Mass estimation of mango fruits (Mangifera indica L., cv. ‘Nam Dokmai’) by linking image processing and artificial neural network

• Main Authors: Katrin Utai, Marcus Nagle, Simone Hämmerle, Wolfram Spreer, Busarakorn Mahayothee, Joachim Müller
• Format: Journal
• Published: 2018
• Subjects: Agricultural and Biological Sciences; Chemical Engineering; Engineering
• Online Access: https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85055892420&origin=inward; http://cmuir.cmu.ac.th/jspui/handle/6653943832/62542
• Institution: Chiang Mai University

© 2018 Asian Agricultural and Biological Engineering Association Computer-aided estimation of mass for irregularly-shaped fruits is a constructive advancement towards improved post-harvest technologies. In image processing of unsymmetrical and varying samples, object recognition and feature extraction are challenging tasks. This paper presents a developed algorithms that transform images of the mango cultivar ‘Nam Dokmai to simplify subsequent object recognition tasks, and extracted features, like length, width, thickness, and area further used as inputs in an artificial neural network (ANN) model to estimate the fruit mass. Seven different approaches are presented and discussed in this paper explaining the application of specific algorithms to obtain the fruit dimensions and to estimate the fruit mass. The performances of the different image processing approaches were evaluated. Overall, it can be stated that all the treatments gave satisfactory results with highest success rates of 97% and highest coefficient of efficiencies of 0.99 using two input parameters (area and thickness) or three input parameters (length, width, and thickness).