Study On The Feasibility Of Rapid Measurement Of Fat And Bacillus Cereus In Liquid Milk By Hyperspectral Technique

It has been a hot topic for the detection of food nutrition and food safety,which is of great significance to meet the needs of consumers' food quality and safety.The traditional methods of food detection are time-consuming,complex,and destructive testing methods.In order to realize rapid and nondestructive detection of food composition,quality and safety,a new quantitative detection method based on hyperspectral imaging technology was proposed and studied,which was applied to test Bacillus cereus and fat content in liquid milk in this paper.Combined with image processing technology,spectral analysis technology and chemical metrology technology,the feasibility of constructing the prediction models of the fat content and the content of Bacillus cereus in milk was explored.The main research contents and conclusions were summarized as follows:(1)The image processing technology was used to select the sample area,and the average spectrum was extracted.The PLS and N-PLS prediction model were established to analyze the fat content in milk.The results showed that the correlation coefficients between the calibration set and the prediction set in the PLS prediction model were 0.98 and 0.99 respectively,and RMSEC and RESEP were 0.31 and 0.24 respectively.The correlation coefficients between the calibration set and the prediction set in the N-PLS prediction model were 0.99 and 0.99 respectively,and RMSEC and RESEP were 0.03 and 0.12 respectively.The accuracy of N-PLS model was higher than that of PLS model,especially in the case of low fat in milk.(2)The image processing technology was used to select the sample area,and the function of the energy value(Energy)texture feature was applied to reduce the dimension of hyperspectral data.The spectral characteristic values were gotten based on the texture feature analysis.And then,the PLS model was built to predict the content of bacillus cereus in milk.The correlation coefficients between the calibration set and the prediction set in the PLS prediction model were 0.92 and 0.91 respectively,and RMSEC and RESEP were 0.73 and 0.81 respectively.The results showed that the PLS prediction model could only identified the high and low concentration of Bacillus cereus in the milk.Therefore,two-dimensional correlation technology combined with N-PLS method was proposed,and the N-PLS prediction model was built.The correlation coefficients between the calibration set and the prediction set in the N-PLS prediction model were 0.99 and 0.99 respectively,and RMSEC and RESEP were 0.02 and 0.09 respectively.The results showed that the N-PLS model had higher accuracy and was able to achieve quantitative analysis of Bacillus cereus in milk.