Classifying Forest Vegetation Using Sub-region Classification Based on Multi-temporal Remote Sensing Images
基于分区和多时相遥感数据的山区植被分类研究

It is very difficult to classify forest vegetation in mountain areas because of the impact of complex terrain.In this paper a new method,sub-region classification based on multi-temporal remote sensing images,is proposed to deal with the classification of forest vegetation.Firstly,sunshiny and shadowy region was classified using terrain factors and reflectance data.This technology could avoid the problem of "different spectrum with the same feature" and "different feature with the same spectrum" in some region.Secondly,the forest vegetation could get better classification precision by avoiding the interactions of different plants with multi-temporal images.So it was enhanced that the separability of coniferous forest and broadleaf forest.Finally,the classification result showed that accuracy could be greatly improved by using sub-region classification based on multi-temporal remote sensing images.The overall accuracy and kappa coefficient was 81.3% and 0.72,respectively.So the method delivered in this essay has obviously technological advantages and important application potentiality in forest vegetation classification.