The study on increasing the identification accuracy of waxed apples by hyperspectral imaging technology

Abstract

Hyperspectral imaging technology is applied to nondestructive quality determination of agricultural and food products. It has a greater advantage of combining spatial image and spectral measurement which can determine both external and internal quality of the product. To increase the classification accuracy and stability of the prediction model, the spectral correlation analysis of each pixel was used to determine quality of the sample’s hyperspectral image in this study. 400 hyperspectral image ROIs were extracted from 40 apples (20 apples with waxed and the other 20 apples without any waxed) were studied. Two effective wavelengths (806.85 nm, 1073.97 nm) were screened by spectral correlation analysis of pixels in 7 peak wave bands (639.90 nm, 806.85 nm, 973.80 nm, 1073.97 nm, 1197.99 nm, 1269.54 nm, 1441.26 nm). When the spectral correlation degree is less than 0.9 among all the pixels in the same sample, this sample data should be eliminated before inputting into the model training group. The least squares support vector machine (LS-SVM) model and BP neural network were used to establish the classification model between the hyperspectral image and waxed situation. The prediction result showed the classification accuracy was increased from 82% to 100% when the low-quality sample data for training were filtered by spectral correlation analysis. By evaluating the quality of the hyperspectral image measured, more reliable prediction results can be obtained, which can make the noninvasive discrimination of food safety come to the practice application sooner.