Tree Species Classification Based on the Analysis of Hyperspectral Remote Sensing Data

摘要

Current tree species classification algorithms operate on a stand-wise level and therefore high-resolution satellite data is sufficient. Local forest inventories at a small-scale level, as well as detailed biodiversity monitoring, need single tree based classification to meet their demands on granularity. The optimal choice of multispectral bands can highly improve classification results. In order to find suitable bands and to evaluate their impact on tree species classification a set of images, satellite data and hyperspectral data is analyzed and the result is tested in a simple decision tree approach to species classification