Development of automated real-time mango grader using machine vision technique

Abstract

Grading of agricultural produce, especially mango, is a significant factor in Bangladesh, crucial for maintaining quality standards in domestic and international markets. This study aims to construct an automated conveyor system with a machine vision system to acquire moving mango images. An image processing algorithm was developed to extract the area feature from the captured moving mango images and optimized an ejection system for real-time mango grading based on the extracted area feature of the mango. The images were acquired using image acquisition, and a size-based algorithm developed using MATLAB-GUI, which was both simple and successful in categorizing mangoes into three sizes based on area, achieving accuracy for Large (83.7%), Medium (85.5%), and Small (79.7%). The ejection accuracy was achieved for Small-Medium (83.9%), Small-Large (82.8%), and Large-Medium (90.1%). Finally, when all the mango samples were together, the ejection system achieved an accuracy of 89.1%. The automated system demonstrated its potential, achieving accuracy in the grading of mangoes. The prototype’s have shown the feasibility and potential benefits of future mechanization in mango grading processes in Bangladesh.