Compound Gauss-Markov random fields for image estimation
- 1 March 1991
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Signal Processing
- Vol. 39 (3) , 683-697
- https://doi.org/10.1109/78.80887
Abstract
Algorithms for obtaining approximations to statistically optimal estimates for images modeled as compound Gauss-Markov random fields are discussed. The authors consider both the maximum a posteriori probability (MAP) estimate and the minimum mean-squared error (MMSE) estimate for image estimation and image restoration. Compound image models consist of several submodels having different characteristics along with an underlying structure model which govern transitions between these image submodels. Two different compound random field models are employed, the doubly stochastic Gaussian (DSG) random field and a compound Gauss-Markov (CGM) random field. The authors present MAP estimators for DSG and CGM random fields using simulated annealing. A fast-converging algorithm called deterministic relaxation, which, however, converges to only a locally optimal MAP estimate, is also presented as an alternative for reducing computational loading on sequential machines. For comparison purposes, the authors include results on the fixed-lag smoothing MMSE estimator for the DSG field and its suboptimal M-algorithm approximation.<>Keywords
This publication has 14 references indexed in Scilit:
- A parallel image segmentation algorithm using relaxation with varying neighborhoods and its mapping to array processorsComputer Vision, Graphics, and Image Processing, 1987
- Image Estimation Using Doubly Stochastic Gaussian Random Field ModelsPublished by Institute of Electrical and Electronics Engineers (IEEE) ,1987
- A parallel approach to the picture restoration algorithm of Geman and Geman on an SIMD machineImage and Vision Computing, 1986
- Optimization by Simulated AnnealingScience, 1983
- Anisotropic Nonstationary Image Estimation and Its Applications: Part I--Restoration of Noisy ImagesIEEE Transactions on Communications, 1983
- Estimation and choice of neighbors in spatial-interaction models of imagesIEEE Transactions on Information Theory, 1983
- Digital image restoration using spatial interaction modelsIEEE Transactions on Acoustics, Speech, and Signal Processing, 1982
- Adaptive nonlinear image restoration by a modified Kalman filtering approachIEEE Transactions on Acoustics, Speech, and Signal Processing, 1981
- Kalman filtering in two dimensionsIEEE Transactions on Information Theory, 1977
- Two-dimensional discrete Markovian fieldsIEEE Transactions on Information Theory, 1972