Discrimination of Golab apple storage time using acoustic impulse response and LDA and QDA discriminant analysis techniques

Firmness is one of the most important quality indicators for apple fruits, which is highly correlated with the storage time. The acoustic impulse response technique is one of the most commonly used nondestructive detection methods for evaluating apple firmness. This paper presents a non-destructive method for classification of Iranian apple (Malus domestica Borkh. cv. Golab) according to the duration of storage. Several data preprocessing methods were tested: normalization, detrending, Savitzky-Golay smoothing, first and second derivatives, multiplicative scatter correction, standard normal variate and moving average. It was observed that the maximum average Fβ value of classification on the test dataset (0.84) belongs to non-preprocessing. In this study, principal component analysis (PCA) technique was performed to determine the key variables that explain most differences in the spectra. Seven principal components were used to calibrate linear discriminant analysis (LDA) and quadratic discriminant analysis (QDA) classifiers. The classification accuracy for LDA and QDA models were about 80.56% and 83.33%, respectively. The results indicated that the acoustic impulse response method is potentially applicable for the detection of apple firmnes