Hierarchical segmentation of range and color image based on Bayesian decision theory

This paper describes recent work on hierarchical segmentation of registered color and range images. The algorithm starts with an initial partition of small first order regions using a robust fitting method constrained by the detection of depth and orientation discontinuities in the range signal and color edges in the color signal. The algorithm then optimally group these regions into larger and larger regions until an approximation limit is reached. The algorithm uses Bayesian decision theory to determine the local optimal grouping and the complexity of the models used to represent the range and color signals. Experimental results are presented.