Remote Sensing Lithologic Classification of Multispectral Data Based on the Vegetation Inhibition Method in the Vegetation Coverage Area

It is the top priority for rock mapping in the vegetation coverage area to eliminate the vegetation interference effect since the growth of vegetation limits the application of remote sensing in geology. Taking Dong Ujimqin Banner of Inner Mongolia as the study area, this paper compares vegetation inhibition method and maximum likelihood method in lithologic classification. Firstly, ASTER (advanced spaceborne thermal emission and reflection radiometer) data are chosen for vegetation index calculation with the soil factor and the vegetation index without the soil factor for principal component analysis respectively in the study area. Then, the principal component which shows the vegetation information is suppressed for lithologic classification. Furthermore, a comparative analysis is conducted and the lithology classification performance of the two methods is evaluated. It is found that the overall classification precision of the vegetation inhibition method reaches 82.946 8%, while that of the maximum likelihood classification reaches 76.364 3%. It shows that it is feasible to use the vegetation index to suppress the vegetation information in the vegetation coverage area. Compared with the conventional classification method of maximum likelihood method, the vegetation inhibition method greatly improves the accuracy of interpretation.