The methods research of deriving Bamboo information based on IKONOS Image

摘要

Bamboo is an important forest resource at sub-tropical regions in China, which continuous expansion has taken place in China during the last 20 years. Monitoring its distribution has important significance. It provide a possibility that we can extract accurate information among vegetation types as result of detailed texture features, patterns, optical information can be obtained from IKONOS image. In this paper, Anji Country in Zhejiang province was selected, deriving bamboo information from IKONOS image by using Iterative Self-Organizing Data Analysis Techniquc(ISODATA) classify method, Decision Tree method based on NDVI and texture and Object-oriented classification algorithm based on texture and spectral bands were discussed. The results showed that Decision Tree method based on NDVI and texture demonstrated high accuracy (Kappa=0.7428). Additional, Object-oriented classification algorithm based on texture and spectral bands suggested the highest user's accuracy 93% and producer's 78% at deriving bamboo information, which can be adapted to deriving bamboo information from IKONOS image at subtropical regions in China.