摘要
以酿酒葡萄赤霞珠果实为研究对象，利用高光谱成像技术检测葡萄果皮中的花色苷含量。采集60组样本的900～1700nm近红外波段高光谱图像，并用pH示差法测量样本果皮中花色苷含量。选取高光谱图像中葡萄果实区域作为感兴趣区域（ROI），计算其平均光谱，并采用sG平滑、归一化、多元散射校正等预处理方法提高光谱的信噪比。然后采用偏最小二乘回归（PLSR）、支持向量回归（SVR）和BP神经网络算法建立花色苷含量预测模型。研究表明：基于PLSR模型推荐的13个隐含变量建立的BP神经网络模型的预测决定系数和预测均方根误差分别为0．9102和0．3795。
This work aimed to determine the anthocyanin content in skin based on hyperspectral imaging technology. The grapes of Cabernet Sauvignon （ Vitis vinifera L. ） produced in Shaanxi province were used as experimental materials. Hyperspectral images of 60 groups of grape samples were collected by near infrared hyperspectral camera （900 - 1 700 nm）. After then, the anthocyanin content of skin was detected by pH-differential method. The grape berry regions of hyperspectral images were extracted as region of interest （ROI） in which its average spectrum was calculated. Moreover, different preprocessing methods were used to improve the signal noise ratio （SNR） including Savitzky- Golay smoothing, normalization and muhiplicative scatter correction, et al. Prediction model was established for determining anthocyanin content by the partial least squares regression （PLSR） , least squares support vector regression （SVR） and BP neural network （BPNN）. It was shown that prediction coefficient of determination （P-R2 ） of BPNN model built by the thirteen latent variables recommended by PLSR model was 0. 910 2 and the root mean square error of prediction （RMSEP） was 0. 379 5.
出处
《农业机械学报》
EI
CAS
CSCD
北大核心
2013年第12期180-186,139,共8页
Transactions of the Chinese Society for Agricultural Machinery
基金
国家自然科学基金资助项目(61003151)
'十二五'国家科技支撑计划资助项目(2012BAD31B07)
中央高校基本科研业务费专项资金资助项目(QN2011099
QN2013062
QN2013055)
国家葡萄产业技术体系酿酒葡萄栽培岗位子项目(CARS-30-02A)
关键词
酿酒葡萄
花色苷
高光谱图像偏最小二乘回归
支持向量回归
BP神经网络
Winegrape Anthocyanin Hyperspeetral image Partial least squares regression Support vector regression BP neural network
作者简介
刘旭，讲师，博士，主要从事酿酒葡萄果实生理与质量控制研究，E-mail：liuxu@nwsuaf．edu．cn
通讯作者：宁纪锋，副教授，博士，主要从事计算机视觉、模式识别与农业信息化研究，E-mail：jf_ning@sina．com
