Method for extracting altered mineral at vegetation-covered areas by hyperspectral remote sensing

Abstract

The invention relates to a method for extracting altered mineral at vegetation-covered areas by hyperspectral remote sensing. The method comprises the following steps: a detection gun is directed at leaves of broadleaved plants to directly measure reflectance spectrum of leaves, measures reflectance spectrum of herbaceous plant canopy and measure reflectance spectrum of fresh surface of soil; uncalibrated and steam-influeced waveband removal, absolute radiation value conversion, bad wire restoration and atmospheric correction are carried out on Hyperion data; two principal components which have large absolute values but are opposite in sign are found according to principal component analysis characteristic value and a 2-D point diagram is made according to the two principal components; and abnormity is delineated according to the point diagram. Through the comparison of spectral feature fitting method and mineral extraction model, multilayer information separation of background, interference and abnormal information is adopted, and the key is to select the best waveband for spatial feature analysis. Through analysis of hyperspectral data characteristics and vegetation-covered area geographical features and the best waveband selection used during spatial feature optimization, vegetation information is better inhibited, and alteration information characteristics are enhanced.

Description

technical field [0001] The invention relates to an application of remote sensing images in geology, in particular to a method for extracting altered minerals using hyperspectral remote sensing images in vegetation coverage areas. Background technique [0002] Hyperspectral remote sensing has a very continuous band, and the spectral resolution reaches nanometer level. Different types of minerals can be identified through the small spectral differences between different minerals. At present, the qualitative identification of hyperspectral remote sensing minerals is based on spectral similarity and band decorrelation. The hourglass model has been well validated in rocky outcrops. However, in vegetation-covered areas, rocks and minerals appear as weak information and small targets on remote sensing images, which are difficult to identify. How to use remote sensing data to extract these weak information is a difficult problem in modern remote sensing geological application techno...

AI Technical Summary

Problems solved by technology

In the vegetation coverage area, the extracted alteration anomaly contains vegetation information which cannot be effectively removed

Using hyperspectral data in vegetation areas, the extraction effect is relatively poor

[0010] To sum up, the existing methods and models mostly use experience and experiments to select the band, which is highly arbitrary and lacks the basis for band selection reference, and in this way, the vegetation spectrum cannot be well separated from the highly reflective vegetation background without completely suppressing the vegetation spectrum. weak information

Images