Computer Vision System for Grading Mangos in Bangladesh

Authors

Md Abdul Momin, North Dakota State University--Fargo
Md Towfiqur Rahman, University of Nebraska-LincolnFollow
Mst Sabrina Sultana, Bangladesh Agricultural University
Igathinathane Cannayen, North Dakota State University--Fargo

Document Type

Abstract

Date of this Version

7-2015

Citation

Abstract

2015 ASABE Annual International Meeting (July 26 – 29, 2015: New Orleans, Louisiana)

Comments

Copyright 2015, the authors. Used by permission

Abstract

Mango (Mangifera indica) is one of the most common, important, and popular fruits among fruits produced in Bangladesh. In Bangladesh, mango was cultivated in about 0.078 million acres with an annual production of 0.8 million tonne. However, the technologies used in post-harvest processing (e.g., grading, sorting, etc.) of mango are mostly traditional involving human labors; and the methods are far from satisfactory, in terms of accuracy and throughput. In order to grade mango according to size, we developed a machine vision system with image processing capabilities that can measure sizes of mango fruits from the images. In this system images were acquired using a XGA format color camera (The Imaging Source: DFK 41AU02) of 8-bit gray levels using ten sets of fluorescent lamps. Results showed that the fluorescent lighting based machine vision system is practical and simple to use. An image processing algorithm based on color binarization combined with median filter and morphological analysis was developed, and the proposed algorithm is a simple and able to discriminate different sizes of mango (e.g., large, medium, and small) with high accuracy. The algorithm was implemented with all image sets using Visual C++ software and OpenCV (Open Source Computer Vision) library functions. Practical experience and results indicate that it is feasible to use the developed machine vision system or its modified versions successfully in commercial applications for grading mangos.