Development of a prototype for classification of potato mini-tubers based on artificial vision

This article presents a methodology that was followed for the design and implementation of a new artificial vision system for the grading of potato tubers. This system uses hardware devices and algorithms used in machine vision applications. This work consists of two main stages. The first stage describes the hardware elements used for image acquisition (lighting, camera and optics) and the algorithms for image processing implemented to carry out the potatoes size (diameter) measurements. The second stage describes the mechatronic system used for classifying the potatoes and the user interface developed to run the system. The result of this work is a prototype for measuring and grading with the capability to classify 750 potatoes per hour with an accuracy of 2/10 mm.