for three-dimensional scenes based on Markov mesh models. Signal process 22, 313-328 (1991)
[37] Qian, W.; Titterington, D. M.: Multidimensional Markov chain models for image textures. J. roy. Statist. soc. B 53, 661-674 (1991) · Zbl 0800.62512
[38] Spiegelhalter, D. J.; Cowell, R. G.: Learning in probabilistic expert systems. Bayesian statistics 4, 447-465 (1992)
[39] Spiegelhalter, D. J.; Lauritzen, S. L.: Sequential updating of conditional probabilities on directed graphical structures. Networks 20, 579-605 (1990) · Zbl 0697.90045
[40] Tjostheim, D.: Statistical spatial series modelling. Adv. appl. Probab. 10, 130-154 (1978) · Zbl 0383.62060
[41] Tjostheim, D.: Statistical spatial series modelling II: Some further results on unilateral processes. Adv. appl. Probab. 15, 562-584 (1983) · Zbl 0525.62084
[42] Welberry, T. R.: Solution of crystal growth disorder models by imposition of symmetry. Proc. roy. Soc. London A 353, 363-376 (1977)
[43] Whittaker, J.: Graphical models in applied multivariate statistics. (1990) · Zbl 0732.62056
[44] Whittle, P.: On stationary processes in the plane. Biometrika 41, 434-449 (1954) · Zbl 0058.35601
[45] Whittle, P.: Stochastic processes in several dimensions. Bull. int. Statist. inst. 40, 974-994 (1963) · Zbl 0129.10603
[46] Woods, J. W.: Two-dimensional discrete Markovian fields. IEEE trans. Inform. theory 18, 232-240 (1972) · Zbl 0235.60063
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. It attempts to reflect the references listed in the original paper as accurately as possible without claiming the completeness or perfect precision of the matching.