A Scene Reconstruction Method of Airborne LiDAR

This article builds a simple, low-cost airborne LiDAR system, collecting and processing the actual data of large-scale outdoor scenes. Aiming at the actual characteristics of data, a Generalized-ICP registration algorithm based on ground segmentation is proposed to calculate the precise transformation relationship between point clouds. On this basis, this paper proposes a reconstruction strategy of group fusion. First, the original frames are registered and merged into groups of point cloud blocks with relatively uniform resolution, which overcomes the impact of uneven resolution on the global registration. In this process this paper proposes to use Mahalanobis distance to detect abnormal matching and proposes a correction method based on track fitting to adjust mismatched matrixes. For the global fusion of formed blocks, a sliding window method is proposed to obtain direct relationships from blocks to the global coordinate system. Frame fusions and block fusions are executed alternately and serially until the complete reconstruction result is achieved. Experiments show that the registration algorithm in this paper has high accuracy and practicability, and the reconstruction method can restore real, dense and accurate scene information.