基于三维几何特征的点云处理技术及应用研究

Three-dimensional optical metrology and its application catches more attention in the manufacturing as well as other related fields. It is more convenient to acquire the three dimensional data (point cloud). As compared with the traditional two-dimensional image, point cloud has one more dimension and the data is far more than before. The non-uniform distribution of the point cloud makes the neighboring relationship is more complex and thus the retrieval more difficult. All these give the challenge to the point cloud processing. This proposal proposes to study the point cloud processing framework based on the three-dimensional geometric features. The local geometric features are extracted from the point cloud first and then registered to accomplish measurement. After solving the key-techniques including filtering and de-noising based on fast neighbor points search, segmentation based on local surface geometric parameters, line point cloud feature extracting based on attitude standardization and max absolute main curvature, and machining allowance evaluation based on plane feature point cloud, the point cloud processing software will be developed, which gives a new automatic and effective way to the evaluation analysis of the machining allowance of the precision cast as well as other point cloud based measurement. It establishes the fundamental for the optical three dimensional measurement and is meaningful both theoretically and practically.

光学三维测量及应用在制造业等诸多领域受到越来越多的关注，目前已可以便捷地获取被测对象的三维数据（即：点云）。相较于传统的二维图像，三维点云数据维数更多，数据量更大；且数据点在空间非均匀分布，邻接关系更复杂，数据检索更困难；对点云数据处理提出了更大的挑战。本项目针对三维精密测量对点云处理的迫切需求，提出研究具有通用意义的基于三维几何特征的点云处理技术：先从点云中提取对应于被测对象局部结构的几何特征点云，再对特征进行配准实现测量。拟重点突破基于邻近点云快速搜索的点云滤波去噪技术、结合点云局部几何参数迭代生长的点云分割技术、基于姿态标准化和最大绝对主曲率的点云特征提取技术、基于平面特征的加工余量估计等关键技术和应用难点，开发相应的点云获取和处理系统。本项目研究结果将为精铸件毛坯加工余量估计等基于点云的测量应用提供创新的解决思路，可提高测量效率和自动化程度，具有重要的理论研究意义和实用价值。

光学三维测量及应用在制造业、学科交叉前沿等诸多领域受到越来越多的关注，目前已可以便捷地获取被测对象的三维数据（点云）。相较于传统的二维图像，三维点云数据维数更多，数据量更大；且数据点在空间非均匀分布，邻接关系更复杂，数据检索更困难；对点云数据处理提出了更大的挑战。.本项目针对三维精密测量对点云处理的迫切需求，研究通用的基于三维点云几何特征的点云处理技术：先从点云中提取被测对象局部几何特征点云，再对特征点云进行配准实现测量。.在对三维点云数据处理调研的基础上，提出了基于点云空间关系的滤波及去噪方法，结合点云局部几何参数迭代生长的点云分割方法，点云分割及三角化一体化方法；在点云法向估计及曲率估计的基础上，提出了三维点云几何特征提取方法；提出了基于离散旋转图像拼接、基于归一化对齐径向特征的多视场点云拼接方法；优化构建了数据采集装置；实现了三维点云几何特征提取数据处理软件；以精铸件毛坯加工余量分析为应用场景，开展了基于特征的点云处理技术应用，验证了前述技术的有效性。该项目成果为基于三维点云处理技术的测量应用奠定了基础，可提高测量及数据处理的效率和自动化程度，为智能制造、前沿交叉学科提供新型测量手段，具有重要的理论研究意义和实用价值。