抄録
This research aims at the development of the mineral identification technique to inquire into the resource using hyper spectral data in arid or semiarid regions. We designed the method by two steps where the minerals were identified from the spectrum. At the first step, to identify the absorption position from the spectrum, we use first derivation spectrums. However, the derivation method emphasizes a noise in the spectrums. To solve this problem, we designed the improved one i.e. giving a median filter to the differentiated spectrum. This method has two advantages, i.e. a rigorous atmospheric correction is not required, and identification is unaffected from the noise of the data. Once the algorithm detects the absorption, we applied least squares fitting of a quadratic curve to around wavelength region of it, for accurate identification of the absorption position.
The next step is identification and quantification of the minerals, using the position and the depth of the absorption features. The content of the minerals is calculated by comparing the content turned out in the identification score and the results of in situ investigation. As a result, we developed the method by which 13 kinds of minerals were able to be identified. Especially, the calcite and the kaolinite were able to calculate the content on the ground.
