Multi-scale map generation method, system and terminal based on generative confrontation network

Abstract

The invention belongs to the technical field of map drawing, and discloses a multi-scale map generation method, system, and terminal based on a generative confrontation network, which collect map tiles of various levels and the highest-level remote sensing image tiles with continuous scales about a specific area Data, and build a training set; build a map generation model, use the constructed training set to train the map generation model, and input high-level remote sensing images of the target area into the trained map generation model to obtain a multi-scale map with continuous scales. The invention adopts GAN to carry out cartographic synthesis, improves the quality of map generation, and obtains a multi-scale online map covering a certain scale interval. It is verified by experiments that the multi-scale maps generated by the method proposed in the present invention are better than the results of directly and simply applying GAN to generate maps of each level.

Description

technical field [0001] The invention belongs to the technical field of map drawing, and in particular relates to a multi-scale map generation method, system and terminal based on a generative confrontation network. Background technique [0002] At present, with the development of science and technology, the demand for online maps in people's daily life, scientific research activities, construction deployment, etc. is increasing day by day. Traditional mapping methods rely more on manual operations, and their efficiency may not be able to adapt to rapid urban construction and other geographical changes. Therefore, there is a need for a fast and cheap method that can generate online maps in real time (including multiple levels to meet observation needs at different scales), fill incomplete maps in time, and alleviate the conflict between traditional methods and urgent tasks. [0003] In response to this idea, CycleGAN and pix2pix have made an attempt to directly generate onli...

AI Technical Summary

Problems solved by technology

[0004] Through the above analysis, the problems and defects of the existing technology are: the quality of the existing multi-scale maps generated is not good, which is more obvious in the low-level, and there are problems such as blurred edges, road interruptions, and loss of details.

The difficulties in solving the above problems and defects are as follows: 1. The distribution of features between the remote sensing image domain and the map domain is very different, which makes it difficult to convert information between the two domains, and the model is difficult to fit; 2. According to the cartographic synthesis rules , low-level map characteristics and low-resolution characteristics of low-level remote sensing images, resulting in low-level image-to-map conversion process, the signal-to-noise ratio of effective information in remote sensing images is very low, increasing the difficulty of direct conversion

Images