【作者】 王志举； 贾克斌； 陈嘉平；

【Author】 WANG Zhiju;JIA Kebin;CHEN Jiaping;Faculty of Information Technology, Beijing University of Technology;Beijing Laboratory of Advanced Information Networks;

【机构】 北京工业大学信息学部； 先进信息网络北京实验室；

【摘要】 车辆的高效检测方法在保证车辆整体性能与提高国防装备水平中都发挥着关键作用,车辆底盘的特性是衡量车辆移动性能的关键参考标准。设计并实现了一套高精度、自动化的快速车辆底盘轮廓重建系统,使用遥控机器人搭载多线激光雷达对车辆底盘轮廓进行扫描,提出了一种基于ICP算法结合直通滤波的底盘点云重建方法,基于随机抽样一致性算法剔除地面点云,并提出了一种下采样结合统计滤波器的底盘点云去噪方法。实验结果表明:基于机器人移速方法重建的点云聚类平均欧氏距离为11.62 mm,且点云出现了形变,而提出的基于ICP算法结合直通滤波方法重建的点云聚类平均欧氏距离为4.43mm,降低了61.8%,融合紧凑度更高,且点云无形变,更好地满足了车辆底盘高精度重建的问题,为车辆工程中底盘分析与优化提供参考,对提高车辆检测工程的自动化、智能化水平具有实际的工程意义。更多还原

【Abstract】 Efficient vehicle detection methods play a key role in ensuring the overall performance of vehicles and improving the level of national defense equipment. The characteristics of vehicle chassis is the key reference standard to measure the mobile performance of vehicles. In this paper, a high-precision, automatic and fast vehicle chassis contour reconstruction system is designed and implemented. Firstly, the remote control robot equipped with multi line lidar is used to scan the vehicle chassis contour, and then a chassis point cloud reconstruction method based on ICP algorithm and direct filter is proposed. Finally, the ground point cloud is removed based on random sampling consistency algorithm, and a de-noising method based on down sampling and statistical filter is proposed. The experimental results show that: the average Euclidean distance of the point cloud reconstruction based on the robot moving speed method is 11.62 mm, and the point cloud appears deformation,while the average Euclidean distance of the point cloud reconstruction based on the ICP algorithm combined with the direct filter method proposed in this paper is 4.43 mm, which is reduced by 61.8%, the fusion compactness is higher, and the point cloud is invisible. It can better meet the problem of high-precision reconstruction of vehicle chassis, provide reference for chassis analysis and optimization in vehicle engineering, and has practical engineering significance to improve the automation and intelligent level of vehicle detection engineering.更多还原