摘要
针对传统作物健康监测效率低与精度不足的问题构建基于深度学习的农业无人机作物健康监测系统,采用搭载多光谱相机的六旋翼无人机获取小麦冠层影像,利用改进ResNet-50模型进行病虫害自动识别与健康分级并通过关联规则挖掘技术分析作物健康时空分布规律。以华北平原冬小麦拔节期—灌浆期为试验对象,系统对叶部病害识别准确率达92.3%,健康指数分级精度达89.7%,单架次监测效率较人工巡检提升15倍。数据挖掘结果揭示白粉病发生与土壤湿度呈0.76正相关与条锈病与日均温差呈0.68负相关的规律为精准植保决策提供数据支撑。
Aiming at the problems of low efficiency and insufficient accuracy of traditional crop health monitoring,an agricultural UAV crop health monitoring system based on deep learning was built.A six-rotor UAV equipped with a multi-spectral camera was used to acquire wheat canopy images,and the improved ResNet-50 model was used to carry out Automatic identification and health classification of pests and diseases,and the spatio-temporal distribution of crop health were analyzed through association rule mining technology.Taking the jointing and filling stages of winter wheat in the North China Plain as the test object,the system's accuracy in identifying leaf diseases reached 92.3%,and the health index grading accuracy reached 89.7%.The efficiency of single inspection was 15 times higher than that of manual inspection.Data mining results reveal that the occurrence of powdery mildew has a positive correlation of 0.76 with soil moisture,and the occurrence of stripe rust has a negative correlation of 0.68 with the daily average temperature difference to provide data support for accurate plant protection decisions.
作者
邓丽霞
李伟光
Deng Lixia;Li Weiguang
出处
《智慧农业导刊》
2026年第4期9-12,共4页
JOURNAL OF SMART AGRICULTURE
关键词
深度学习
农业无人机
作物健康监测
数据挖掘
精准农业
deep learning
agricultural UAV
crop health monitoring
data mining
precision agriculture
