Naïve Screw Nut Classifier Based on Hu's Moment Invariants and Minimum Distance

Abstract

In this paper, an algorithm for classification of screw nuts by means of digital image processing is presented. This work is part of a project where a production line was built, and is focused on the quality assessment section. The algorithm presented classifies among good and poor quality screw nuts passing by a conveyor belt, by computing Hu's moment invariants of its picture. Those moment invariants are the input of a minimum distance classifier, obtaining very competitive results compared with some other classification algorithms of the WEKA plattform.