Research On Quality Parameters And Grade Evaluation Of Hanfu Apples Based On Hyperspectral Imaging

Although China is the world's largest producer of apples,its export volume is only less than 3% of its total output.Compared with other major apple producing countries and exporting countries,there is a big gap.The reason is because Apple's post-harvest commercialization is low,and grading plays a central role in post-natal treatment.At present,the classification of Hanfu apple is mainly based on manual sorting,but its sorting subjectivity is strong and the detection accuracy is not high.Therefore,this study uses hyperspectral imaging technology to evaluate the comprehensive quality grade of Hanfu apple.The results are of great significance for improving the postpartum grading level of Hanfu apple.The key achievements are contained in this research as follows:(1)In this study,the apple image is transformed from the RGB color space to the HIS color space,the H component is extracted,the coloring area is calculated using the pixel point conversion method,and the coloring degree is calculated.The apple image is preprocessed,the center of mass of the apple outline is found and the diameter is calculated.The minimum circumscribed circle and the largest inscribed circle of the apple outline are obtained,and the area ratio of the two circles is calculated to obtain the fruit shape index.(2)This study examined the integrity of apple fruit stems.And the spectral information of the region of interest was extracted by hyperspectral imaging.The Stepwise multiple linear regression(SMLR)was used to extract 5 characteristic wavelengths from the whole wavelength band(450~970nm),and the successive projections algorithm(SPA)was used to extract 7 characteristic wavelengths from the whole wavelength band(450~970nm).And then the 4 texture features of energy,entropy,moment of inertia and correlation were extracted in the region of interest.The three sets data of spectral characteristics,texture features and spectral characteristics combined with texture features,respectively,which were as the input vector of support vector machine(SVM)and BP artificial neural network(BPANN),were used to identify the integrity of apple fruit stem.The experimental results show that only using the spectral features as the input vector recognition effect is better,which SPA-SVM method to identify the best,recognition accuracy rate of 91.7%,and the data calculation is small.This research provides the theoretical basis for apple quality grade evaluation.(3)This study proposes the evaluation of the application of the fuzzy membership function to the quality of apples.The input,size,fruit stem integrity,disease defect,hardness,fruit shape index,coloring area and sugar index were used as input vectors to establish decision tree C4.5,TS fuzzy neural network(TS-FNN),and fuzzy least squares respectively.Support Vector Machine(FLSSVM)Model to Assess the Comprehensive Quality Level of Hanfu Apple.The decision tree C4.5 model has the best discriminant effect,and the correct rate of the verification lumped is 95.47%.Among them,the rate of discriminate veracity is 100%,that of the first grade is 95.5%,and that of the second grade is 94.%,the exact rate of determination of the cut fruit is 92.4%.The discriminant effect of FLSSVM model is the second highest,and the correct rate of verification lumped is 93.6%.The experimental results show that the decision tree C4.5 model has certain advantages in multi-index decision making and can adapt to the small changes in data.