基于改进端元提纯模型的MODIS森林类型识别研究

Spectral Unmixing of MODIS Data Based on Improved Endmember Purification Model for Forest Type Identification

MODIS遥感数据具有很高的光谱辐射精度,以及成本低、覆盖面积广、获取容易、周期短等数据特征,可以实现全覆盖大尺度区域森林类型信息快速提取,但由于其空间分辨率较低,遥感数据中存在混合像元。利用混合像元分解模型进行分解可得到较好的分类结果,但混合像元分解的端元组分直接影响分类的精度。利用决策树分类模型改进端元提纯,分析各地物的MODIS时间序列植被指数变化规律及物候变化规律,利用决策树模型分类的结果进行端元组分的提纯,最后进行混合像元分解。研究结果表明:分类精度最高的是线性混合像元分解,其次是最大似然分类,最差的是非线性混合像元分解,其中带约束和不带约束的线性分解模型的精度相差不大。

Because of high spectral resolutions,large coverage and low cost MODIS(Moderate Resolution Imaging Spectroradiometer)data has been widely used to quickly extract information of forest types at re-gional,national and global scales.However,its coarse spatial resolution often leads to mixed pixels and low classification accuracy of forest types.Using spectral unmixing model can,to some extent,increase the accuracy of classification.But how to accurately extract pure endmembers for a study area is often a great challenge.The selection of linear or non-linear spectral unmixing algorithm is another challenge.In this study,a method to extract endmembers from MODIS images was developed.In this method,the time series of MODIS derived vegetation index was first derived and the phenological variation of forest types was ana-lyzed.Decision tree classification was then conducted and the obtained results were used to extract end-members.In addition,for comparison,the classification was also made using a widely used classifier-maximum likelihood.It showed that linear spectral unmixing was the best regardless of with and without constraints,the maximum likelihood classification and non-linear spectral unmixing were in the 2nd and 3rd places respectively.