Identifying Qualities And Varieties Of Postharvest Peaches And Nectarines By Using Dielectric Spectra/Near-infrared Spectra Technology

Peach as one of major fruit production in China, its production and consumption has allalong been in the front rank of the world. However, the Chinese peach is less competitive inmarket and its proportion in international trade has been low. Compared with developedcountries, the mainly problems were low domestic prices, domestic consumption-based, poorvisual quality, varying quality and other issues. All these results come from the backwardcommercialization of postharvest handling. Thus, it is particularly important to graspaccurately the fruit quality information, to learn fruit growing, to understand qualitycharacteristics change during picking and storage, and then to carry on rapid non-destructivetesting fruit quality research.In this study, adopting the coaxial open-ended probe technology based on networkanalyzer and Fourier transform near infrared diffuse reflectance techniques (FT-NIR), usingdifferent varieties of peaches and nectarines, their dielectric parameters (including relativedielectric constant ’, dielectric loss factor ’’) at the frequency range from20MHz-4.5GHz,their absorbance at wavelength range from800-2500nm, and their qualityparameters(including soluble solids content reflecting sugar content, moisture content,firmness and pH) were measured at25℃, and then quality testing models and recognitionmodels of different varieties of fruits were established in order to compare the twonon-destructive testing methods advantages and disadvantages in terms of quality testingpeach, comprehensively and systematically. Gains the following conclusions:(1)In the quantitative analysis of peach quality indicators, models based on near-infraredspectroscopy to predict the SSC of peach was better than which based on dielectric spectrum,specially, FS-BP model performed best, and its correlation coeffiencient of calibration (Rc),root mean square error of calibration (RMSEC), correlation coeffiencient of prediction (Rp),root mean square error of prediction (RMSEP) of the model were0.98,0.96,0.16,0.77,spectively.(2) Models based on dielectric spectrum to predict firmness of peach did better thanwhich based on near-infrared spectroscopy, specially, SPA-BP model performed best, and itsRc, Rp, RMSEC, RMSEP were0.83,0.93,1.99,1.04, spectively.(3) In the quantitative analysis of nectarine quality indicators, models based onnear-infrared spectroscopy to predict the quality parameters always did better than which based on dielectric spectrum; specially, UVE-LSSVM model performed best at SSCprediction, and its Rc, Rp, RMSEC, RMSEP were0.98,0.97,0.57,0.78, spectively;FS-LSSVM model performed best at firmness prediction,and its Rc, Rp, RMSEC, RMSEPwere0.90,0.91,2.54,2.18, spectively.(4) All models based on dielectric spectrum or near infrared spectrum can be accuratelyand quickly identified the varieties of peach; When discriminated peach varieties based on thedielectric spectrum, SPA method was the best method to extract information, and LSSVM wasthe optimal discriminant model; When based on the near-infrared spectroscopy, UVE was theoptimal information extraction method and LSSVM was the optimal discriminant model.(5) All models based on dielectric spectrum or near infrared spectrum also can beaccurately and quickly identified the varieties of nectarine; When discriminated nectarinevarieties based on the dielectric spectrum, full spectrum modeling performed best, and SPAwas the best information extraction method among three information extraction methods, andBP model was the best discriminant model; When based on the near-infrared spectroscopy,both UVE and SPA methods were optimal information extraction methods, moreover, bothELM and BP models were the best discriminant models.