| Abstract |
The construction of smart cities demands high-quality point cloud data. On the one hand, SLAM technologies register the data based on time series, it is hard to optimize the consistency of point cloud globally, on the other hand, putting on a high-end IMU is sometimes difficult due to the high cost. Obtaining high-quality point clouds on low-cost IMUs is a key challenge. To improve the point cloud quality, this paper proposes a method by segmenting sub-maps, building loop constraints, and then solving the equations with point cloud matching constraints, IMU pre-integration constraints, GNSS PPK constraints, and LiDAR odometry constraints. This method optimizes point cloud and corrects the trajectory globally. Meanwhile, the multi-source data from a helmet-based mobile mapping system is used to verify the approach. The comparison and analysis results demonstrate the effectiveness of this method and its upcoming potential for wider applications. © 2024 Editorial Board of Medical Journal of Wuhan University. All rights reserved. |