Research and Application of Urban Road Extraction from Remote Sensing Images Based on Convolutional neural network U-net
Xuechun Feng, Ziwen Zhang · 2023
Abstract: The urban road network is the backbone of a city, and the development speed of a city largely depends on whether the planning of the urban road network is reasonable. How to accurately obtain road distribution has profound significance for urban road planning in China, and the remote sensing image classification method for extracting urban road information utilization is a very effective method. Aiming at the problems of low recognition accuracy and weak generalization ability of traditional methods for urban roads, this paper proposes the research and application of urban road extraction methods from remote sensing images based on Convolutional neural network U-net The maximum likelihood method and the Convolutional neural network U-net method based on the depth learning method are used to extract urban roads from the experimental data. By evaluating the accuracy of urban road extraction results, it is concluded that the U-net method based on Convolutional neural network is reliable and superior for urban road extraction from remote sensing images.