Research On Identification Method Of Navel Orange Origin And Disease Based On Hyperspectral Imaging Technology

China is a big fruit producer.With the increase of people’s income,the demand for highquality fruits is becoming stronger and stronger.The methods of detecting the origin of the fruit and rapid disease detection are closely watched by consumers and processors.The quality of fruit directly affects the quality,grade and sales of processed products.Therefore,an efficient and non-destructive technical method is needed to detect fruit quality.Compared with traditional detection technology,the use of hyperspectral imaging technology to detect fruit quality has the characteristics of non-destructive and high efficiency.Taking navel orange as the research object,this article uses hyperspectral technology,combined with a variety of chemometric methods,mainly to identify the disease of the navel orange.The identification of the origin and the shelf life of navel oranges have been tested.The main research contents and conclusions of this paper are as follows:(1)Using hyperspectral technology to detect and identify different diseases of navel oranges.Taking normal navel oranges,Huanglong disease navel oranges,and Shar Pei navel oranges as the research objects,the collected sample data was analyzed by principal component clustering,and then GA and CARS algorithms were used to select the characteristic wavelengths of the original spectrum.PLS-DA and LS were used respectively.LS-SVM algorithm for modeling.The results of the experiment show that the CARS band screening method,the LS-SVM model established by the selected characteristic wavelengths has the best effect,improves the efficiency of model establishment and reduces the misjudgment rate,and can effectively classify and identify normal and diseased navel oranges,and the overall error The judgment rate is 0%.(2)Using hyperspectral technology to detect and identify navel oranges from different origins.Taking navel oranges from Jiangxi’s Gannan,Hubei’s Zigui,and Chongqing’s Fengjie as the research object,the principal component clustering analysis was performed on the collected sample data,and then GA and CARS algorithms were used to select characteristic wavelengths of the original spectrum,and PLS-DA,LS-SVM,ELM algorithm for modeling,the experimental results show that the GA band screening method,the LS-SVM model established by the selected characteristic wavelength is the best,the overall misjudgment rate is 15%,and the modeling speed and model are improved The predictive accuracy of this method shows great potential for the detection of navel oranges from different origins.(3)Using hyperspectral technology to detect and identify navel oranges with different shelf life.Taking navel orange as the research object,the samples were collected on the 0th day,7th day and 14 th day of the shelf life of the hyperspectral image,and the original spectral characteristics and image characteristics(including RGB,HSI color characteristics and Graylevel co-occurrence matrix features),the PLS-DA and LS-SVM models were established with spectrum,image,and mixed features.The experimental results showed that the effect of establishing the LS-SVM model with mixed features is the best and improves the correct identification of navel oranges with different shelf life.It can realize the accurate and effective classification and identification of navel oranges with different shelf life,and the misjudgment rate is 1.33%.The research results show that hyperspectral technology can better realize the detection and identification of navel oranges with different diseases and different origins,and discriminate navel oranges with different shelf life.It can provide a certain reference and basis for the external quality of fruit sorting,and it can provide a certain reference and basis for fruit external quality sorting.Sales,storage and deep processing companies have a certain degree of reference.