Technical section: A potential-based generalized cylinder representation

  • Authors:

  • Jen-Hui Chuang;Narendra Ahuja;Chien-Chou Lin;Chi-Hao Tsai;Cheng-Hui Chen

  • Affiliations:

  • Department of Computer and Information Science, National Chiao Tung University, Hsinchu, Taiwan 30010, ROC;Department of Computer and Information Science, National Chiao Tung University, Hsinchu, Taiwan 30010, ROC;Department of Computer and Information Science, National Chiao Tung University, Hsinchu, Taiwan 30010, ROC;Department of Computer and Information Science, National Chiao Tung University, Hsinchu, Taiwan 30010, ROC;Department of Computer and Information Science, National Chiao Tung University, Hsinchu, Taiwan 30010, ROC

  • Venue:
  • Computers and Graphics
  • Year:
  • 2004

Quantified Score

Hi-index 0.00

Visualization

Abstract

Generalized cylinder (GC) is a class of parametric shapes that is very flexible and capable of modeling many different types of real-world objects, and have subsequently been the focus of considerable research in the vision community. Most of the related works proposed previously have dealt with the recovery of 3D shape description of objects based on the GC representation from one or more 2D image data. Different from the objective of the previous works, in this paper, we will propose a new approach to obtain a GC-based shape description of 3D objects. The proposed approach of deriving the GC axis is a further extension of the potential-based skeletonization approach presented in (IEEE Trans. Pattern Anal. Mach. Intell. 22(11) (2000) 1241). Simulation results demonstrate that the derived GC representation will yield better approximation of object shape than that based on simpler subclasses of GC since there is, in principle, no restriction on the topology of the GC axis and the shape of the cross-sections.