利用高光谱图像技术鉴别庐山云雾茶产地  

Identification of Geographical Origins of Lushan Mist Tea by Hyper-spectral Imaging Technology

作　　者：艾施荣[1,2] 吴瑞梅[3] 吴彦红[3] 严霖元[3] 

机构地区：[1]同济大学软件学院,上海200096 [2]江西农业大学软件学院,江西南昌330045 [3]江西农业大学工学院,江西南昌330045

出　　处：《江西农业大学学报》2014年第2期428-433,共6页Acta Agriculturae Universitatis Jiangxiensis

基　　金：江西省科技计划项目(20112BBF60019);江西省教育厅科学基金项目(GJJ11081)

摘　　要：提出了基于高光谱图像技术的庐山云雾茶产地鉴别方法。利用高光谱成像系统采集地理标志庐山云雾茶和广西、四川、福建3个其他产地云雾茶的高光谱数据,采用主成分分析法从原始高光谱数据块中选取3个特征波长:792.20,831.47,870.97 nm,分别提取每个特征波长下的灰度图像的纹理特征,利用BP神经网络方法建立庐山云雾茶产地鉴别模型。模型训练时的回判识别率为97.25%,预测时的识别率为95%,说明利用高光谱图像技术追溯庐山云雾茶产地可行。A rapid method was developed for discrimination of the geographical origins of Lushan mist tea by hyper-spectral imaging technique .The mist tea with Lushan geographic signs and mist tea from other places （ Guangxi ,Sichuan and Fujian ） were investigated .A hyper-spectral imaging system was used to collect hyper-spectral image data of the tea .Three feature wavelengths namely ,792.20,831.47 and 870.97 nm were opti-mized by principal component analysis method from the raw hyper-spectral images .The texture features were extracted from each gray image of each characteristic wavelength .BP-ANN model was used to develop a dis-crimination model.The correct discrimination rate of the model was 97.5%in the training set,and the discrimi-nation rate of model was 95%in the prediction set .The overall results sufficiently demonstrated that hyper-spectral imaging technology could be used to identify rapidly the geographical origins of mist tea .

关 键 词：高光谱图像 BP神经网络 主成分分析 庐山云雾茶 产地鉴别 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]