Document Type : Research Article
Associate Professor, Department of Biosystems Engineering, Faculty of Agriculture and Natural Resources, University of Mohaghegh Ardabili, Ardabil, Iran
PhD Graduated, Department of Biosystems Engineering, Faculty of Agriculture and Natural Resources, University of Mohaghegh Ardabili, Ardabil, Iran
Regarding this fact that peroxidase (POD) activity is considered as one of the important qualitative parameters of apple fruits, in this study, the effect of cold storage on POD activity of Red Delicious apples were investigated during 60 days. Hyperspectral reflecting imaging in range of 400-1000 nm has been applied while POD of samples were measured according to standard methods. After discarding noises using principal component analysis (PCA), to improve spectrum, different primary pre-processing had been applied and their effects were investigated. The suitable model was obtained via Partial Least Square method (PLS). Important wavelengths were selected based on regression coefficient of the best model includes large absolute values of weighted regression coefficients (RC) and sequential predictions algorithm (SPA) and using various techniques were modeled. Concerning the PLS analysis, the best results were obtained through smoothing Savitzky-Golay pre-processing with mean square root error (RMSE) of 0.475 and 0.518 and coefficient of determination (R2) of 0.948 and 0.940 for calibration and validation data, respectively. According to RC and SPA, 9 wavelengths were determined as the best. In modeling by efficient wavelength, artificial neural network (ANN) and SPA Combined Model gave the best result. The results indicated that hyperspectral imaging could be considered as a valuable tool for POD activity prediction and the selected wavelengths could be potential resources for instrument development.
- Near-infrared hyperspectral imaging was used to evaluate peroxidase activity in ‘Red Delicious’ apples during cold storage.
- The influence of different Pre-processing method on PLS model were studied.
- Effective wavelengths for POD activity discrimination of apples were selected based on regression coefficient (RC) of best PLS model and sequential prediction algorithm (SPA).
- Different regression models created with effective wavelengths effective wavelengths and their performance was compared.