A method and system for extracting cities and towns from remote sensing images with ten-meter-level spatial resolution

Abstract

The invention discloses a ten-meter spatial resolution remote sensing image town extraction method and a ten-meter spatial resolution remote sensing image town extraction system. The extraction methodcomprises the following steps: collecting a ten-meter-level spatial resolution optical remote sensing image of a summer time phase and night light data of an urban area; establishing a town and non-town multi-size remote sensing image training sample set according to the optical remote sensing images; training an urban scene image classifier according to the deep learning features to obtain a training classifier; calculating an optimal threshold value of the night light data, wherein the optimal threshold value is a segmentation point numerical value used for segmenting a town region type anda non-town region type in the night light data; carrying out urban scene classification of the multi-scale grid by adopting a training classifier in a space range constrained by the initial urban region distribution mask map to obtain an urban classification result; and performing post-processing operation on the town classification result to obtain a town boundary extraction result. The distribution range of urban and non-urban region types is obtained by adopting the optimal threshold to segment the night light data, and the accuracy of urban boundary extraction results is improved by usingurban scene classification of a multi-scale grid.

Description

technical field [0001] The invention relates to the field of remote sensing, in particular to a method and system for extracting cities and towns from remote sensing images with ten-meter-level spatial resolution. Background technique [0002] In recent years, with the rapid development of our country's economy, urban agglomerations have become the main places where the spatial redistribution and transfer of production factors under the new situation of economic globalization take place. Urban agglomerations are the product of industrial spatial integration. occupies a central position. In traditional urban geography research, the definition and definition of urban agglomerations mainly depend on socioeconomic statistics. There are high requirements for the completeness and accuracy of each indicator of the statistical data, and it cannot directly reflect the spatial relationship between towns within the urban agglomeration, the spatial form and spatial pattern of the urban...

AI Technical Summary

Problems solved by technology

Traditional methods usually use the multi-spectral features of remote sensing images to classify urban land use. However, due to the electromagnetic waves radiated by ground objects, they must pass through the atmosphere to reach the sensor. The absorption and scattering of the atmosphere will affect the electromagnetic waves of the target objects. During classification processing, due to the difference in atmospheric composition and humidity at different times, the scattering effect is also different. Due to the difference in lighting conditions during imaging, the amount of electromagnetic radiation of the same ground object is different for images of different scenes, and the coverage types of the underlying surface are various. The spatial resolution of the sensor is limited, and the gray value in the remote sensing image cannot fully reflect the characteristics of the electromagnetic wave radiated by the target object

On the other hand, under the ten-meter-level spatial resolution imaging conditions, the interior of a town is a complex of various types of ground features including buildings, roads, grasslands, and water bodies. Therefore, only the spectral features of remote sensing images are used to classify towns. Classification, the accuracy of classification results is low

Images