Development of a computer vision based Eggplant grading system

The popularity of automated fruit and vegetable grading has increased in most of the countries because the businessmen can make more benefits from graded fruit and vegetable than without graded fruit and vegetable. The business dealers in Bangladesh distribute and sell fruits and vegetables after manually grade or without grade. Thus, both sellers and customers are deprived everyday in Bangladesh. This paper presents an automated computer vision based Eggplant grading system. After image acquisition, median filtering is used to remove noise from the RGB image. The diseased effected areas of the Eggplant are segmented using Otsu and Binary Transformation methods. The size, shape, color and percentage of diseased area features are extracted using gray level co-occurrence matrix to construct the template of the healthy and unhealthy Eggplants. Four categories such as healthy, partially defected, moderately defected and unhealthy are considered in this research. The KNN method is used to classify the Eggplants into its respective grade. It is recommended to use this automated fruit and vegetable grading system because the remarkable classification success rate is 88% and hence, the customers will get good quality products as well as the satisfaction of customers will lead a successful business.