利用可见近红外光谱技术对小米产地进行溯源研究
目的 利用可见/近红外光谱技术对产自不同地区的晋谷21号小米进行溯源研究。方法 使用近红外光谱仪获取产自洪洞、浮山、沁县3个不同地区的晋谷21号小米400～1004 nm波段范围内的漫反射光谱; 对光谱分别进行多元散射校正法(multiple scattering correction, MSC)、一阶导数法(first derivative, 1St-D)预处理; 对预处理光谱进行主成分分析, 全交叉验证确定最佳主成分数量, 获取主成分; 同时选择预处理光谱特征波长。使用马氏距离法、线性判别法建立判别模型, 最后用未知样品的验证准确率来表示模型的判别效果。结果 原始光谱和MSC处理光谱提取特征波长分别建立的产地判别模型对3个不同产地的小米判别完全准确; 1St-D处理光谱基于7个主成分结合马氏距离法和基于9个主成分结合线性判别法建立的2种判别模型对3个不同产地的小米亦实现完全准确判别。结论 可见/近红外反射光谱技术用于小米产地的判别具有可行性, 本研究可为小米产地的快速判别应用中提供技术基础。
Objective To discriminate the Jingu21 millet of different regions by visible/near infrared spectroscopy (VIS/NIR). Methods The infrared diffuse reflection spectrum were obtained at 400~1004 nm from 3 different areas of millet, including Hongdong, Fushan and Qinxian. The millet spectrums were pretreated by multiple scattering correction (MSC), first derivative (1st-D) method, respectively. Then the best number of principal components was determined by principal component analysis of pretreated spectra. At the same time, the characteristic wavelengths of preprocessing spectrum were collected. Then the discrimination models were established based on the Mahalanobis distance and linear distance methods, respectively. Finally, the validity of the model was proved by the accuracy of the unknown samples. Results Based on characteristic wavelength of original spectrum and MSC spectrum, the results of millet discrimination models were completely accurate, respectively. Based on the Mahalanobis distance and the linear discriminant analysis, the 1st-D spectra discrimination models of 7 principal components and 9 principal components also had the best results. Conclusion The VIS/NIR can be used to identify the origin of millet, which can provide a certain technical basis for the application of VIS/NIR spectroscopy technique in the rapid identification of millet.
标题：利用可见近红外光谱技术对小米产地进行溯源研究
英文标题：Discrimination of the origin of millet by visual/near infrared reflectance spectroscopy
作者：
李佳洁 山西师范大学食品科学学院
吴建虎 山西师范大学食品科学学院
张海波 山西师范大学食品科学学院
中文关键词:小米,可见/近红外反射光谱,特征波长,主成分分析,
英文关键词:millet,visible/near infrared reflective spectroscopy,characteristic wavelength,principal component analysis,
发表日期：2017-04-10
- 文件大小:
- 24.03 MB
- 下载次数:
- 60
-
高速下载
|
|
