Rolling Clustering Algorithm for Open-Pit Mine Road Network Extraction Based on Vehicles Trajectory

The construction of the road network is a very critical issue in the transportation industry. An accurate road network can effectively provide vital preconditions for logistics transportation, traffic diversion, vehicle scheduling, etc. At present, it is still a difficult problem to achieve rapid and accurate extraction of road networks in different transportation environments. In order to solve the problem of road network automatic extraction in open-pit mines, this paper proposes a Rolling Clustering Algorithm (RCA) based on truck GPS trajectory data. The algorithm combines the advantages of road intersection recognition and trajectory clustering, which improves the accuracy of road network extraction while ensuring the topology. First, the original data are preprocessed to eliminate the influence of noise points. Next, all trajectories are divided into road segments through the identification of road intersection nodes, and rolling clustering is performed to extract road skeletons. Finally, a complete road network is generated by connecting the segments and intersection nodes. This study evaluated RCA's performance by comparing it with several representative road inference algorithms. The results show that the proposed algorithm outperformed others in terms of precision and recall. This algorithm achieves the best extraction accuracy while ensuring the road network topology. In the final validation phase, the GPS trajectory data of open-pit mine trucks are adopted for practical application. The proposed framework based on GPS trajectory provides a new solution for the road network extraction problem.