Research On No-destructive Testing Of Fruit Quality Based On Hyperspectral Imaging Technology

China is the largest fruit producer in the world,with the annual output ranking the first in the world.However,the proportion of fruit export in the output is still at a low level compared with some developed countries,and one of the important factors affecting the fruit export is the fruit quality.At present,the quality of fruit has become the basic factor of whether fruit can enter the domestic high-end market and foreign market.In addition to improving planting technology to improve fruit quality,refined fruit quality detection technology and evaluation standards are also urgently needed in the market.In recent years,hyperspectral imaging technology,as a fast,nondestructive and simple imaging technology,has been widely used in the non-destructive testing of fruit quality,and this technology has the advantages of"integration of atlas",which is more suitable for the non-destructive testing of fruit quality.Therefore,in this study,strawberry and Feicheng peach were taken as the research objects,and the hyperspectral data were obtained by portable hyperspectral imager and measured by instrument Content（Soluble solids content and firmness）,combined with the chemometrics method,the qualitative discrimination model of strawberry and Feicheng peach at maturity and the quantitative prediction model of fresh-keeping Strawberry（SSC）and postharvest Feicheng peach（SSC and firmness）were established,and the internal quality of Feicheng peach was visualized and analyzed.The main research conclusions are as follows:（1）Field evaluation of strawberry ripeness based on hyperspectral imaging technology,the hyperspectral data of unripe,mid-ripe and ripe strawberries were obtained in the field and the data was collected in the laboratory for comparison.The spectral data was preprocessed using multiplicative scatter correction（MSC）,The x-loading weight（XL）method,continuous projections algorithm（SPA）and competitive adaptive reweighted sampling（CARS）were used to extract the characteristic wavelengths and establish the partial least squares discrimination analysis（PLS-DA）and least-squares support vector machine（LS-SVM）ripeness discriminant models.It was found that the accuracy of the PLS-DA and LS-SVM discriminant models established on the field collected data was 91.7 to 96.7%,and the discriminant accuracy of the model collected on the laboratory was 93.3 to 100%.Among them,among the models established by field data,the PLS-DA discriminant model established by full-spectrum data and the CARS extraction of 7 characteristic wavelengths（423nm,585nm,614nm,715nm,828nm,886nm,973nm）established the LS-SVM model recognition accuracy both reached 96.7%.Therefore,it is feasible to use hyperspectral imaging technology to evaluate the maturity of strawberry in the field,and it also provides technical support for the field detection of fruit and vegetable maturity and the development of detection instruments.（2）Detection of soluble solids content in strawberry coated with chitosan by hyperspectral imaging.For 0%,0.5%,and 1%concentrations of chitosan-coated strawberries,stored for 1,2,and 4 days respectively.Hyperspectral imaging and measurement of SSC found 0.5%And 1%chitosan coating can inhibit the reduction of strawberry SSC and prolong the fresh taste of strawberry.Then,32,30 and 20 characteristic wavelengths of three concentration samples were optimized by CARS algorithm,11,8 and 16 characteristic wavelengths were optimized by SPA algorithm,and then PLSR and SVR models were established to predict SSC of coated strawberry.The results show that SPA-SVR model has better performance.For 0%CTS coating samples,the modeling set coefficient²₍(8（8）=0.865,prediction set coefficient²=0.835;for0.5%CTS coating sample,modeling set factor²₍(8（8）=0.808,prediction set coefficient²=0.799;for 1%CTS coating sample,modeling set factor²₍(8（8）=0.834,prediction set coefficient²=0.875;further analysis of the principal component image of coated strawberries showed that the morphology of 0.5%CTS coated samples was relatively complete with the increase of storage days,which also showed that chitosan coating had different effects on the shelf life of strawberries.Therefore,based on hyperspectral imaging technology and different data processing methods,SSC detection of chitosan coated strawberry can be realized,which provides theoretical guidance for the quality detection of fresh fruits.（3）The quality of Feicheng peach was visualized and the hyperspectral imaging test was performed at maturity stage.Spectral information,SSC and firmness of 80 Feicheng peaches were collected from seven mature and nine mature.Characteristic wavelengths were selected using CARS and SPA,and a multiple linear regression（MLR）model was established.The analysis of the results found that the CARS-MLR model has better performance.The CARS-MLR model for predicting SSC has modeling set accuracy of²₍(8（8） and prediction set accuracy of²of 0.8191 and 0.8439,respectively.The ratio of standard deviation of the validation set to standard error of prediction（RPD）was 2.The CARS-MLR model for predicting firmness has a modeling set accuracy of²₍(8（8） and a prediction set accuracy of²of 0.9518 and 0.8772,respectively,RPD=2.1.Through the generated SSC visual distribution map,nine mature peaches were mainly dark yellow,SSC was mainly concentrated in 12¹5°Brix,seven mature peaches were mainly light yellow,SSC was mainly concentrated in 10¹3°Brix;in the firmness visual distribution map,nine mature peaches were mainly light green,firmness was mainly concentrated in 3⁸kg/cm²,seven mature peaches were mainly light yellow and green,firmness was mainly concentrated in 6¹2kg/cm²;three characteristic wavelengths（493nm,530nm,720nm）were extracted by SFS algorithm,and the artificial neural network（ANN）maturity discrimination model was established with the characteristic band data as input,with the classification accuracy of 98.3%,which lays a data foundation for the development of online quality detection equipment for postharvest peaches and other fruits and vegetables.