THE CLASSIFICATION AND IDENTIFICATION OF ANCIENT SITE AREA BASED ON SPECTRA DATABASE

Based on operational modular imaging spectrometer (OMIS-II) hyperspectral image data of Qinshihuang's mausoleum area, the profile spectral curve data of the typical ground objects and the ancient site are collected and purified. Then the ancient site spectra database can be established. Taking this spectral database as a reference, the methods of spectral angle mapping (SAM) and spectral information divergence (SID) are used to identify and classify the ground objects and ancient site in study area. Finally, the classification map of the ancient site area is obtained. The results show that the classification and identification of the ancient site area based on spectra database of hyperspectral images (that is, the reference end-members can be obtained directly from the images), is not only reliable in feature selection, but also has a good correlation with image data. The classification accuracy of the above methods is also better than the traditional methods of minimum distance 'classification (MDC) and maximum likelihood classification (MLC). For the fine classification of typical ground objects in the ancient site area, the methods used in this paper are feasible and have important practical significance for further extracting the precise information of ancient sites.