高分解能衛星画像を用いたディシジョンツリー法に基づく樹種分類法の開発

抄録

A forest species classification method based on a decision tree method which used high resolution satellite imagery is proposed in this study. The proposed method consists of two parts. In the first part the extraction of forest areas is accomplished by the maximum likelihood method. In the second part 10 kinds of forest species are classified in the extracted forest areas using the decision tree method and IKONOS pan-sharpen multi-spectral data. In the decision tree method the parameters to divide two branches are obtained from the spectral analysis or the texture analysis of the IKONOS data for each forest species.
The study was carried out for a mountain area located in the Yamato and Minami district, Gifu Prefecture, in Japan. The results of forest species discrimination were evaluated by using independent field survey data. It is concluded that the comprehensive extraction accuracy in all forest species is about 67% and the proposed method is applicable to the forest monitoring and the planning as a useful tool.