Bayesian edge regularization in range image segmentation

We present in this paper a new method for improving range image segmentation, based on Bayesian regularization of edges produced by an initial segmentation. The method proceeds in two stages. First, an initial segmentation is produced by a randomized region growing technique. The produced segmentation is considered as a degraded version of the ideal segmentation, which should be then refined. In the second stage, image pixels not labeled in the first stage are assigned to the resulting regions by using a Bayesian estimation based on some prior assumptions on the region boundaries. The image priors are modeled by a new Markov Random Field (MRF) model. Contrary to most of the authors in range image segmentation, who use surface smoothness MRF models, our MRF model is based on. the smoothness of region boundaries, used to improve the initial segmentation by a Bayesian regularization of the resulting edges. Tests performed with real images from the ABW database show a good potential of the proposed method for significantly improving the segmentation results.