基于三维点云无网格处理的大型复杂锻件结构特征曲线重建

项目名称： 基于三维点云无网格处理的大型复杂锻件结构特征曲线重建

项目编号： No.51305390

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 机械、仪表工业

项目作者： 林洪彬

作者单位： 燕山大学

项目金额： 24万元

中文摘要： 大型复杂锻件是核电安全化与国防装备大型化的基础和关键，锻件结构特征尺寸的有效提取对提升锻件质量、提高锻件合格率具有重要作用。随着三维测量手段的不断完善，利用测得三维数据重建大型复杂锻件结构特征成为当前研究热点。迄今大型复杂锻件测量数据处理仍局限于传统的网格化方法，大型复杂锻件测量数据自身的散乱性、强噪声、海量性、局部化和结构复杂性特点使得网格化处理方法的效率、精度和可靠性得不到保障。人们试图通过网格化处理重建大型复杂锻件结构特征的努力遇到巨大挑战。本项目提出通过无网格处理实现三维点云保特征消噪、精简和鲁棒拼接，为大型复杂锻件结构特征精确、快速、可靠重建提供保障；通过构建三维点云多尺度张量特征指标，实现采样点特征属性的有效识别，结合Snake模型实现大型复杂锻件结构特征有效重建。本项目可为大型复杂锻件结构特征重建提供新途径，对深化低质量大型复杂三维实体结构特征重建基础研究深化有重要科学意义。

中文关键词： 大型复杂锻件；点云处理；点云配准；点云精简；结构特征提取

英文摘要： Complex large forgings are foundation and key to safety of nuclear power and maximization of national defense equipments. Effective extraction of structural characteristic scale from complex large forgings plays important role in improving quality and qualified rate of large complex forgings. With the development of 3D measurement technology, reconstructing structural feature from measured 3D data of complex large forgings becomes research hotspot. Classical mesh based methods are still used in complex large forgings 3D data processing as yet. However, the efficiency, accuracy and reliability of these methods are restricted due to the inherent confusion, strong noisy, massive, localization and structural complexity characteristic of 3D data measured from complex large forgings. People are challenged to reconstruct the structural feature of complex large forgings based on meshed methods. A novel conception is proposed in this subject, and mesh free methods will be researched in depth to feature preserving de-noise and simplify the original 3D point clouds, and robustly register the 3D point clouds. This will lay a foundation for rapid, accurate, reliable reconstruction of structural feature curve of complex large forgings. Further more, multi-scale tensor feature index will be constructed and used in identificati

英文关键词： Complex Large Forgings；Point Cloud Processing；Point Clouds Registation；Point Clouds Simplification；Structrual Feature Extraction