Firmness And Damage Detection Of Zucchini Based On Visible/Near Infrared Spectroscopy And Hyperspectral Imaging Technology

The fruit of zucchini is cylindrical,and fresh and tender fruits are picked for vegetables.Zucchini is loved by consumers because of its thin skin,thick flesh,juicy and rich moisture.Greenhouse planting zucchini has also become an important source of income for farmers in some areas.Because of the quality problem of zucchini,the economic value can not be optimized,and then to achieve the fast classification and nondestructive testing of Zucchini quality has become the necessary prerequisite for the development of Zucchini planting industry in China.The purpose of this study is to provide a theoretical basis for the industrialization of Zucchini and the comprehensive evaluation of quality nondestructive testing indicators of zucchini.In this paper,the research object is zucchini in the harvest period.By using visible/near-infrared spectroscopy,hyperspectral imaging,chemometrics and physical-chemical analysis,the internal quality of zucchini is quantitatively tested,the slight damage of zucchini is identified,and the corresponding qualitative discrimination and classification model is established The quantitative analysis model of visible/near infrared spectrum and the mathematical discrimination model of hyperspectral imaging for quality index.The main results and conclusions are as follows:(1)The firmness nondestructive testing model of Zucchini was studied by diffuse reflectance near infrared spectroscopy.Firmness can be used as one of the important indexes to judge the maturity and quality of fruits and vegetables.Based on the spectral data of Cucurbita pepo samples,the PCR,SMLR and PLSR prediction models of Cucurbita pepo pulp firmness were established.The results show that the PLSR firmness prediction model based on convolution smoothing and standard normal transformation(S-G+SNV)is the best,the correlation coefficient of correction set is 0.979,and the correlation coefficient of prediction set is 0.976.The correlation coefficient is 0.886,and the root mean square error is 0.126.In this study,it is feasible to predict the firmness index of Zucchini by using visible/near infrared spectroscopy,which provides a theoretical basis for on-line nondestructive testing of fruit and vegetable firmness index in the future.(2)The damage detection model of Zucchini was studied by diffuse reflectance near infrared spectroscopy.In order to identify the slight damage of zucchini,the visible near infrared spectroscopy was used to detect the damage of zucchini.By using visible near infrared spectroscopy,combined with partial least square(PLS),principal component regression(PCR)and least support vector machine(LS-SVM),the damage prediction results of SNV+PLS model are 100%.The accuracy of SNV+PCR and SNV+LS-SVM reached 93.3% and 96.7% respectively.(3)By analyzing and comparing the influence of linear PLS,PCR and nonlinear LS-SVM,different spectral preprocessing and different modeling bands on the detection accuracy of quantitative analysis model of Zucchini firmness,it is concluded that the PLS model based on full band original spectrum is the most suitable for quantitative analysis of Zucchini firmness,at this time,RC,RMSEC,RP and RMSEP are 0.872,1.301n and 0 respectively 796,1.546n;the LS-SVM model established by the original spectrum is most suitable for quantitative analysis of Zucchini firmness.At this time,RC,RMSEC,RP and RMSEP are 0.885,1.229n,0.807 and 1.402n,respectively.The results show that it is feasible to use hyperspectral technology to predict the firmness index of zucchini.(4)Based on hyperspectral imaging technology,a slight damage discrimination model of Zucchini was established,and a partial minimum support vector machine damage discrimination model of full band and characteristic wavelength was established through different preprocessing.The results show that the damage detection with characteristic spectrum including key spectral information has better robustness,and the prediction result of damage model reaches 83.3%,which is conducive to the application of the model in online damage detection.