摘要
本研究的目的是建立马奶葡萄果实表皮色泽的可见/近红外漫反射光谱检测数学模型,并评价其应用价值。以明度L*和色度坐标a*、b*为评价指标,讨论应用不同化学计量学建模方法、不同的光谱预处理方法、不同有效波长区间建立定标模型的效果,并用最优模型对未知样品进行预测,以验证模型精度。结果显示,在全光谱范围内,应用偏最小二乘法(PLS)结合二阶导数、去散射的光谱预处理方法的L*值模型效果最好,其交互验证误差SECV、交互验证相关系数R2CV分别为0.4591、0.9476;a*、b*值模型宜采用一阶导数结合加权多元离散校正处理,SECV分别为0.1239、0.4496,R2CV为0.9508、0.8648。实验证明,应用可见/近红外漫反射技术对马奶葡萄果实表皮色泽进行快速无损检测是可行的,L*、a*模型稳健,b*模型精度有待进一步提高。
The aim of this study was to establish the mathematical model based on visible / near infrared diffuse reflectance spectroscopy for detecting the color of Manai grape,and to evaluate its application value. With the brightness L＊and chromaticity coordinates a＊,b＊as evaluation indexes,different stoichiometry modeling methods,different spectral preprocessing methods and different effective wavelength intervals were discussed,and the unknown samples were predicted with the optimal model in order to validate the accuracy. The results showed that in the whole spectral range,L＊got the best result with partial least squares（ PLS） combined with the second order derivative,scattering spectra pretreatment methods,the SECV,R2 CVwere 0.4591,0.9476; optimal methods for a＊,b＊were the first derivative combined with the weighted multiple discrete correction processing,the SECV were respectively 0.1239,0.4496,and the R2 CVwere 0.9508,0.8648.Application of quick and nondestructive detection based on visible / near infrared reflectance technique was feasible in measuring surface color of Manai grape,models of L＊and a＊were moderate,the accuracy of b＊model remained to be further improved.
出处
《食品工业科技》
CAS
CSCD
北大核心
2015年第19期308-311,378,共5页
Science and Technology of Food Industry
基金
"十二五"国家科技支撑计划项目(2012BAD38B01)
天津市高等学校创新团队培养计划项目(TD12-5049)
关键词
可见/近红外漫反射光谱
马奶葡萄
表面色泽
无损检测模型
visible / near infrared diffuse reflectance spectroscopy
Manai grape
surface color
detection model
