Multispectral remote sensing image road extraction method based on generative adversarial network

Abstract

The invention discloses a multispectral remote sensing image road network extraction method based on a generative adversarial network, and the method comprises the steps: labeling a multispectral remote sensing image, generating a label image, and carrying out the preprocessing of the image; then cutting and preprocessing the image; establishing a generative adversarial network through a deep learning model for training, and storing the model when the network converges; and finally, obtaining a final result image of the to-be-tested image through the generator model. Compared with the prior art, semantic segmentation is carried out on the remote sensing image in a generative adversarial network mode, pixel-by-pixel classification is carried out, and finally an extracted road result is obtained.

Description

Technical field [0001] The invention belongs to the technical field of image recognition, and particularly relates to a method for extracting a road network from a multispectral remote sensing image based on a generation countermeasure network. Background technique [0002] Traditional road data acquisition mainly comes from manual methods, such as real-time inspections and surveys by field collection teams. Although this method can obtain more accurate road information, it requires a lot of manpower and material resources and the update speed is slow. At the same time, it is accompanied by a certain degree of danger. [0003] In recent years, with the continuous increase and improvement of the number and technology of remote sensing satellites launched in the world, the resolution of remote sensing satellite images has also been greatly improved, so the use of high-resolution satellite remote sensing images has become a reality. At the same time, high-resolution satellite remote ...

AI Technical Summary

Problems solved by technology

However, the running speed of this method needs to be improved, and at the same time, it lacks the mining of contextual features

[0009] For the above methods, most of the methods need to manually select features, and the quality of feature selection will largely affect the extraction effect of roads.

Images