摘要
中国是世界重要的棉花产地,而棉花黄萎病是影响棉花产量的重要原因。棉花染病后轻则减产,重则绝收,因此对黄萎病的识别检测研究有重要意义。目前对棉花黄萎病的研究方法主要是通过遥感技术采集棉花图像,分析其高光谱特征;根据不同的特征,建立相应的光谱估测模型,进而对黄萎病进行识别检测。但由于高光谱分析仪器价格昂贵,难以普及化应用,因此市场应用价值不高。本文结合实际需求,利用机器学习方法建立了三种识别模型,极大提升了检测效率和精度,具有广泛的应用前景。
China is one of important cotton producing places accross the world,and cotton verticillium wilt is an important factor affecting cotton production.Cotton yield may be reduced after being infected,even cause total crop failure.Therefore,it is of great significance to identify and detect verticillium wilt.At present,research method of cotton verticillium wilt mainly is to collect cotton images with remote sensing technology,analyze its hyperspectral characteristics;establish corresponding spectral estimation model according to different characteristics,identify and detect verticillium wilt.But hyperspectral analysis instrument is not of market application value for expensive price and and difficulty of popular application.The paper establishes three recognition models according to actual needs with machine learning method,which improves detection efficiency and accuracy greatly,and has broad application prospect.
作者
陈梦媛
CHEN Meng-yuan(Electronic Information Engineering School,Anhui University,Hefei,Anhui 230601)
出处
《新型工业化》
2020年第2期136-139,共4页
The Journal of New Industrialization
关键词
机器
学习
检测模型
Machine
Learning
Detection model