论文部分内容阅读
本文利用2002年陕西省ETM卫星影像及相关植被地理信息,在其影像上选取覆盖全省包含不同地物类型的85个地面样地(Ground Truth),通过2003年陕西省MODIS卫星影像获取所选样地全年的NDVI时间序列曲线。采用决策树分类方法,结合NDVI时间序列曲线,实现基于时相和波谱信息的植被分类。最后通过混淆矩阵与Kappa系数等方法对分类结果进行正确率评价,结果表明,文中所给方法优于传统分类方法,所得结果与其他调查结果相一致。
Based on the ETM satellite images and related vegetation geographic information of Shaanxi Province in 2002, this paper selected 85 Ground Truths that cover different province types in the province and selected the MODIS satellite images of Shaanxi Province in 2003 NDVI time series curve of the whole year in the plot. Using the decision tree classification method and NDVI time series curve, vegetation classification based on time-phase and spectral information is realized. Finally, the correctness of the classification results was evaluated by confusion matrix and Kappa coefficient. The results show that the proposed method is superior to the traditional classification methods, and the results are consistent with other findings.