¡@

In this paper, we present a free-form deformation (FFD) approach to manipulate point-sampled surface. Our approach provides users the powerful free-form deformation directly on point-sampled surface. In the preprocessing, we use a robust extension of hierarchical clustering scheme to partition a point-sampled model into clusters, fit a moving least square (MLS) surface to each of them, and define a valid polygonal area on the base domain of the MLS surface. At run time, we apply the conventional free-form deformation technique on these clusters. While users adjust the grids of the FFD lattice, we propose a novel interpolation method, which responses to the curvature variation during object deformation, to curve the base domain of the MLS surfaces of those clusters concerned. Finally, we resample the new base domain to produce the deformed point set model. The proposed technique is very intuitive, effective and easy to implement. With this technique, high frequency details from the original point-set surface can be maintained well on the deformed surface and several interesting deformed results of pointsampled models are demonstrated to verify the proposed scheme.

Keywords: point-sampled models, clustering, moving least square (MLS) surface, freeform deformation (FFD), interpolation

Retrieve PDF document (200705_06.pdf)

Received April 1, 2005; revised July 4, 2005; accepted August 3, 2005.
Communicated by Pau-Choo Chung.
^*This project was supported by the National Science Council of Taiwan, R.O.C., under contracts No. NSC-93- 2213-E-006-026, NSC-94-2213-E-006-063, NSC-94-2213-E-006-005, and NSC-94-2213-E-218-016. The bunny, David, Lion models are courtesy of Stanford University and the Octopus and Igea models are courtesy of ETH Zurich.