| In this study,Ningxia Jingyuan yellow beef was taken as the research object,the moisture and chroma of beef was determined as the physical and chemical index.Combined with hyperspectral imaging technology,the hyperspectral prediction research and content visualization analysis were carried out on the moisture and chroma of Ningxia Jingyuan yellow beef.The spectral prediction model of beef key quality was constructed,and the online detection system and equipment were completed.And described working principle and composition of the software and hardware about on-line detection system.The results can provide technical support for on-line rapid detection of moisture and chroma of Jingyuan yellow beef,and provide theoretical basis and technical route for online detection of meat quality and meat identification.The results are as follows:(1)N Near infrared hyperspectral imaging technology combined with physical and chemical value detection method was used to determine moisture and chroma of Jingyuan yellow beef aged 18-22 months.The results showed that the moisture content of 18,20,and 22 month old beef was 75.84%,71.48%,and 68.20%,respectively.The moisture content of Jingyuan yellow cattle decreased with the increase of age;the L*value and a*value of chroma increased with the increase of age,and the B*value decreased with the increase of age.Comprehensive comparison showed that the overall content of beef at 22 months old was balanced and the quality was good.(2)The region of interest(ROI)was extracted and the average spectral value was calculated;seven different preprocessing methods are used to preprocess the original spectrum,and the abnormal sample values are eliminated by Monte Carlo cross validation method;the best preprocessing method is selected to preprocess the original spectral,and continuous projection algorithm(SPA),competitive adaptive reweighting algorithm(CARS)and no signal are used.In order to optimize the prediction model of beef chroma value,a partial least squares regression(PLSR)model based on different characteristic wavelengths was established.The results show that obvious differences existed between the 400-1000 nm average spectral,indicating that there was a certain correlation between the spectral and the physical and chemical indexes.Most of samples,the model with L*value pretreated by resolving method had the best results,(?)was 0.9790,(?)was 0.9766,red-green(a*)value was the best,(?)was 0.8070,(?)was 0.9196,blue-yellow(b*)value was the best,(?)was 0.9311,(?)was 0.9506;(3)Near infrared spectrum has strong absorption of moisture,and the detection of beef moisture content can achieve high accuracy by using hyperspectral imaging technology.Using vs2015 program development environment,combined with opencv2.4.14.Software,calling matlab characteristic wavelength extraction and prediction model,establishing DLL data analysis dynamic link library file,developing beef moisture online rapid detection system combining software and hardware,achieving beef spectral data collection and moisture content model analysis.The prediction model of beef moisture based on CARS-PLSR method was established,with(?)and(?)values of 0.814 and 0.750 respectively,and the determination coefficient of correction set((?))and prediction set determination coefficient((?))were 0.817 and 0.850 respectively;12,27 and 27 characteristic wavelengths were selected by CARS,SPA and UVE respectively;the advantages of PLSR model based on full spectrum and characteristic spectrum were compared The results showed that the root mean square error(RMSEC)of correction set and root mean square error of prediction set(RMSEP)were 0.477 and 0.555,respectively.Finally,the moisture content of each pixel of sample was calculated by CARS-PLSR model,and the moisture distribution of sample was visualized by using pseudo color map,so as to achieve the rapid detection of beef moisture content and visual expression of its distribution.;(4)Based on the research of meat on-line detection for many years,we independently developed an online beef moisture content detection system and equipment.Using the moisture prediction and content distribution model,the composition and working principle of the software and hardware of the online detection system were described,and the prediction model of meat chroma value was implanted and tested... |
