摘要
以57份小麦籽粒为样本,分别采用Vertex 70傅立叶近红外光谱仪(单籽粒测样附件)和近红外增强型高光谱成像系统采集光谱,建立预测小麦籽粒蛋白含量的模型。基于近红外的小麦单籽粒模型相关系数为0.52,交叉校验均方根误差为0.807;而基于高光谱建立的模型相关系数为0.81,交叉校验均方根误差为0.7035。结果表明:在样本量少且为籽粒形态时,可优先考虑高光谱技术替代传统的近红外单籽粒采样模式来检测样本内部品质,但其实用性还有待进一步验证。若深入结合图像信息,高光谱技术在农产品内外品质检测方面有更广阔的应用。
57 unit wheat grain sample＇s spectrums are collected separately by Fourier near infrared（NIR） spectrometer（single grain test sample accessories）and HyperSIS-NIR system,and the wheat grain protein content models built respectively by NIR and hyperspectrum is mainly compared,and the model correlation coefficient based on NIR is 0.52 and error of cross validation（RMSECV） is 0.807,while the model correlation coefficient based on hyperspectrum is 0.81 and RMSECV is 0.703 5.The results show that when the sample is few and in grain form,hyperspectrum technology has more advantages to test sample internal quality than NIR,but its practicality remains to be further verified.If further combined with image information,hyperspectrum technology has more an application in detecting internal and external quality of agricultural products.
出处
《传感器与微系统》
CSCD
北大核心
2013年第2期60-62,共3页
Transducer and Microsystem Technologies
基金
北京大学生科学研究与创业行动计划建设项目(PXM2012-014213-000067)
北京市教委科研基地科教创新平台项目(PXM2012-014213-000023)
关键词
高光谱
近红外
小麦籽粒
蛋白
hyperspectrum
near infrared（NIR）
wheat grain
protein
