Unsupervised Image Classification with a Hierarchical EM Algorithm

This work takes place in the context of hierarchical stochastic models for the resolution of discrete inverse problems from low level vision. Some of these models lie on the nodes of a quad-tree which leads to non-iterative inference procedures. Nevertheless, if they circumvent the algorithmic drawbacks of grid-based models (computational load and/or great dependance on the initialization), they admit modeling shortcomings (cumbersome and somehow artificial). We investigate a new hierarchical stochastic model which takes benefit from both the spatial and the hierarchical prior modelings. The independance graph is based on a tree which has been pollarded with the nodes at the coarsest resolution exhibiting a grid-based interaction structure. For this class of models, we address the critical problem of parameter estimation. To this end, we derive an EM algorithm on the hybrid structure which mixes an exact EM algorithm on each subtrees and a low cost Gibbsian EM algorithm on the coarse ...

• Type: Conference
• Year: 1999
• Where: ICCV
• Authors: Annabelle Chardin, Patrick Pérez
• Keywords: Computer Vision, Exact Em Algorithm, Gibbsian EM Algorithm, Grid-based Interaction Structure, Hierarchical Prior Modelings, Hierarchical Stochastic Models, ICCV 1999