AN INTEGRATED APPROACH FOR RANGE IMAGE SEGMENTATION AND REPRESENTATION

This paper proposes an efficient method for the segmentation and repre sentation of 3D rigid, solid objects from its range images using diffe rential invariants derived from classical differential geometry. An ef ficient algorithm for derivation of surface curvatures, which are affi ne invariants, at smooth surface patches is proposed. The surface is a pproximated by Bezier and Beta-splines to compare qualitatively the pr oposed segmentation scheme. This scheme leads to derivation of surface features, which provides a very robust surface segmentation. An integ rated approach represents the surface in terms of plane, quadric and s uperquadric surface. Experiments show excellent performance and togeth er with the inherent parallelism make the scheme a promising one. Pres ent experiments were conducted on some real range images where most of the parts of the object are planar.