• 论文 •    

基于非线性主成分分析的最小包围盒计算方法

陈柏松,叶雪梅,安利   

  1. 第二炮兵工程学院 401教研室,陕西西安710025
  • 出版日期:2010-11-15 发布日期:2010-11-25

Minimum bounding box calculation based on nonlinear principle component analysis

CHEN Bai-song, YE Xue-mei, AN Li   

  1. 401 Staff room,The Second Artillery Engineering College, Xi'an 710025, China
  • Online:2010-11-15 Published:2010-11-25

摘要: 研究了三维模型最小有向包围盒算法,提出了基于非线性主成分分析的最小包围盒计算框架,根据Voronoi区域和Barycentric区域定义了顶点区域面积,并根据非线性主成分分析原理,使用顶点区域面积对各顶点进行调制,消除顶点分布的不均匀性,再使用传统主成分分析得到目标正交坐标系,从而得到最小包围盒。最后,定义了一个相似度系数,来度量最小包围盒的计算效果。通过实验表明,使用顶点面积进行调制可以得到比其他方法更稳定的结果,且计算复杂度较小。

关键词: 非线性主成分分析, 最小包围盒, 顶点面积, 三维模型

Abstract: Minimum bounding box of 3D model was studied, and a calculation framework based on nonlinear Principle Component Analysis (PCA) was proposed. The regional area of the vertex was defined according to Voronoi regions and Barycentric region. Using the theory of nonlinear PCA, the vertex of 3D modal was modulated by the regional area of the vertex to eliminate the uneven distribution of vertex. Then, the objective orthogonal coordinate system was obtained by traditional PCA, thereby the minimum bounding box was calculated. To measure the effect of the minimum bounding box algorithm, a new similarity coefficient was defined. Experiment results showed that the method by using the regional area of the vertex to modulation was more stable with less complicated computations.

Key words: nonlinear principle component analysis, minimum bounding box, regional area of the vertex, 3D model

中图分类号: