The Extraction Of Alteration Information In Remote Sensing Under The Condition Of Spatial Constraints

The wall rock alteration is the reaction of wall rock and hydrothermal, whichmakes a series of changes about the material composition, construct, structure, it cannot only reflect the existence of ore body, but also instruct the location of ore body. Sothe wall rock alteration is an important criterion for prospecting of ore. The alterationrock has unique electromagnetic radiation, so remote sensing can conduct mineralresources investigation and ore-forming prediction by combining the spectralinformation of wall rock alteration with related geological background, at the sametime, it can provide gist for the prospecting target area.In the process of remote sensing imaging, the physical, chemical properties andbackground environment (terrain, irradiance conditions) of features determine thespectral information received by the sensor, the performance of the sensor (spatialresolution, spectral resolution, radiation resolution and temporal resolution)determines identifiable degree of remote sensing image. So terrain factor and pixelscale factor are important factors in the extraction of alteration information.The traditional method of alteration information extraction mainly considers thediagnostic spectral characteristics of alteration information itself, this paper not onlyconsider the spectrum characteristic, but also give full consideration to the spatialconstraint factor, including the terrain factor and the pixel size factor. From the aspectof terrain, Ratio Method, Principal Component Analysis (PCA), Spectral AngleMapping Method, Decomposition of Mixed Pixels have achieved desired results, buttheir applications have some limitations. A large number of experiments show that toa certain extent, Ratio Method can eliminate topographic influence, but the results ofalteration information extraction always appear false information; The alterationinformation extraction results of PCA is seriously influenced by topography whichweakens the information of shade slope, Spectral Angle Method is not sensitive totopographic influence, but may cause the results of low precision because of the wideband spectrum and little band. Terrain factor is one of the most important reasonswhich cause “the same object with different spectrum” or “different objects with thesame spectrum”. As a kind of weak information, the spectrum of alterationinformation is heavily affected by terrain. So the consideration of terrain factor is amust in the alteration information extraction. From the aspect of pixel size, a largenumber of experiments show that multispectral and hyper-spectral data can be a goodinstruction of the alteration wall rock’s diagnostic spectrum, but it cannot perfectlyidentify and express the alteration of wall rock because of the limitation of its spatial resolution. Therefore, as a factor of background parameter, pixel size factor must beconsidered during the extraction of alteration information.This paper studies the methods of alteration information extraction under thecondition of spatial constraints, and the main achievements are as follows.1.This paper improves the traditional experience statistics topographic correction model byusing the method of grade classification(According to the topography of study area, itis divided into6classes, as follows:<5°,5°to10°,10°to15°,15°to20°,20°to30°,>30°.), that is, regression relationship is established for each slope grade levelbetween effective solar incidence angle and the pixel brightness values of radiance,which obtains the corresponding parameter m and b. So it can improve correctionaccuracy under the condition of high spectral fidelity. The improved model is used inthe image of study area, then the terrain corrected image is used to extract alterationinformation by PCA. After validated, the extraction result is highly reliable, themethod has a common applicability.2.According to the relationship between the aspect and the solar azimuth, thestudy area is divided into four blocks of spectrum, as follows: Ⅰ (aspect range is135°to195°), Ⅱ (aspect range is75°to135°and195°to255°), Ⅲ (aspect range is15°to75°and255°to315°), Ⅳ (aspect range is315°to360°and0to15°). Each spectrumarea respectively uses principal component analysis (PCA) to extract alterationinformation. Then it sets up different alteration threshold in different spectral region.This method can effectively remove the influence of terrain factor without changingthe spectral information of altered mineral. After validated, the extraction result ishighly reliable, the method has a common applicability.3. This paper analyzes the feasibility and advantage of high resolution data(Quick Bird) in extracting the alteration information of iron mineralization from theaspects of data characteristics, and compares extraction results in the same study areaby ETM+data, ALOS data, Quick Bird data. The result shows that high resolutiondata Quick Bird can be used to extract the alteration information of ironmineralization, also it can identify and express alteration information distributing inthe small area, and locate alteration zone by conveniently combining with GPS, whichis beneficial to validate the alteration zone at fixed-point.This paper relies on the project of "The study and application of thermal infraredmultispectral/hyper-spectral mineral mapping and abnormal information extractionof remote sensing" and the project of "The remote sensing information extraction forregional geological survey".