Raman chemical feature extraction for quality control of dairy products

摘要

As a fast information acquisition technique, Ramanspectroscopy can be used to control the quality of dairyproducts. Feature extraction is a necessary processingstep to improve the efficiency of Raman spectral data.Principal component analysis is a traditional methodthat can effectively extract the features and reducethe dimension of spectral data. However, it is difficultto analyze the chemical information of the extractedfeature, thus limiting its practical application. In thiswork, Raman spectral chemical feature extractionwas carried out. The quality control of Dingxin dairyproducts (a famous dairy brand in China; purchasedfrom Heilongjiang Zhaodong Tianlong Dairy Co. Ltd.,Heilongjiang, China) was used as an example. Ramanpeak intensity, peak area, and peak ratio were extractedas chemical features and further investigatedusing Euclidean distance and the quality fluctuationcontrol chart. The potential quality discrimination abilityof the Raman feature extraction methods was demonstrated.The results showed that the Puzhen dairyproducts (purchased from Inner Mongolia Yinuo HalalFood Co. Ltd., Inner Mongolia, China) and Xueyuandairy products (used as a control; purchased from InnerMongolia Wulanchabu City Jining Xueyuan Dairy Co.Ltd., Inner Mongolia, China) could be distinguishedfrom Dingxin dairy products when the Raman chemicalfeatures special peak intensity, peak area, and peak ratiowere used, and their discriminatory ability increasedin sequence. Hence, it was shown that Raman chemicalfeature extraction can achieve quality control anddiscriminant analysis of dairy products and that thespectral information is clear.