应用哨兵2A多时相遥感影像对树种的识别

森林树种类别的准确识别和专题图制作对于森林经营管理具有重要作用。以哨兵2A多时相遥感影像数据源,通过对单时相影像以及根据基尼系数(Gini系数)和信息熵进行的不同影像组合,利用随机森林算法对大孤家林场的树种类别进行识别和分类制图。结果表明:哨兵2A多时相影像组合能够显著改善树种分类的精度,多时相影像组合总体分度最高为87.45%,Kappa系数为0.825,比单时相遥感影像分类总体精度提升了4%~7%。

Tree Species Recognition with Sentinel-2A Multitemporal Remote Sensing Image

Accurate identification and thematic mapping of forest tree species are of great importance for forest management and scientific research.The Dagujia Forest Farm was used as the research area to identify the tree species and thematic maps by using the six Sentinel 2 remote sensing images.Random forest algorithm was used to classify the six temporal single-view images and different image combinations based on the Gini coefficient and information entropy.The sentinel 2 multi-temporal combination image can significantly improve the accuracy of tree species classification.The overall classification accuracy of the multi-temporal combination image is up to 87.45%and the Kappa coefficient is 0.825.Compared with the single-phase remote sensing image classification result,the overall accuracy is improved by 4%-7%.