Detection Of Fruit Bruise,sugar Content And Shelf Life By Hyperspectral Imaging

It is important for the detection about external defects and internal quality,especially for the sale,processing and consumption of fruit.compared with the traditional detection technology,using hyperspectral imaging technique to detect the quality of fruit has the advantage of green,fast and efficient.In this paper,taking the yellow peach and pear as test object and detected the hidden damage and internal quality of yellow peach at the same time and did some detection and research for the internal quality of net pear and the shelf life of pear.The main contents and conclusions of this study are as follows:(1)The detection for the bruise and internal quality of yellow peachhas been realized simultaneously with hyperspectral imaging technology.Principal component analysis(PCA)and partial least squares discriminant analysis(PLS-DA)were used to take the qualitatively analyze for the damage of yellow peach and the result of qualitative discrimination was 94.6.At the same time,combined the normal samples spectrum and partial least squares(PLS)to establish the quantitative model and the quantitative model results of RMSEP and r_p were0.975 and 0.792 respectively.(2)From the original hyperspectral image of net pear extracted three specrum which belonged the plastic net,pear sample surface and sample and plastic mesh respectively.According to the difference of thee spectrum getting the best feature images and extraced the spectrum of pear suiface.The prediction model of sugar content was established by using images and spectra respectively combined the partial least squares support vector machine(LS-SVM)and PLS.The results show that the optimal results of RMSEP and r_p based on images were 0.43 and 0.61 respectively and the optimal models of RMSEP and r_p based on spectra were 0.3 and 0.78 respectively.(3)Taking pear as the research object,after the hyperspectral image acquisition for pear samples in the shelf life is the first day,fifth day and tenth day respectively,first of all,using PCA to process the hyperspectral image to obtain the image characteristics of the original pear,then calculated the image features average gray value of all samples by MATLAB,according to the average gray value of the characteristic image of experimental samples and the spectra extraction from the sample combined with LS-SVM and PLS-DA algorithm respectively to establish qualitative discrimination model of pear shelf life.The results show that the error rate in the model based on image and spectrum was 0.