Multiresolution image classification by hierarchical modeling with two-dimensional hidden Markov models

Jia Li, R.M. Gray, R.A. Olshen
2000 IEEE Transactions on Information Theory  
The paper treats a multiresolution hidden Markov model MHMM for classifying images. Each image is represented by feature vectors at several resolutions, which are statistically dependent as modeled by the underlying state process, a multiscale Markov mesh. Unknowns in the model are estimated by maximum likelihood, in particular by employing the EM algorithm. An image is classi ed by nding the optimal set of states with maximum a posteriori probability MAP. States are then mapped into classes.
more » ... ped into classes. The multiresolution model enables multiscale information about context to be incorporated into classi cation. Suboptimal algorithms based on the model provide progressive classi cation which i s m uch faster than the algorithm based on single resolution HMMs. Keywords image classi cation, image segmentation, two dimensional multiresolution hidden Markov model, EM algorithm, tests of goodness of t
doi:10.1109/18.857794 fatcat:3g2bstyf55dclhkvnc6gs6p6ni