Method Research Of Alteration Extraction In Vegetation Cover Area Based On Sentinel-2 Date

A small number of alteration information extraction methods based on vegetation cover mostly rely on prior knowledge,that is measured spectral data or geochemical data.In the absence of these data,mask method or mixed pixel decomposition method is often used to remove or suppress interference factors such as vegetation.While removing vegetation,the mask method often masking the alteration information that may exist on the boundary of high vegetation or in the interior of low and middle vegetation.The partially constrained least squares mixed pixel decomposition usually limits the sum of endmember abundances to 1,and does not limit the range of endmember abundances(0~1),so this method produces negative values with no practical physical meaning in the process of mixed pixel decomposition,which affects the accuracy of alteration extraction.therefore,the method of fully constrained least square mixed pixel decomposition is introduced in this paper.Strictly limit the above two conditions.At the same time,as a new type of data source,there are relatively few studies on the extraction of geological information from Sentinel-2 data.Therefore,taking the vegetation cover area of Huanggangliang in Inner Mongolia as an example,this paper apply mask method and fully constrained least square mixed pixel decomposition method in Seentinel-2 date to extract iron-stained alteration information in this area.Based on the characteristics of lithological ore control in this area,combined with the known metallogenic(mineralization)points,limonitization points and structural information,the effects of iron-stained alteration extraction by the two methods are comprehensively evaluated and analyzed,and the following results are obtained:(1)According to the characteristics of structural ore control in the study area,this paper uses SPOT7 data to enhance and interpret the tectonic information of the study area through three-dimensional geomorphological enhancement and different filtering methods,which lays a foundation for the comprehensive analysis of subsequent alteration information.(2)A fully constrained least squares algorithm is introduced to limit the sum of endmember abundances to 1,and the range of endmember abundances is also limited.From the qualitative and quantitative point of view,the partial constraints and fully constrained least squares algorithms are compared and analyzed.The results show that the fully constrained least squares algorithm has higher accuracy.(3)Using mask method and fully constrained least square mixed image method to remove vegetation and other interference factors,and according to the spectral characteristics of iron-dyed altered minerals,the specific principal component analysis method is used to extract iron-stained alteration information.Finally,the C-A fractal method is used to classify the alteration anomalies.(4)The information of superimposed structures,known metallogenic(mineralization)points,limonitization points and iron-stained anomalies in the study area are synthesized.The anomalous information extracted by masking method and complete constrained least squares mixed pixel decomposition is compared and analyzed in whole and part,as well as its distribution relationship with structure,metallogenic(mineralization)point and limonitization point.The results show that the information of iron-stained alteration extracted by mixed pixel decomposition method is not only more closely related to structure and metallogenic(mineralization)points,but also more consistent with the known location of limonitization points.At the same time,the fully constrained least squares mixed pixel decomposition method can better extract alteration information in low and middle vegetation coverage areas and on the boundary of high vegetation coverage areas.