基于混合固态激光雷达的动态车辆三维模型重建

Dynamic Vehicle 3D Model Reconstruction Based on Hybrid Solid-state Lidar

目前常见的三维车辆模型重建算法,是通过机械旋转激光雷达进行即时扫描采样,获取的点数有限,且由于混合固态激光雷达独特的扫描特性,即时扫描采样的车辆点云会出现点迹杂乱、车辆模型形变等问题。故本文提出一种基于混合固态激光雷达的动态车辆三维模型重建算法,在常规三维车辆模型重建算法的基础上,对目标车辆进行目标追踪、车辆补偿,将车辆经过探测区域的所有激光点回溯到车辆的3D轮廓中,并使用3DReshaper软件进行车辆模型重建。同时,为了验证算法的可靠性,对试验车辆在不同车速下进行多次测试,验证得到示值相对误差均小于1%,重复性均小于3%,说明此算法测量准确高,且具有较好的稳定性。

At present, the common 3 D vehicle model reconstruction algorithm is real-time scanning and sampling through mechanical rotating lidar, and the number of points obtained is limited. Due to the unique scanning characteristics of hybrid solid-state lidar, the real-time scanning and sampling vehicle point cloud will have problems such as point trace clutter and vehicle model deformation. In this paper, a dynamic vehicle 3 D model reconstruction algorithm for hybrid solid-state lidar is proposed. Based on the conventional 3 D vehicle model reconstruction algorithm, the target vehicle is tracked, the vehicle is compensated, all the laser points of the vehicle passing through the detection area are traced back to the 3 D contour of the vehicle, and the vehicle model is reconstructed using 3 dshader software. At the same time, in order to verify the reliability of the algorithm, the test vehicle is tested for many times at different speeds. The results show that the relative error of the indication is less than 1% and the repeatability is less than 3%, indicating that the algorithm has high measurement accuracy and good stability.