三维模型轴向体积分布特征提取及比较算法

Extracting volume distribution feature from 3D models

在分析网格数据结构的基础上,提出一种提取任意三维网格模型体积分布特征的算法.该算法首先应用主元分析法确定模型的主轴方向,再以一组等距的平行平面从三个坐标轴方向对模型进行剖分处理,通过计算截面面积求得模型沿三个主轴方向的体积分布特征.模型在三个坐标轴向上的体积分布描述了模型的几何特征,通过体积分布特征数据得到不同模型间的相似性度量.实验表明,该算法程序运行稳定、快速,可用于提取具有任意几何和拓扑复杂性的各类网格模型的体积分布特征.

The key issue of 3D model retrieval is how to extract features from a given model.Distribution of volume along certain direction in 3D space is one of the most important features for a geometric object.In this paper,we propose an algorithm to extract volumetric characters from arbitrary models based on half-edge data structure.The algorithm first rotates a mesh model to coordinate-axis-aligned position by employing Principal Component Analysis (PCA) method,and then uses a set of parallel equidistant planes to slice the model into several segments along three principal axes,respectively.From each cross section of the model,a polygon set composed of one or more polygons can be obtained and the areas of the cross section can be computed on the fly.After normalizing these areas,we plot three volume distribution curves along corresponding principal axes.The curves give a shape descriptor of the model.Experiments show that our algorithm is efficient,stable,and can process 3D models with arbitrary geometry and topology.