Abstract
(Objective) Fruit classification remains a challenge because of the similarities involved by a large quantities of types of fruits. With the aim of recognizing fruits accurately and efficiently, this paper offered a novel fruit-classification tool. (Method) The proposed methodology consisted of following four processes: (i) A four-step preprocessing was performed. (ii) The color, shape, texture features were combined. (iii) Principal component analysis was employed for feature reduction. (iv) We presented a novel classification method with the combination of "Hybridization of PSO and ABC (HPA)" and "single-hidden layer feedforward neural-network (SLFN)", which was termed as HPA-SLFN. (Results) The experiment results demonstrated that the proposed HPA-SLFN achieved an 89.5% accuracy that was superior to existing methods. (Conclusion) The proposed HPA-SLFN was effective.