Comparative Study On Rapid Detection Of Antioxidants In Mutton Based On Hyperspectral With Different Spatial Resolutions

Antioxidant is an important detection index to measure the quality of mutton.Traditional detection methods are time-consuming,laborious and it is not possible to quickly and accurately detect antioxidants in mutton.Hence,the experiment used hyperspectral and microscopic hyperspectral imaging techniques to obtain map information of different spatial recognition degrees of samples,combined with chemometric methods to determine the soluble protein content,glutathione(GSH)content,peroxidase(POD)activity and catalase(CAT)activity in mutton by modeling analysis.It provided a reference for its application in the field of non-destructive testing of mutton quality.The main research contents and conclusions are as follows.(1)Using the spectral information of hyperspectral imaging technology to conduct rapid and non-destructive testing of mutton soluble protein content,GSH content,POD activity and CAT activity and establish model.Using spectral data extracted from five characteristic wavelengths,multiple linear regression(MLR)and least squares support vector machine(LS-SVM)models were established.The results showed that the model was established by the interval variable iterative space shrinkage approach-Iteratively retaining informative variables(iVISSA-IRIV)characteristic wavelength extraction data had the best effect,and the soluble protein content was the best in the LS-SVM model,RC and RP were 0.9146 and 0.8818,respectively;the GSH content had the best effect in MLR model,RC and RP were 0.8446 and 0.8705 respectively;POD activity was the best in the MLR model,and its RC and RP were 0.8935 and 0.8373,respectively;it showed that the iVISSA-IRIV feature wavelength extraction method was easier to extract the antioxidant-related feature information of mutton.The MLR model was established had the best effect by using CARS characteristic wavelength to extract CAT vitality data,and its RC and RP were 0.9173 and 0.9047,respectively.(2)Using the spectral information of microscopic hyperspectral imaging technology to conduct rapid non-destructive testing of mutton soluble protein content,GSH content,POD activity and CAT activity,and establish MLR and LS-SVM models.The results showed that:the MLR model was shown as the best prediction model in six data sets,among them,the characteristic wavelength data extracted by SPA had the best effect in soluble protein content(RC=0.8406 and RP=0.7544)and GSH content(RC=0.8252 and RP=0.6486).Extracting characteristic wavelength data with CARS method in POD activity(RC=0.8620 and RP=0.8830)and CAT activity(RC=0.8969 and RP=0.9105)were best.It showed that the use of MLR model in microscopic hyperspectral imaging technology was an effective method to predict mutton antioxidants,and the extraction of characteristic wavelength data can replace the full-band data to analysis.(3)The gray level co-occurrence matrix(GLCM)was used to extract spectral image texture information and characteristic spectral information for fusion by hyperspectral imaging and microscopic hyperspectral imaging,which established linear and nonlinear models and studied comparatively.It showed that the LS-SVM model established by the hyperspectral characteristic spectral information was better than the fusion information for the soluble protein content.The GSH content and CAT activity had the best effect in the establishment of the MLR model in the hyperspectral fusion information,which was 2.2%and 3.9%higher than the characteristic spectral information;the MLR of micro-hyperspectral fusion information of POD activity was the optimal model,which was 1.1%higher than the characteristic spectral information.To sum up,at the spectral level,hyperspectral imaging technology had more advantages in the analysis of mutton antioxidants,which may predict its antioxidant indicators quickly and accurately;at the fusion level,the model effects displayed the good results through the fusion of image texture and characteristic spectral information.Therefore,hyperspectral imaging and microscopic hyperspectral imaging are an effective method to predict the antioxidants of mutton.