论文部分内容阅读
树种分布是林业调查工作中的关键环节,遥感影像中对树种信息的提取具有重要意义。论文以福建省长汀县为研究区,收集2016年3月美国陆地卫星影像(Landsat 8 OLI)数据,提出基于知识库的图像膨胀分类方法,结合不同树种类型分布点的影像像元灰度与影像不同波段的重新排列,对杉类、松类、竹类与阔叶树类4个类型提取分布信息,结合地面验证坐标点进行精度评价,同时与专家知识分类法比较。结果显示图像膨胀法结合了地物分布特征与光谱信息,在Landsat 8 OLI影像中是可行的,并提高了精度,整体精度为83.01%,Kappa系数为0.77,与专家知识分类法相比分别提高了8.25%、0.11。该研究为快速精确地使用Landsat 8 OLI影像提取树种分布信息具有一定参考。
The distribution of tree species is the key link in forestry investigation, and the extraction of tree species information in remote sensing images is of great significance. The paper takes Changting County of Fujian Province as the research area, collects the data of Landsat 8 OLI in March 2016, and proposes a method of image blooming classification based on knowledge base. Combining with the gray value of image pixel of different tree species distribution points and Image rearrangement of different bands, the cedar, pine, bamboo and broadleaf tree four types of distribution information extraction, combined with the ground validation coordinate point accuracy evaluation, and expert knowledge classification compared. The results show that the image expansion method combines the distribution features and spectral information, which is feasible in Landsat 8 OLI images and improves the accuracy. The overall accuracy is 83.01% and the Kappa coefficient is 0.77, which increases respectively compared with the expert knowledge classification 8.25%, 0.11. This study may provide some reference for the rapid and accurate extraction of tree species distribution information using Landsat 8 OLI images.