Research Of Persimmon During Shelf Life Based On Hyperspectral Imaging

Persimmons are called the King of fruit, but also have high medicinal value. In recent years, with the improvement of life level, people pay more attention to the development of persimmon fruit nutritional value. But as the persimmon fruit of raw and cooked in a concentrated period, after picking short-term easily variable factors of soft etc., which effect of persimmon fruit circulation, value, the interests of producers, which makes the persimmon lack of competitiveness in the market. Fruit shelf life refers to the end of the storage period of fruits, to enter the market and business links, and eventually had have circulation period. With manual detection of traditional time-consuming, labor-intensive, therefore, quick and nondestructive detection to achieve internal and external quality of persimmon fruit has become a necessary precondition of the development of persimmon fruit market.This subject to harvest persimmon fruit as the research object, through the hyperspectral imaging technology, chemometrics methods and so on .Through the persimmons’ shelf life of the qualitative analysis, at the same time of persimmon fruit internal quality of quantitative analysis, to establish the qualitative and quantitative analysis of classification model, on the basis of which, high spectral pre judgment model is established.The main research contents and conclusions are as follows:(1) With ZOLIX Gaia Sorter hyperspectral sorter (Hyperspectual Sorting system) hyperspectral image information collection of each sample set. Through the hyperspectral image technology of persimmon fruit shelf life for qualitative discrimination, using a variety of acquisition and processing of spectral data pretreatment methods, and then PLS discriminant, PCR discriminant analysis, LS-SVM discriminant linear and nonlinear pattern recognition method, to establish persimmon shelf-life discriminant model, the results show that, the pretreatment of first derivative and convolution take 1 point smoothing, modeling and analysis using stepwise discrimination method of LS-SVM effect is the most stable, and predict a better rate, pre judgment rate reaches as high as 96.94%.(2) With ZOLIX Gaia Sorter hyperspectral sorter (Hyperspectual Sorting system) spectra in hyperspectral image collection of each sample set, analyzing the internal quality of persimmon fruit quantitative, combination of various pretreatment methods to establish high spectrum image information of persimmon soluble solids content (SSC), tannin content and hardness analysis. Comparing with partial least squares regression model PLS, principal component analysis PCR and a number of linear chemical measurement model establishing method for soluble solids content (SSC), the detection performance of hardness and toughness of the epidermis, the result show that the best modeling model is using all band PCR.(3) The persimmon fruit soluble solid all band convolution smoothing analyte quantitation (Savitzky Golay, SG) the PLS model and PCR model are suitable for flat persimmon soluble solids (SSC) quantitative analysis. The SG built flat persimmon all spectrum PLS model Rc2, RMSEC, RP2, RMSEP respectively were 0.69114,0.8218,0.80249,0.5756, all spectrum PCR model Rc2, RMSEC, RP2, RMSEP respectively were 0.69166,0.8211,0.8349,0.5379, obviously the all spectrum PCR model is better than the all spectrum PLS model.(4) Quantitative analysis of persimmon tannin content all band convolution smoothing (Savitzky Golay, SG) quantitative analysis of the PLS model and PCR model are suitable for flat persimmon tannin content. The SG built flat persimmon all spectrum PLS model Rc2, RMSEC, RP2, RMSEP respectively were 0.70346,0.0440,0.79666,0.0263, all spectrum PCR model Rc2, RMSEC, RP2, RMSEP respectively were 0.71321,0.0412,0.79768,0.0263, obviously throughout the all spectrum PCR model is better than the all spectrum PLS model.(5) A all band SG analysis of persimmon fruit hardness quantitative (5)+Der built the PLS model of Rc2,RMSEC, RP2, RMSEP respectively were 0.87067,1.1845,0.92253,0.7667, a all band SG +Der model of Rc2, RMSEC, RP2, RMSEP respectively were 0.87630,1.1643,0.91193,.08387, by comparing both the built model is close to that of the quantitative analysis are suitable for flat persimmon tannin content.