Bayesian pixel classification using spatially variant finite mixtures and the generalized EM algorithm
S. Sanjay‐Gopal, Thomas J. Hebert · IEEE Transactions on Image Processing · 1998
A spatially variant finite mixture model is proposed for pixel labeling and image segmentation. For the case of spatially varying mixtures of Gaussian density functions with unknown means and variances, an expectation-maximization (EM) algorithm is derived for maximum likelihood estimation of the pixel labels and the parameters of the mixture densities, An a priori density function is formulated for the spatially variant mixture weights. A generalized EM algorithm for maximum a posteriori estimation of the pixel labels based upon these prior densities is derived. This algorithm incorporates a variation of gradient projection in the maximization step and the resulting algorithm takes the form of grouped coordinate ascent. Gaussian densities have been used for simplicity, but the algorithm can easily be modified to incorporate other appropriate models for the mixture model component densities. The accuracy of the algorithm is quantitatively evaluated through Monte Carlo simulation, and its performance is qualitatively assessed via experimental images from computerized tomography (CT) and magnetic resonance imaging (MRI).
Cited by 207 works. doi:10.1109/83.701161
Authors
References
- Stochastic Relaxation, Gibbs Distributions, and the Bayesian Restoration of Images
- A new look at the statistical model identification
- Digital Picture Processing
- Maximum likelihood estimation from incomplete data via the EM algorithm
- Scene Labeling by Relaxation Operations
- Statistical Analysis of Finite Mixture Distributions.
- On the Foundations of Relaxation Labeling Processes
- Statistical Analysis of Finite Mixture Distributions.
- Modeling and Segmentation of Noisy and Textured Images Using Gibbs Random Fields
- 10.2 Maximum Likelihood from Incomplete Data