Scene Measurement Method Based on Fusion of Image Sequence and Improved LiDAR SLAM
Dongtai Liang,Donghui Li,Kui Yang,Wenxue Hu,Xuwen Chen,Zhangwei Chen
DOI: https://doi.org/10.3390/electronics13214250
IF: 2.9
2024-10-30
Electronics
Abstract:To address the issue that sparse point cloud maps constructed by SLAM cannot provide detailed information about measured objects, and image sequence-based measurement methods have problems with large data volume and cumulative errors, this paper proposes a scene measurement method that integrates image sequences with an improved LiDAR SLAM. By introducing plane features, the positioning accuracy of LiDAR SLAM is enhanced, and real-time odometry poses are generated. Simultaneously, the system captures image sequences of the measured object using synchronized cameras, and NeRF is used for 3D reconstruction. Time synchronization and data registration between the LiDAR and camera data frames with identical timestamps are achieved. Finally, the least squares method and ICP algorithm are employed to compute the scale factor s and transformation matrices R and t between different point clouds from LiDAR and NeRF reconstruction. Then, the precise measurement of the objects could be implemented. Experimental results demonstrate that this method significantly improves measurement accuracy, with an average error within 10 mm and 1°, providing a robust and reliable solution for scene measurement.
engineering, electrical & electronic,computer science, information systems,physics, applied