A Maximum-Likelihood Approach to Segmenting Range Data

Abstract

This paper addresses the problem of segmenting a range image into homogeneous regions in eacb of which the range data correspond to a difierent surface. The segmentation sought is a maximum-likelihood (ML) segmentation. As most manufactured parts are well approximated by patches of planes, cylinders, and spheres, we consider these three simple surfaces as the only surfaces present in the image. These surfaces have very distinct differences in their symmetries (plane, axis, or point of symmetry) that make them very appealing. The basic approach to segmentation is to divide the range image into Windows, classify each window as a particular surface primitive, and group like windows into surface regions. The window size is chosen such that: i) each window patch is approximalely planar—planar facet—(or a mixture of planes, for windows lying at the boundaries between surfaces); and ii) if we examine a local neighborhood about the window to be classified, enough curvature information is provided to detect the differences in the symmetries of these three surfaces. Mixed windows are detected by testing the hypothesis that a window is homogeneous. Homogeneous windows are classified according to a generalized likelihood ratio test which is computationally simple and incorporares information from adjacent windows. Grouping windows of the same surface types is cast as a weighted ML clustering problem. Finally, mixed windows are segmented using an ML bierarchical segmentation algorithm. The resulting regions and their associated ML surface parameter and boundary estimates could them be used to perform ML object position estimation and matching. A similar approach is taken for segmenting visible-light images of Lambertian objects illuminated by a point source at infinity. © 1988 IEEE