基于支持向量机的云烟87烟叶成熟度高光谱遥感识别  

Spectral Feature Extraction for Flue-cured Virginia Tobacco Leaves of Different Maturity Grades

作　　者：梁寅[1] 张云伟[1] 李军营[2] 

机构地区：[1]昆明理工大学现代农业工程学院,云南昆明650500 [2]云南省烟草农业科学研究院,云南昆明650500

出　　处：《西南农业学报》2013年第3期957-962,共6页Southwest China Journal of Agricultural Sciences

基　　金：国家自然科学基金项目(31060118)

摘　　要：利用烟叶光谱反射率数据研究烟叶的成熟度是将高光谱遥感探测技术应用于烟叶成熟采收的前提。对生理成熟期内云烟87的中部叶成熟度进行了研究,研究是从提取和分析烟叶高光谱信息的角度上进行的。基于提取到的反映烟叶成熟度的光谱特征和支持向量机这一基于数据的小样本学习方法,建立了识别烟叶成熟度的数学模型,这个数学模型是由支持向量机生成的2个最优决策超平面,一个将初熟类烟叶和适熟类烟叶分开,另一个将适熟类烟叶和过熟类烟叶分开,且分类精度均在90%以上。Using spectral reflectance data to research the maturity of tobacco leaf was the premise of the application of hyperspectral remote sensing detection technology on tobacco harvest.The perspectives of extraction and analysis of spectral information was researched on the maturity of tobacco in this paper.Based on the spectral feature and support vector machine,the mathematical model for recognizing the maturity of tobacco had been built.This model contained two optimal decision hyperplanes.One was for initial maturity category and proper maturity category,the other was for proper maturity category and over maturity category.The accuracy of the classifier was more than 90 %.

关 键 词：烟叶成熟度 光谱特征 支持向量机 最优决策超平面 分类精度 

分 类 号：S572[农业科学—烟草工业]