A road extraction method for intelligent dispatching based on MD-LinkNeSt network in open-pit mine

Authors:
Publisher website
Google Scholar
Abstract
Road extraction in open-pit mine is one of the important tasks of intelligent dispatching and manless driving. However, traditional methods for extracting open-pit roads cannot capture detailed information about roads, such as shape and surrounding environment. An MD-LinkNeSt (Multi Dilation-LinkNeSt) network is proposed to extract the open-pit road. First, the classification of UAV remote sensing image in mining area was completed based on mine road features. Second, ResNeSt and LinkNet are used for the encoding and decoding of MD-LinkNeSt network. The MD module includes encoding and decoding to form MD-LinkNeSt network.MD-LinkNeSt, a network framework for road extraction from open-pit mines, is constructed. Third, to enhance the extraction effect of MD-LinkNeSt network on open-pit road, the road loss function is added to MD-LinkNeSt network. Compared with the networks of Unet, DUnet, LinkNet and D-LinkNet, the network in this paper has higher IOU (Intersection over Union) and better continuity. Finally, the road network model is generated by using the open-pit mine road extraction method. The results show that this method can not only extract the road details better than the traditional method, but also meet the intelligent extraction accuracy requirements of the main road network in the open-pit mine.
Funding Information
  • National Natural Science Foundation of China (Grant No. 51774228)
  • National Natural Science Foundation of China (Grant No. 52074205)
  • Distinguished Young Scholars (Grant No. 2020JC-44)
  • Shaanxi Provincial Natural Science Basic Research Program (Grant No. 2019JLP-16)

This publication has 16 references indexed in Scilit: