基于人体线点云的光顺去噪.docVIP

基于人体扫描线点云的光顺去噪 李晓志(，李晓久，王玉秀 （天津工业大学 艺术与服装学院服装工程系,天津 300387） 摘 要 利用课题组新研制的便携式三维人体测量系统获得人体扫描线点云。在测量过程中，由于受扫描设备的物理特性、扫描环境、光照和标定算法本身的缺陷等因素的影响，点云中不可避免混有噪声点。根据噪声点的特点将其分为3类：与人体真实数据相隔较远的点为第1类噪声点，点云中重叠的点为第2类噪声点，夹杂在人体真实数据点中而导致点云不光顺的点为第3类噪声点。对于第1类和第2类噪声点，采用基于距离的选点删除法；对第3类噪声点采用基于离散点云曲率及曲率变化的Savitzky-golay滤波方法。实验结果显示，以上2种方法能有效去除人体扫描线点云中的噪声点，并能保持点云固有的几何细节。 关键字 人体；三维人体扫描；扫描线点云；光顺去噪；选点删除法；Savitzky-golay滤波 中图分类号:TP391.72 文献标识码:A 3D body scan line point cloud data denoising and smoothing LI Xiaozhi LI Xiaojiu WANG Yuxiu (Department of Art and Clothing, Tianjin Polytechnic University, Tianjin 300384, China) Abstract Our team developed a portable three dimension (3D) body measurement system. By measuring the body using the system, the body scan line point clouds are reconstructed. Because of physical properties of measuring device, environment of measurement, light and the defects of calibration algorithm, the initial point clouds include some noisy points. Base on the characters of noisy points, they are divided into three types: the first type has a far distance from true body data points, the second type is the lapped points and the third type mixes with true body data to cause the point clouds unfair,. The first and second type noisy points are deleted by the distance-based algorithm, and then the third type noisy points are removed by the method of Savitzky-golay filter based on the curvature and curvature’s change. The results show the methods are effective on smoothing point clouds and the details of point clouds are remained. Key words Body; 3D body scanning; Scan line point clouds; Denoising and smoothing; Choose point and delete it; Savitzky-golay filter  非接触式三维人体测量已成为现代人体测量技术的主要特征，它以现代光学为基础，融光电子学、计算机图像学、信息处理、计算机视觉等科学技术为一体的测量技术，具有比传统测量方法更快速、准确、效率高等特点[1]。目前，大多数三维人体测量系统的工作原理都是通过光照系统往人体上投射光源（如激光、红外光、多条纹光等），使用视觉设备来捕获人体的外形，然后利用系统软件来提取人体在三维空间中的数据并对数据进行处理以及重构数字化三维人体模型。 通常把三维空间中数据点的集合称为点云[2]，根据测得的初始点云中点的分布特征，可以将点云分为散乱点云和扫描线点云[3]。扫描线点云由一组扫描线组成，扫描线上的点位于扫描平面内，沿扫描方向非常密集