There are dead-end roads in the road network, and many of them have the function
of indicating specific target clues, which is of great significance in the
fields of military, urban construction, and disaster relief and rescue. However,
many of the important cut-offs are in mountainous or wilderness areas, and
surveying them is difficult and costly. The research objective of this project
is to extract the breakpoints in the road network using high-resolution Google
satellite imagery, so as to provide clues and indications for the subsequent
relevant work. Firstly, the image is corrected and pre-processed to highlight
road edge information.Then the phase grouping method is improved by setting a
double-angle threshold, filtering the edge operator to reduce the calculation
error of the gradient angle, and the road network is extracted by the improved
phase grouping method. And finally screens out the dead-end road points with the
eight-neighbourhood method, and marks them on the experimental image, so as to
realize the identification and positioning of the dead-end roads.The algorithm
has a certain degree of universality, after replacing the experimental area and
image, it can still locate the dead-end road breakpoints accurately, and the
universal accuracy rate reaches more than 80%, and after adjusting the threshold
value for a specific area, the extraction accuracy rate can reach more than 90%.
This project realizes the technology of identifying and locating the dead-end
road based on the phase grouping method, which can intelligently and
automatically identify and locate the dead-end roads on the remote sensing
image, and the improvement of the phase grouping method significantly improves
the extraction precision of the road network and the extraction precision of the
broken point, which is of practical value.