Research On The Detection Technology Of Machine Vision And Raman Spectroscopy For Egg Freshness

More and more people care more about food quality along with the improvement of life.Freshness and safety are two important criteria for consumers to buy ingredients.Eggs have the largest occupancy of marketed egg products because of their extensive consumption and rich nutrition.As the storage time increases,the internal and external physical parameters and various nutritional indicators will change accordingly,which will reduce the consumption performance,and the detection of egg freshness is an important guarantee to determine the edible performance of eggs.At present,the detection of egg freshness is still dominated by manual testing in China which has some disadvantages,such as strong subjective factor interference,irregular detection and high labor intensity.How to get rid of the traditional manual detection mode of egg freshness and trend into a convenient and efficient way step by step is a research direction with practical value and significance.On the one hand,the convenient and efficient detection technology can protect the rights and interests of consumers,and on the other hand,it can provide effective information to egg sales market for the reasonable operation,to ensure the quality meets its price.The detection technology of Machine vision has the advantages of safety,high efficiency and nondestructive examination,it can measure and analyze the internal and external quality of sample with the shape,size and color texture.In the process of data acquisition by Raman spectroscopy,the steps of collection are simple while the sample size is small,and the Raman spectra indicates the constitution and frameworks of substance molecules.Eggs with different freshness have different shape parameters of yolk and chamber,and the molecular composition of protein in albumen is different,so there are differences in image and Raman spectrum information can be observed.Therefore,machine vision and Raman spectroscopy can be used to analyze the quality of eggs in different storage periods.In this paper,the changes of freshness index of eggs in different storage time were analyzed by the detection technology of machine vision and Raman spectroscopy independently,combined with different data processing and modeling methods.The main research work includes:(1)The changes of relevant physical and chemical parameters with the changes of egg freshness were analyzed according to the biological knowledge,and relevant parameters of egg quality were determined.(2)The physical parameters of the egg were measured after the image had been collected in a nondestructive way,the image information was processed with filtration,binarization,color space conversion and extraction to extract the characteristics of yolk and gas chamber.(3)The physical parameters and feature extraction parameters were analyzed by SPSS software,which was to establish the model of egg weight and protein height.(4)Taking the egg weight and protein height as input,the egg freshness grade as output,established the BP neural network,RBF neural network,support vector machine,and deep belief network machine learning model to predict the degree of egg freshness,at the same time,verified the forecast accuracy of each model by the data of validation set sample.(5)Raman spectra of the albumen stored for different days were collected after the physical parameters had been measured,the characteristic peak intensity was extracted after the de-noising treatment.(6)The characteristic peak intensity of Raman spectrum and the measured yolk index,chamber height,protein height and Hu value were analyzed by SPSS software,which was to study the relationship between the index of egg freshness and the peak intensity of albumen's Raman spectrum.The experimental results show that,deep belief network model achieved an accuracy of93.3% in the detection system of egg freshness based on the technology of machine vision,the prediction accuracy was high and the model was stable,which means that it can be used as a reference for selecting the modeling algorithm to predict the fresh quality of eggs in the future;The detection technology which was based on Raman spectroscopy,showing that the relationship between the index of egg freshness and the peak intensity of eggsamples at different storage times was linear,which was analyzed by establishing the relationship model between the peak intensity of characteristic peaks and the yolk index,protein height,chamber height,and Hu number,the results show that it is feasible to detect the egg freshness by Raman spectroscopy,which can be used as a reference for the determination of fresh quality of eggs in the future.