| The maturity of apple is an important part of growth period of fruit quality,analysis of op tical properties SSC based on spectroscopy and imageing technology has an important signific ance for monitor scientific production of apple orchard and postharvest classification.Near inf rared spectroscopy and imaging technology are used to collecting spectrum and image informa tion of Xinjiang Fuji apple,analysis of geometry and optical character of apple maturity,the model were establish for predicting the soluble solids content based on the image and spectral information.The research contents and conclusions are as follows:(1)Analysis of the change laws of geometric and optical properties of apples at maturity.The gray mean value and the standard deviation of R,G,B,Y and S component grayscale ima ges after masked by binary image,as well as the geometric information of apple,were analyze d and obtained after done by preprocessing apple image,and the conversion between the color space of RGB,YUV and HSV.The trend graphs of the characteristics of the apple maturity w ere also drawn.The results show that the geometrical characteristics of apple,including theare a,perimeter,length of the long axis and minor axis of apple region,were not changed signific antly at the maturity,changes are mainly concentrated in the transformation of internal nutrien ts;and the gray mean value and the standard deviation of R,G,B,Y and S component graysca le images were obviously change during the maturity,and that of the red and saturation images shown a significant growth trend while that of green,blue and brightness images shown a signi ficant decrease.The change of the apple in the mature period was mainly for the performance of the internal nutrient with increasing redness at 0°,20°and 30°and with significant decreasin g yellow-green at 60°,70°and 80° in days of maturity.(2)Prediction model for apple soluble solids content based on spectral information MSC?SNV and other methods were used to preprocess the original spectra,and the partial leaet squa es method was used to compare the results of five kinds of wavelengthselection results,such as correlation coefficient method:UVE,SPA,etc.The results show that the spectral pretreatmen t of the 15-point moving window smoothing method is better than others for predicition model s,the wavelengths selection results of GA-PLS is better used to calculate soluble solid content of apples,calibration the correlation coefficient,root mean square error of calibration and pred iction correlation coefficient and root mean square error of prediction are 0.9472,0.7018,0.95 42,and 0.5482.(3)Prediction model of soluble solids content of apple based of image information.The co rrelation between gray mean value and the standard deviation of the grayimage of each compo nent from different color space were compared,these features can be applied to predict the sol ube solids content of the apple.the calibration set andthe prediction set are divied by these sel ection methods,such as: Rank-Sel,SPXY,KS,Rank-KS and Duplex.The prediction model of was established by using partial least squares regression and support vector machine.The resul ts show that the prediction results are better for soluble solids content of apple when the set of the correction and prediction obtained by the set division ofduplex algorithm.Model of partical least squares regression and support vector machine are established by these feature of gray m ean value and the standard deviation of the gray image of each component from different color space.The correlation coefficients of the prediction set were 0.9343 and 0.9372. |
PDF Full Text Request