The Quality Detection In Polluted Egg Liquid And Egg Based On Hyperspectral Technology

As an important source of high quality protein for human,eggs have the high value of nutrition and health care.However,due to the encapsulation of eggshell,consumers can't judge the freshness of eggs intuitively,meanwhile,because of the rich nutrient in egg liquid,it is vulnerable to be polluted by the environment in storage process.As a new detection technology,the hyperspectral technology has many advantages,such as the integration of spectrum and image,green environmental protection,fast analysis and so on.In order to explore the feasibility of detecting egg quality based on hyperspectral technology,this paper,which was supported by the project of China Youth Science Foundation of National Natural Science Foundation(C200701),chose the egg liquid and egg as the objects respectively and built the quantitative prediction models of the polluted egg liquid and egg quality based on visible-near infrared and near infrared hyperspectral technology,the results can provide reference for the development of egg safety detection and spectral identification technology.Specific research contents and results are as follows:1.The analysis of quality index change under different storage temperaturesThe samples,which inoculated with mixed suspension of Pseudomonas aeruginosa and Escherichia coli(1:1)were stored at low temperature(4?),room temperature(20?)and high temperature(35?)respectively,and then measured the changes of quality indexes during storage.The results showed that the change trend of egg liquid and egg quality indexes were the same at different temperatures,and there were significant correlation between them and storage days(p<0.05).For egg liquid samples,the yolk index and protein content decreased with time,while the albumen pH,TVB-N content and the total number of colonies increased with time.For egg samples,the Haugh unit,yolk index and protein content decreased with time,and the albumen pH,TVB-N content and total number of colonies increased with storage time.The higher the temperature,the more obvious the change of quality indexes.But at the end of the test,there is no significant difference in the value of quality index between different temperatures.2.The quality detection in polluted egg liquid based on Visible-Near Infrared hyperspectral technologyVisible-near infrared hyper-spectral information of egg liquid stored at room temperature(20?)was collected,the partial least squares regression(PLS)and support vector machine regression(SVM)were built to quantitatively predict the quality indexes of polluted egg liquid based on the full-band spectroscopy and characteristic bands screened by successive projections algorithm(SPA).The results showed that,the PLS prediction model based on full-band had a better result on yolk index with the RP and RMSEP were 0.94 and 2.06%,respectively.The SPA-PLS prediction model had a better result on egg albumen pH with the RP and RMSEP were 0.77 and 0.10,respectively.The SPA-SVM prediction model had a better result on protein content with the RP and RMSEP were 0.87 and 0.44 g/100 g,respectively.And for TVB-N content,the SVM prediction model based on full-band was better,the RP and RMSEP were 0.85 and 3.49 mg/100 g,respectively.Further,in order to investigate the detection of microorganisms in different parts of egg liquid,the prediction models of total colonies in egg albumen,yolk and whole egg liquid were established,respectively.The results showed that the better prediction models were all the SPA-SVM models,in which RP was 0.81,0.83 and 0.76,RMSEP was 0.63,0.47 and 0.71 1g CFU/g,respectively.3.The quality detection in polluted egg based on Infrared hyperspectral technologyThe near infrared hyper-spectral information of egg stored at room temperature(20?)was collected,and the same methods was used to build the models to quantitatively predict the quality indexes of polluted egg.The results showed that the SPA-SVM prediction model had a better result on Haught unit with RP and RMSEP were 0.87 and 3.94,respectively.For yolk index,the PLS prediction model based on full-band was better,the RP and RMSEP were 0.78 and 3.02%,respectively.For egg albumen pH,the SPA-SVM prediction model was better,the RP and RMSEP were 0.70 and 0.11,respectively.For protein and TVB-N content,the PLS prediction model based on full-band had better result with RP were 0.85 and 0.88,RMSEP were 0.54 g/100 g and 3.39 mg/100 g,respectively.The SPA-SVM prediction model showed better result on the total colonies with RP and RMSEP were 0.83 and 1.00 lg CFU/g,respectively.Furthermore,combined with the quality index prediction model and image processing technology,we used different colors to represent the content of different quality indexes.After pseudo-color processing of the egg images,the visualization of freshness and pollution degrees of egg were realized,which provided a new scheme and idea for the quality and safety control of eggs.