基于无人机遥感和机器学习的车前种植面积信息提取研究OA

To construct a high-precision method for identifying the planting area of Plantago asiatica L.based on UAV remote sensing technology,realize the accurate extraction of Plantago asiatica planting areas and provide technical support for the dynamic monitoring of Chinese medicinal materials planting information and the planting area monitoring of herba-ceous Chinese medicinal materials.Taking the main producing area of Plantago asiatica L.in Tangzhou town,Taihe county,Jiangxi province as the research area,based on UAV re-mote sensing data and field survey classification sample data,five supervised classification al-gorithms,including Support Vector Machine(SVM),Random Forest(RF),Maximum Likelihood(ML),Minimum Distance(MD)and Mahalanobis Distance(ME),were adopted to classify and extract the planting areas of Plantago asiatica L.A confusion matrix was constructed to systematically evaluate the overall accuracy,Kappa coefficient,producer ac-curacy and user accuracy of each algorithm.The results showed that the classification accu-racy of the five algorithms ranked in the order of SVM>ML>RF>ME>MD.Among them,SVM achieved the optimal performance with an overall classification accuracy of 95.57%and a Kappa coefficient of 0.938 3,which had high classification accuracy.The SVM classification method based on UAV remote sensing presents obvious superiority in the identification of Plantago asiatica L.planting areas,which can meet production and practical needs,has ap-plication potential in the dynamic monitoring of Chinese medicinal materials planting infor-mation.It provides a scientific technical approach and methodological support for the monitoring of planting areas of Plantago asiatica L.and other herbaceous Chinese medicinal materials.