基于非均匀细分模型和特征保持的海量车身线扫描点云采样方法研究

项目名称： 基于非均匀细分模型和特征保持的海量车身线扫描点云采样方法研究

项目编号： No.51205015

项目类型： 青年科学基金项目

立项/批准年度： 2013

项目学科： 机械工程学科

项目作者： 周煜

作者单位： 北京航空航天大学

项目金额： 25万元

中文摘要： 以海量线扫描车身点云的数据采样为研究对象，以空间微分几何相关理论为理论基础，从线扫描车身点云数据的空间结构特点出发，通过分析线扫描数据拓扑结构与点云曲率之间的关系，建立以法矢变化量为判据的非均匀细分模型，研究一种可同时满足"四度"（时间复杂度、曲率适应度、边界保护度、自动分块度）要求的数据采样算法。在此基础上开发适用于海量车身点云数据的集成采样系统。该方法不需要借助附加曲面和三角片模型，而是利用线扫描数据的拓扑结构实现点云的非均匀采样。这对于车身点云的高速预处理具有重大的现实意义和工程应用价值。其研究成果为进一步开发新型的空间线扫描点云集成预处理系统提供理论基础，也同样适用于具有相似结构特点的大尺寸复杂曲面的数据采样，具有广阔的应用前景。

中文关键词： 非均匀细分；数据采样；车身点云；特征保持；线扫描

英文摘要： Based on spatial differential geometry theory, by analyzing the relations between point cloud curvature and topological structure of line-scan data, after establishing non-uniform subdivision model based on vector-change, the algorithm of data reduction for massive line-scan automobiles-bodies point cloud was presented which had lower time-complexity and the ability of curvature-adaptability, boundary-protection, automatic segmentation. The software system of data reduction for massive automobiles-bodies point cloud was developed by this algorithm. Without any fitted surface and triangulation, it carried out the non-uniform data reduction with the help of the topological structure of line-scan data. It had great practical value and engineering significance for high-speed automobiles-bodies data pre-processing. The research result provided a theoretical basis for the development of integrated software system of spatial line-scan data pre-processing. It also suitable to the data reduction of large-size complex surfaces which had similar structure. So it has a broad applied prospect.

英文关键词： Non-Uniform；Data Reduction；Automobile-bodies；Feature-Preservation；Laser Scanning