利用荧光高光谱图像技术无损检测猕猴桃糖度  

Nondestructive Determination of Sugar Content in Kiwifruit Using Hyperspectral Fluorescence Imaging Technology

作　　者：吴彦红[1] 严霖元[1] 吴瑞梅[1] 杨勇 

机构地区：[1]江西农业大学工学院,江西南昌330045 [2]深圳市方大装饰工程有限公司,广东深圳518057

出　　处：《江西农业大学学报》2010年第6期1297-1300,1306,共5页Acta Agriculturae Universitatis Jiangxiensis

基　　金：江西省自然科学基金资助项目(0630023);江西农业科技支撑计划资助项目(2009BNB05705)

摘　　要：将405nm激光照射到猕猴桃样品上,当激光透过样品内部时,部分单色光被样品内部成分吸收,释放出荧光,再用高光谱成像系统采集诱导出的荧光散射图像。在荧光散射图像上选取感兴趣荧光区域(ROIs),提取感兴趣区域在波长400～1000nm范围内的特征变量。当提取12个特征变量时,建立的猕猴桃糖度多元线性回归(MLR)模型的校正集相关系数Rc为0.932,预测均方根误差(RMSEC)为0.4764°Brix,预测集相关系数Rp为0.8227,预测均方根误差(RMSEP)为0.5645°Brix。研究结果表明,采用激光诱导荧光成像技术无损检测猕猴桃糖度是可行的。This paper attempts the feasibility to determine the sugar content in kiwifruit using hyperspectral fluorescence imaging technique. A continuous wave laser （405 nm） was used as an excitating source for inducing fluorescence in kiwifruits. Fluorescence scattering images were acquired from kiwifruit by a hyperspectral imaging system. Fluorescence scattering images were preprocessed,then the regions of interest （ROIs） of pixels were selected from the preprocessed images. Character variables were selected in the wavelength range of 400~1 000 nm attained by mean grey value of the ROIs. The MLR model was achieved with Rc=0.932 and RMSEC=0.476 4 °Brix in the calibration set,Rp=0.822 7 and RMSEP=0.564 5 °Brix in the prediction set when 12 character variables were selected.The study demonstrated that laser-induced fluorescence hyperspectral technology could be successfully used in determination of sugar content in kiwifruits.

关 键 词：高光谱图像 荧光 无损检测 猕猴桃 糖度 

分 类 号：S123[农业科学—农业基础科学] O433.2[机械工程—光学工程]