基于植被物候特征与监督分类的青南高原信息提取

针对大尺度区域的植被信息提取，由于范围广阔、地形复杂、气候迥异，分类精度的提高是个亟待解决的问题；通过对青南高原采用分区处理,利用植被指数的特性，将基于时间序列的NDVI数据所反映的植被物候知识，辅助信息DEM和GIS数据加入监督分类系统，进行植被信息提取，并进行了分类精度评价。研究结果表明，利用该方法对青南高原的3个地区分类后，其分类精度都达到了83.3%以上，达到了较好的分类结果。在监督分类的训练区选取过程中，将植被物候特征作为知识，结合目视解译和DEM辅助知识帮助选取训练区的方法，同时参考GIS土地利用数据，使得训练区的选取更准确可靠，可进一步提高分类精度。

Vegetation Information Extraction in the South Qinghai Plateau Using Phenology and Supervised Classification

Because of wide ranges,complicated terrains and disparate climates,it is an important problem to increase the classification accuracy for vegetation information extraction in wide ranges.In this paper,it uses division processing and joins vegetation phonological knowledge reflected by NDVI series data and auxiliary information including DEM and GIS data into the supervised classification system to extract vegetation of South Qinghai Plateau.The classification accuracy has reached more than 83.3% by using the method mentioned above and achieved better classification results.It is reliable to help select training areas,using vegetation phonological knowledge,visual interpretation and DEM data and taking land-use data into account.It makes training areas more accurate and improves the accuracy.