Research on Object Oriented Algorithm for Road Extraction in High-Resolution Remote Sensing Image

In view of the characteristics of high-resolution remote sensing images, an automatic road extraction method based on object-oriented thought was proposed. Firstly, the remote sensing image is bilaterally filtered to smooth the details and retain the road edge information. Then, the image is segmented by the FCM algorithm to obtain independent ground objects, and the candidate road segments are obtained by filtering each object according to the geometric features of the road. The regional growth algorithm is used to form the road network, and finally the morphology method is used to finish and refine the road network. Experiments show that this method can effectively extract road targets from remote sensing images of different scenes without manually selecting road seed points.