A New Urban Road Detection Method in High-resolution Images Based on Bayesian Network

In view of the fact that road network detection effect in high-resolution image is not satisfying, an approach for road network detection based on Bayesian Network is put forward in this paper. First, under the guidance of existing GIS data, extract roads from remote-sensing images, and obtain most of the unchanged road edge information and suspected road edge information. Then, making use of the reasoning ability of Bayesian Network, collect strong evidence for identifying road network. Judge and make an inference from road edge information with the method of hypothetical test to extract road network in the remote-sensing image, and change information of the road network can also be obtained. * Corresponding author: XU Chuan, PhD candidate, majors in SAR image segmentation. Email: [email protected]