2ND-ORDER IMPLICIT POLYNOMIALS FOR SEGMENTATION OF RANGE IMAGES

In this paper, we present an algorithm for segmentation and feature ex traction for noisy range images. We use the edge- and region-based app roach for obtaining reliable segmentation maps. For edge detection, re gions having much higher values of planar fit error are assumed to con tain edges. These edge regions are explored for detection of crease ed ges, which can be located at the maxima of the curvature. However, due to noise, the maximas are generated at a number of points other than the actual edges. These extraneous curvature maximas have a substantia lly smaller value of curvature compared with the curvature of the actu al edges. The true edges are detected using an approach based on local thresholding within edge regions. The detected edges segment the rang e image into smooth patches. On each of the smooth patches, a planar/q uadric fit is obtained using a novel non-iterative procedure. To obtai n reliable fits, we had to restrict the permissible class of quadrics to surfaces of revolution. The quadric fit obtained gives a very relia ble estimate of the position and orientation of the axis for curved su rfaces. The approach has been demonstrated on real range data and has been found to give good results. (C) 1996 Pattern Recognition Society. Published by Elsevier Science Ltd.