Abstract: Terrestrial laser scanners, serving as real-time and high-precision three-dimensional reconstruction tools, play a significant role in various fields such as intelligent robotics and surveying. However, their high cost has limited their widespread application. Laser scanners developed with consumer-grade lidar and cameras offer a more affordable alternative for achieving three-dimensional scene reconstruction. Nevertheless, this approach encounters challenges like the calibration of relative sensor poses and the fusion of heterogeneous sensor data. In this study, the extrinsic calibration between the lidar and the motor and camera, is accomplished using a method based on point cloud edge line feature matching and another method that optimizes the intensity-gray normalized information distance. Additionally, post-processing is performed on the scanning results, and a point cloud coloring fusion algorithm based on pixel distance weighting is proposed to enhance the quality of the final scanned point cloud. The experimental findings indicate that the measurement accuracy of the developed terrestrial panoramic laser scanner can be maintained within the original accuracy range (1 cm-2 cm) of the utilized lidar, and the scanning results can fulfill the requirements for panoramic three-dimensional reconstruction in common scenarios.

Key words: LiDAR, camera, extrinsic calibration, laser ranging, sensor fusion, point cloud processing

摘要： 地面式激光扫描仪作为一种实时的高精度三维重建工具，在三维重建和智能机器人和测绘等领域有着重要应用，因其价格昂贵，难以广泛应用。使用消费级激光雷达和相机研制的激光扫描仪能够在成本相对较低的情况下同样实现场景的三维重建，面临传感器相对位姿标定和异构传感器数据融合等问题。使用基于点云边缘线特征匹配的方法和基于强度-灰度归一化信息距离优化的方法分别完成激光雷达与电机、相机之间的外参标定，同时对扫描结果进行了后处理并提出了一种基于像素距离加权的点云着色融合算法，改进最终的扫描点云质量。实验结果表明，研制的地面式全景激光扫描仪测量精度能够保持在所使用激光雷达的原测量精度范围内（1 cm~2 cm），扫描结果能够满足一般场景下的全景三维重建要求。

关键词: 激光雷达, 相机, 外参标定, 传感器融合, 点云处理