基于局部形状分布的三维CAD模型检索算法

3D CAD Model Retrieval Using Local Shape Distributions

为了对三维CAD模型的检索结果能够体现模型局部细节特征的相似性差异,提出一种基于局部形状分布的三维CAD模型检索算法。在模型表面选取随机采样点;根据采样点与其邻域内各点的测地距离,构建模型在该区域相应的局部形状分布直方图;通过对比不同模型各自对应的局部形状分布直方图,实现对模型的相似性检索。实验结果表明:本文算法对三维CAD模型的检索性能优于整体形状分布算法和球面谐波算法,使模型的局部细节特征具有更高的分辨能力。

A novel algorithm using local shape distribution was proposed to address the problem of indicating the similarity of model＇s local parts in 3D CAD model retrieval. Firstly, sampling points were randomly chosen in the surface of a model. Then, for each sampling point and all points in its neighborhood, their geodesic distances were calculated, and a local shape distribution histogram was created accordingly. Finally, the similarity between two models was determined by the likeness of their all corresponding local shape distribution histograms. The experimental results demonstrate that the proposed method obtains better retrieval performance compared to the state-of- the -art.