Taste characterization of orange using image processing combined with ANFIS

Alternative methods for taste evaluation of fruits are of great interest to the food industry. This paper deals with the application of ANFIS-based image processing to characterize orange taste. For this purpose, images of 300 orange samples (Bam, Khooni and Thompson varieties) were acquired using a camera and the relevant features were extracted. The features were RGB component, HSV component, texture features, major and minor diameters, area, circumference, R/G and R/B color component ratio and diameter ratio. A sensory test performed by ten panelists was used to provide the reference data for image analysis. Then, the features were entered as input to ANFIS, and taste class of the fruit was the output of ANFIS. Based on the results, the success rate for the taste classification of Bam, Khooni and Thompson orange varieties were found to be 96.6%, 93.3% and 93.3%, respectively.