三维点云预处理与网格模型多分辨率实时绘制技术的研究-光学工程专业论文.docxVIP

摘要 基于激光三维扫描技术的人体数字化己广泛应用于服装工业、康复工程、产 品设计以及虚拟现实等领域，具有童耍的应用价值和广阔的发展前景。 本文采用激光王维人体扫描系统和高分辨率人股扫描系统获得的三维点云， 实现了数据重来样、光顺去噪、数据精简等点云预处理算法以及网格的多分辨率 实时绘制技术，主要工作与创新点如下 z ，队1.采用基于主曲线与 cardinal 三次样条插值的点云童来样算法，实现了多 传感器统一、杂散噪声消除。通过人体自动分割、单层点分块等预操作可以将该 ，队 2. 提出了基于点云曲率特征混合分类的双边滤波平滑算沽，该方法结合双 边滤波和平面投影的方法，针对所划分的平面类、次特征类、富特征类三种邻域 类型的点集进行自适应滤波，并采用变参数双边滤波器，同时避免了法向调整的 费时运算。 3. 实现了基于周部曲事的体栅格点云精简算法，并对基于通归 BSP 树的层 次聚类算法进行改进，在其基础上实现了采用优先级队列的聚类精简算法，提高 了运行效草和可操作性。 4. 采用基于二次误差度量的动态简化方法构建有序渐进附格，并在此基础 上采用信息查找表、三角形队列以及顶点映射表等一系列数据结构实现了基于视 距远近、视窗大小以及模型旋转、平移的王维网格模型的 LOD (多层次细节) 实时给制。 .、Jq .、Jq iaF ABSTRACT .. ，t(俑 ， 4 伽 s Digital human modeling，which is based on the technology of3D laser scanning， is of important value in application and broad prospects for development.lt is therefore widely used in g即ment industry，health and rehabi1itation， production design， and virtual reality，础。鸣others. η山pap时 ado严s 3D point-cloud ca阴阳d by 3D laser human body scanner and high resolution 3D 伽ce scanne弘 Plenty of algorithms 町e realized such as data resampl吨， denois吨， decimation and real-time LOD rendering of tríangle model.The main innovations in the research are as follows: A data resamping algorithm based on principal curves and cardinal cubic spline interpolation is adopted. This method effectively solved multi哺ensor data me咆ing and denoising.By body sengmentation and data partition，批 is also e班iciently applied in complicated surface such 部 3D human body data. An effective and feature-preserving 3D point-c1oud smoothing method，based on the multifeatured c1assífication of curvature characteristic ，is proposed. This method originated from local plane 伽íng and bilateral denoising. Accordíng to the features，the data set are divided into plane，suh-featured and rich-featured neighborhoods. Considering the different neighborhood types，a bilateral filtering of variable param创ers is proposed. And the large calculation of normal directíon adjustment 盯ffectively avoided. A dat r