Automatic extraction of road network in open-pit mine based on GPS data

The traditional grid method for extracting road network has a low accuracy,especially for the extraction of open-pit road network,the loss and offset of road is more significant. For solving this problem,the conventional solution is to enlarge the grid so that the connectivity can be ensured. However,in this paper,by assuming the GPS bias as a normal distribution,a method of rastering the GPS data by calculating the probability of track points on the road was proposed. On this basis, the median filter algorithm was deployed to preprocess the raster image. The index table thinning algorithm was improved,and this improved thinning algorithm was used to refine the grid image of the road network. Finally,the road network was translated into vectorization. Experimental results show that the coverage ratio of this method has been improved by 6. 43% to 11. 54% compared with the traditional grid method,and the error rate has been reduced by 42. 13% to 83. 02% . This paper provides an effective method for road network extraction and re- veals the important influence of grid size on road network extraction results.