Method for quickly building road network model of open-pit mine based on grid refinement

Constructing the road network in open-pit mining areas is a crucial step for intelligent scheduling and autonomous driving of trucks in these regions. However, given the intricate road conditions, extensive GPS trajectory data from mining vehicles, and numerous outliers, modeling the road network poses several challenges. To solve this problem, a method for quickly building road network model of open-pit mine based on grid refinement was proposed. Firstly, denoising method for grid data, utilizing an enhanced dilation algorithm, is proposed to clean the road network grid generated by binarizing trajectory points. The algorithm fills in gaps within low-connectivity grids, mitigating the effects of grid fragmentation. Subsequently, a model for extracting the road network's backbone is developed, leveraging the refined Zhang-Suen algorithm. This model identifies morphological features within the grid area and extracts a skeleton map with a consistent grid width, eliminating burrs and redundancies left by the original algorithm. Taking advantage of trajectory temporality, a backbone connection algorithm is then devised to pinpoint the navigable roads within the network. This resolves issues stemming from the inherent grid method's abnormal connectivity, enhancing overall road connectivity. Lastly, in accordance with practical application needs and the structure of the road network, a point-road-point road network model structure is proposed to significantly reduce the complexity and computational scale of the road network while ensuring that the logical structure of the road network remains unchanged. And use folium to visualize the road network. The experiment shows that the accuracy and completeness of the road network constructed by this method are 95.45% and 96.43%, respectively, and the program running time is 2.697 seconds, which meets the requirements of fast generation and high accuracy of open-pit mine road network models.