Discrimination On Maturity Of Plums Based On Hyperspectral Imaging Technology

Plums, exquisitely carved, shape beautiful, sweet taste, is one of the most popular traditional fruit. It can be made into canned food, fruit, and high medicinal value. In this paper, fruit from the orchard of Shanxi Taigu country as the research object, Nondestructive detection of plum maturity based on hyperspectral imaging technology. In order to provide the theoretical basis for the nondestructive detection and discrimination of plum maturity.The main research contents and results are as follows:(1) Spectrum information (420-1000nm) of plum fruit sample was collected based on high spectrometer of different mature degrees of plum fruit (unripe, ripe, mature and overmature) samples were based on the full spectrum, principal component analysis and characteristic wavelengths of the PLS modeling analysis. The correction set and forecast effect of the plum maturity under each model were obtained and compared with the results. The model based on the full band spectrum has the highest accuracy. However, considering the experimental computation and complexity, Based on the characteristic wavelength of the model to determine the accuracy of the optimal.(2) SSC (soluble solids content) and hardness values of plum samples were analyzed by SPSS statistical software, Eliminating abnormal sample and carries on the Shapiro-Wilk normality test. Indicates that the two factor is representative of the maturity of the sample. Then the PLS model of plum maturity is established based on the two factors. Results show that the highest accuracy rate is based on the hardness value, Reached 84.5%,75.6%,91.70% and 96.5%.(3) Image information of plum sample is collected based on hyperspectral image, median filtering denoising for image. RGB and HSV color image model of different components by using Matlab software (average) and standard deviation as the color feature value, And establish the RGB?HSV and RGB-HSV color characteristic value of the plum maturity PLS discriminant model and make correction and prediction. The results show that the accuracy rate of discriminant model based on RGB-HSV color feature value is better than that of RGB and HSV, The accuracy rate reached 98.35%?90.00%?85.85% and 90.85%?...