| As a nutritious and delicious seawater fish,salmon is loved by consumers all over the world.In recent years,there has been a strong demand for salmon in China’s market,and the import volume has increased continuously.The import mode mainly includes chilled and frozen.Compared with frozen salmon,chilled salmon maintain more excellent quality,but at the same time,the cost is higher and the price is more expensive.Therefore,some illegal traders sell frozen-thawed salmon as chilled salmon to make more profits.Such fraudulent behavior not only seriously harms the interests of consumers,but also goes against the healthy development of salmon consumption market in China.At present,most of the identification methods for chilled and frozen-thawed meat are tedious and time-consuming,which are difficult to realize large-scale detection.In the study,the optical and electrical characteristics of chilled and frozen-thawed salmon were analyzed by using hyperspectral imaging technology and impedance spectroscopy technology,and combined with chemometrics methods to identify chilled and frozen-thawed salmon.Besides,based on the sensitivity of spectrum to the detection of physicochemical indicators of meat,the fast detection and distribution visualization of quality indexes of chilled and frozen-thawed salmon were studied by using hyperspectral imaging technology.The main research contents and conclusions are illustrated as follows:(1)Identification of chilled/frozen-thawed salmon and the number of frozen-thawed times based on hyperspectral imaging technology.Taking chilled salmon and salmon with different frozen-thawed times as the research object,hyperspectral image data of objects were collected by hyperspectral imaging system firstly.Then the ENVI 4.5 software was used to extract the average spectrum of the region of interest(ROI)in the hyperspectral images of the sample,and the gray level co-occurrence matrix(GLCM)method was used to extract the texture information of the first three principal component images.After the spectral information was preprocessed and the spectral wavelength was screened,the identification model of chilled salmon and frozen-thawed salmon was established based on spectral information,image information,and fusion spectral-image information.The results showed that the extreme learning machine(ELM)model based on spectral information had the best recognition effect for chilled and frozen-thawed salmon,and the recognition rates of its calibration set and prediction set were 100.00%and 95.00%respectively.In the identification of frozen-thawed times of salmon,the best model was also the ELM model based on spectral information.The recognition rate of calibration set and prediction set was 97.50%and 91.67%,respectively.(2)Fast detection of salmon quality index and visualization of its distribution based on hyperspectral imaging technology.Taking chilled salmon and salmon with different frozen-thawed times as the research object,hyperspectral image of objects were collected first and then the average spectrum of ROI was extracted.Then the physicochemical truth value of the quality index in the sample was determined by the routine test method.The partial least squares(PLS)prediction model for each quality index was established after the spectral information was preprocessed and the wavelength was screened.The results of the model showed that the PLS quantification model based on the spectrum after the selection by the competitive adaptive reweighted sampling(CARS)could better predict the quality index of salmon.In the prediction of a*value,shear force,equivalent umami concentrations(EUC)and K value,the correlation coefficient(r_p)of the test set were 0.9243,0.8854,0.8093 and 0.9493,respectively,and the root mean square error of prediction(RMSEP)were 0.78,0.64 N,1.13 gMSG/100g and 1.46,respectively.After determining the optimal prediction model corresponding to each quality index,the corresponding quality index value was calculated by the spectral value of each pixel in the hyperspectral image of the sample.Finally,according to the coordinate information of pixels and the corresponding index value,the distribution of salmon quality index on the image was reconstructed in the form of pseudo color image,and the distribution visualization of salmon quality was realized.(3)Identification of chilled/frozen-thawed salmon and the number of frozen-thawed times based on impedance spectroscopy.Given the situation that the detection effect of impedance spectroscopy technology in meat detection was reduced due to the individual differences of samples,this study took chilled salmon with different frozen-thawed times as the research object.The differences of four impedance spectroscopy between chilled salmon and frozen-thawed salmon were analyzed and compared.Then,the identification model of chilled and frozen-thawed salmon based on different impedance spectra was established,and an efficient and robust identification method of chilled and frozen-thawed salmon based on impedance spectroscopy was identified according to the recognition rate.It was found that the CARS-BPANN model based on phase spectroscopy had the best effect on the identification of chilled and frozen-thawed salmon,the recognition rate of the calibration set and prediction set reached 100.00%,which could effectively distinguish chilled and frozen-thawed salmon;CARS-ELM based on phase spectroscopy had the best effect on the frozen-thawed times identification of salmon,the recognition rates of calibration set and prediction set were 91.67%and 86.67%,respectively. |
