Markov random field modeled range image segmentation

In this paper range image segmentation is cast in the framework of Bayes inference and Markov random field modeling. To facilitate the inference from distance measurement to labeling set, we introduce the set of surface function parameters as another estimation and construct a novel model accordingly. Subsequent study shows that range image segmentation can be formulated as a combinatorial optimization problem. This model-based optimization will be used as a postprocessing module after an edge-based initial segmentation in our hybrid segmentation scheme. Qualitative improvement is observed from experimental results. 2003 Elsevier B.V. All rights reserved.