多源遥感数据森林信息的提取和比较分析

Extraction and comparative analysis of forest information of multi-source remote sensing data

基于遥感的森林信息提取的信息源比较多,这些信息源由于本身参数的差异,其有效提取结果并不相同。以湖南省株洲市攸县黄丰桥林场为研究对象,利用SPOT5多光谱数据、TM和CBERS02星多光谱数据,分别开展森林信息提取,并对结果进行比较;同时还开展了决策级融合比较。研究结论如下:SPOT5多光谱波段提取针叶林信息最好,精度达到了84.00%;利用TM提取阔叶林信息效果最好,精度达到了85.42%;决策级融合后各地类分类精度都有所提高,尤其灌木林分类精度达76.92%;决策级融合后总体精度达85.08%,Kappa系数达0.813 2,基本达到SPOT5数据2.5 m(87.29%,0.838 5)的分类准确度。

Numerous remote sensing sources can be used on forest information extraction, among them there are lots of differences in parameters, so the extraction results are different greatly. The forest information of Huangfengqiao forest farm in Zhuzhou city, Hunan province has been extracted, analyzed and compared by using the remote sensing data such as SPOT5 multi-spectrum, TM and CBERS02 multi-spectrum. At the same time, the decision level data fusion was carried out, then the precision of four images were compared. The results show that the effects of SPOT5 multi-spectral band to extract coniferous forest information was the best, the precision reached 84.00%; after the decision level data fusion had been made, the classification precision of every land types improved, especially shrubbery＇ s went up to 76.92%, the overall accuracy of decision level fusion reached 85.08%, and the Kappa coefficient 0.8132, which basically reached the precision （87.29%） of SPOT5 with 2.5 meter resolution.