Acoustic detection of apple mealiness based on support vector machine

Abstract

Mealiness degrades the quality of apples and plays an important role in fruit market. Therefore, the use of reliable and rapid sensing techniques for nondestructive measurement and sorting of fruits is necessary. In this study, the potential of acoustic signals of rolling apples on an inclined plate as a new technique for nondestructive detection of Red Delicious apple mealiness was investigated. According to destructive confined compression tests, the mealiness of apples was evaluated by the hardness and juiciness measurements. In addition, support vector machine (SVM) models were developed to classify apples. The radial basis function (RBF) as the kernel was used in SVM models. According to exhaustive search method, the model with nine features combination was found to be the best model. Results indicated overall accuracy of 85.5 % to classify apples in mealy and healthy groups. The results indicated that this method is potentially useful for apple mealiness detection.