Nondestructive Prediction of Mechanical Parameters to Apple Using Hyperspectral Imaging by Support Vector Machine

This study proposes a method for predicting the mechanical parameters of apple after impact damage based on hyperspectral imaging with the range of 900–1700 nm. A damaged region detection algorithm combined with the principal component analysis (PCA) was developed and applied to the PC images, and the spectra were extracted based on the selected region. PCA and the successive projection algorithm (SPA) were used to select characteristic wavelengths. Nondestructive testing of mechanical parameters was conducted using the support vector machine (SVM) model which was established based on full-band and selected wavelengths. The results show that, based on PC5 image combined with threshold segmentation algorithm, the damaged region can be identified. The results of the SVM model based on characteristic wavelengths are relatively close to those of the model based on full-band wavelengths. The prediction results based on the full-band wavelengths show that the coefficient of determination and root mean square error of the absorbed energy, contact load, and damaged area are 0.890 and 0.218 J, 0.846 and 57.826 N, and 0.861 and 104.821 mm 2 , respectively. This study provides a reference for using hyperspectral imaging to analyze the mechanical properties of postharvest fruit and other agricultural products.