Integrating edge and surface information for range image segmentation

Abstract

The authors show that the synergetic combination of both surface- and edge-based segmentation processes can lead to more accurate segmentation results than just either one alone. A segmentation algorithm which integrates edge and surface information to come up with a high-level description of a scene given a range image is proposed. It is shown that the surface parameters can be used for accurate computation of surface discontinuity parameters. The presence or absence of a predicted surface discontinuity in the image can be used to guide and correct the process of extracting homogeneous surface regions in the range image. Experimental results on real range images indicated the advantages of the proposed technique.<>