Visual Detection of SSC and Firmness and Maturity Prediction for Feicheng Peach by Using Hyperspectral Imaging

Feicheng peach is prone to spoilage due to its surface color changing rapidly after harvest, which will degrade its quality. Hyperspectral imaging technology was used to detect the soluble solid content (SSC), firmness and maturity of Feicheng peach for improving its quality and price. There were 80 maturity 70% and 90% Feicheng peach were used for hyperspectral images (400~1 000 nm), SSC and firmness collection, respectively. These samples were split into calibration set and validation set with a ratio of 2: 1 by samples set partitioning based on joint X-Y distances method after the outliers were eliminated by using Monte Carlo-partial least squares method. MLR detection models were established using feature wavelengths selected by competitive adaptive reweighted sampling (CARS) and successive projections algorithm (SPA), respectively. The more effective detection results was emerged by CARS-MLR model, with a determination coefficient of calibration set (Rc2) of 0.819 1, a determination coefficient of validation set (Rv2) of 0.843 9 and a residual prediction deviation (RPD) of 2.0 for SSC assessment, Rc2 of 0.951 8, Rv2 of 0.877 2 and RPD of 2.1 for firmness assessment. Visualization maps for SSC and firmness were generated by calculating the spectral response of each pixel on peach samples. Furthermore, the artificial neural network model was provided to predict the maturity of Feicheng peach using feature wavelengths selected by the sequential forward selection algorithm, with total recognition accuracy of 98.3%. It can be concluded that hyperspectral imaging technology can be applied to determine the SSC, firmness and maturity of Feicheng peach, laying a foundation for the on-line nondestructive quality monitoring and timely harvest of Feicheng peach. © 2020, Chinese Society of Agricultural Machinery. All right reserved.