Journal of Agricultural Machinery Engineering ISSN:2799-8673(Print) 2799-8819(Online) 농업기계공학

In this study, we aimed to examine the non-destructive measurement technologies for banana firmness, and two methods, such as digital image-based technology and compressed air-based deformation measurement technology, were used to nondestructively predict the banana firmness. A total of 130 mature green ‘Cavendish’ bananas imported from the Philippines were used as sample, and were stored in a incubator with temperature of 20℃ and relative humidity of 90%. The firmness of banana was measured using destructive method after taking digital image and deformation of bananas, and ten bananas were used for each experiment. L*a*b* color space of banana surface was extracted from digital images using image processing, and the deformation of banana was measured using a deformation acquisition system. L*a*b* color space and deformation data of bananas were used as input data to develop banana firmness prediction model based on the support vector regression (SVR). As a result, the prediction models based on L*a*b* color space and deformation achieved accuracy with R2 values of 0.617, 0.764 and RMSE of 12.518 N, 9.827 N, respectively. Therefore, it is considered that the technologies can be used to non-destructively predict the banana firmness. Nevertheless, it is necessary to carry out more experiments to improve the prediction accuracy in the future.