Rapid And Non-destructive Detection Of Cured Meat During Drying Processing Using Hyperspectral Imaging Technique

Cured meat as one of the representative traditional Chinese meat products,which has been won the preference of the public for a long period.At the same time,for the reseaeches,the quality of cured meat has also become a hot spot,but the current detection technology is still stuck in the period of damaged,directly traditional detection technology.As an emerging technology,hyperspectral-imaging(HSI)technique has attracted the attention of the researcheres due to it's integration of conventional computer imaging and spectral analysis to attain spatial and spectral information from an object,which can easily conquer the shortcomings of spectroscopy and computer vision mentioned above.However,the detection of food-related hyperspectral researches which are mostly about physical and chemical indicators and characteristics of food,while processing control are reported relatively rarely.In this paper,we will have a research about the spectral characteristics of the cured meat during the heating processing,and then detect and analyze the quality of the bacon in the heating processing.It can make full use of the advantages of HSI technology about the rapid,non-destructive and non-contact characteristics.The main contents and results which can be listed as follows:(1)The use of visible/near-infrared HSI technology in tandem with stoichiometry and the ENVI image processing technology for having a research on the effects of cured meat about the moisture,fat and protein of cured meat under the different heating times.Multivariate calibration models were developed using partial least-squares regression(PLSR)and least-squares support vector machines(LS-SVM)in the full spectral range.In order to simplify the calibration model,the optimal wavelengths identified using the regression coefficient(RC)was selected to build simplified model.The prediction accuracy of the PLSR model(Rp2 = 0.823,RMSEP = 0.568)is better than the MLR model in fat,while the MLR model has better prediction accuracy in water content(Rp2 = 0.943,RMSEP = 2.73)and protein content(Rp2 = 0.891,RMSEP = 2.57).(2)Rapid detection of the NPN and TVB-N value due to protein degradation in cured meat during heating processing based on HSI technology to know the the degree of hydrolysis of protein and the changes of TVB-N and spectral,and then choose the optimal wavelengths.Results indicated that the prediction modle of NPN(Rp2 = 0.933,Rp2 = 0.922)and TVB-N(Rp2 = 0.825,Rp2 = 0.861)by using PLSR and MLR modle both are stability and reliability.In addition,the dynamic variations of TVB-N content were finally visualized and obtained.(3)To study and establish the rapid detection method of fat corruption in cured meat heating processing,this paper mainly studies the representative index of TBA during the heating processing,and then establish a quantitative model for hyperspectral detection.Combining the image texture information with the spectral information under the characteristic wavelength,and then the influence of the image texture information on the model establishment is discussed.