Probabilistic Modeling of Discrete Images
We present a methodology for constructing probabilistic models of discrete images using Gibbs distributions. The method differs from previous approaches in that the formulation is suited for the modeling of discrete images, and hence readily applicable to discrete tomography problems. Second, the distribution is “image-modeling” in the sense that random samples drawn from the distribution are likely to share important characteristics of the images in the application area. We propose a technique to estimate parameters of the model,based not only on local characteristics of the images, but also on their global properties. Using it as image priors in a Bayesian framework, we apply the model to the problems of recovering discrete images corrupted by additive Gaussian noise and discrete tomographic reconstruction. We demonstrate the usefulness of the proposed model and compare its effectiveness with previous approaches.
Unable to display preview. Download preview PDF.
- J. Besag “On the statistical analysis of dirty pictures (with discussion).” Journal of the Royal Statistical Society Series B 48, 259–302 (1986).Google Scholar
- S. Geman and D. McClure, “Statistical methods for tomographic image reconstruction,” In Proceedings of the.46th Session of the ISI, Bulletin of the ISI, volume 52 pp. 5–21, 1987.Google Scholar
- R. Leahy and X. Yan, “Incorporation of anatomical MR data for improved functional imaging with PET”In 12th International Conference on Information Processing in Medical Imaging, pp. 105–120, 1990.Google Scholar
- J. Besag “Spatial interactions and the statistical analysis of lattice systems (with discussion),” Journal of the Royal Statistical Society Series B 36, 192–236 (1974).Google Scholar
- Z. Zhou, R.M. Leahy, and E.U. Mumcuoglu, “Maximum likelihood hyperparameter estimation for Gibbs priors from incomplete data with applications to PET,” In R. Di Paola Y. Bizais, and C. Barillot, Information Processing in Medical Imaging (Kluwer Academic Publishers, Dordrecht, The Netherlands) pp. 39–52, 1995.Google Scholar
- H. Gould and J. Tobochnik, An Introduction to Computer Simulation Methods: Application to Physical Systems (Part 2) (Addison-Wesley Publishing Company, Reading, MA), 1988. Chapter 12.Google Scholar
- G. T. Herman, M. Chan, Y. Censor, E. Levitan, R. M. Lewitt, and T. K. Narayan, “Maximum a posteriori image reconstruction from projections,” In S. E. Levinson and L. Shepp, Image Models (and their Speech Model Cousins), (Springer-Verlag, New York) pp. 53–89, 1996.Google Scholar
- A. Papoulis, Probability, Random Variables, and Stochastic Processes, Second Edition (McGraw-Hill, New York), 1984.Google Scholar
- C. Yang “Efficient stochastic algorithms on locally bounded image space,” CVGIP: Graphical Models and Image Processing 55, 494–506 (1993).Google Scholar