Robust Morphing of Point-Sampled Geometry: Research Articles

CX Xiao,WT Zheng,QS Peng,AR Forrest

DOI: https://doi.org/10.1002/cav.22

IF: 1.01

2004-01-01

Computer Animation and Virtual Worlds

Abstract:We propose a novel morphing algorithm for objects represented by point-sampled geometry. The fundamental problem of point-sampled geometry morphing is how to set the correspondence between points of the two objects which are usually of different size. The two objects are first parameterized by projecting the sample points onto a common parametric domain. As both objects are densely sampled, we present a novel accelerated parameterization algorithm employing the technique of LOD. The common parameter domain is then split recursively into clusters. The correspondence between sample points of the two objects is established by performing a local mapping in each cluster. As for complex geometries, the establishment of correspondence is facilitated by decomposing the geometry into patches using geodesic decomposition curves.To preserve the features during morphing, a process of features assignment is incorporated. By re-sampling the in-between object dynamically and adaptively, the cracks that would occasionally occur during morphing are successfully eliminated. Experiment results show that our algorithms are fast, stable and easy to implement. High-quality morphing is produced. Copyright © 2004 John Wiley & Sons, Ltd.