Graph neural network-based greenway line selection method

The invention belongs to the technical field of greenway line selection, and relates to a greenway line selection method based on a graph neural network, which comprises the following steps of: 1, selecting and extracting points; step 2, constructing a node matrix required by the graph neural network based on selection of points; 3, constructing a two-layer GCN network, and achieving the precise solving of the greenway; the method has the advantages that various data of traditional resistance factors are utilized, subjective factors of pedestrians, such as preferences of the pedestrians to natural landscapes and recreation, are combined, selection of greenways is achieved through the graph neural network model, and after the model is fully utilized for training on a large-scale data set, the greenways can be selected through the graph neural network model. Complex geographic features, terrain and environmental conditions and pedestrian activity and landscape density can be learned, and high-precision greenway