Object-oriented multiple scale mountainous city land coverage information obtaining method

Abstract

The invention discloses an object-oriented multiple scale mountainous city land coverage information obtaining method, comprising the steps of: S101, establishing a classification system in advance; S102, performing multiple scale image segmentation on a shot remote sensing image; S103, comparing images after each time of segmentation to determine a best segmentation scale image, wherein the image after each time of segmentation is composed of a plurality of pattern spots of different shapes; S104, establishing the fuzzy classification standard of each land type, and correspondingly classifying the pattern spots in the best segmentation scale image meeting the fuzzy classification standard to corresponding land types; S105, establishing a nearest classification standard, and performing nearest classification on the pattern spots not classified in the best segmentation scale image so as to classify the pattern spots not classified to a land type with nearest characteristics, thereby realizing image object complete classification; and S106, generating classified images, merging classifications and generating classified images.

Description

technical field [0001] The invention relates to the field of image processing, in particular to an object-oriented method for acquiring land cover information of multi-scale mountainous cities from remote sensing images. Background technique [0002] With the development of remote sensing satellite technology, remote sensing images with high spatial resolution are widely used in the extraction of urban land change / land cover information due to their clearer details and richer information. The phenomena of the same spectrum and different objects and the same object and different spectrum are serious in mountainous cities, with various types of land use, rich textures, intricate spatial relationships between land types, and difficult to extract land cover information. Traditional remote sensing image classification methods mainly use the spectral characteristics of pixels, and seldom use the texture and geometric characteristics of ground objects. The classification effect o...

AI Technical Summary

Problems solved by technology

Traditional remote sensing image classification methods mainly use the spectral characteristics of pixels, and seldom use the texture and geometric characteristics of ground objects. The classification effect of medium and low resolution remote sensing images is better, while the spectral characteristics of high spatial resolution image pixels are relatively single. The same kind of ground objects present great spectral heterogeneity, and the spectral features of different ground objects overlap with each other, and there are different objects of the same spectrum and different spectra of the same object. The traditional pixel-based remote sensing image classification method has been difficult to obtain satisfactory results.

Images