Artificial intelligence and image processing based techniques: A tool for yarns parameterization and fabrics prediction

This paper presents a new innovative technological solution idea to automatically quantify the yarn mass parameters (hairiness, diameter and mass), the yarn production characteristics (snarls length, number of cables, fibres orientation and cables orientation) and the yarn surface porosity, as well as the yarn associated fabrics prediction, using Image Processing (IP) and Artificial Intelligence (AI) techniques. The presented approach suppresses the constraints of the traditional commercial testers used for yarn quality parameterization measurement, as it is characterized by its low cost, low weight, low volume, higher resolution and precision, high technological stability, reduced maintenance and lower hardware complexity, presenting the possibility of online use for control during production. Moreover, as a result of the superior resolution and elevated accuracy, the automatic determination of some new yarn relevant parameters will be introduced (e.g. protruding/loop fibres length and number, irregularities length, absolute number of cables and surface porosity, among others). Finally, the results of this project will establish, among other benefits for the textile industry, a new level of parameterization, allowing increased products' quality and superior efficiency, contributing to an economic recovery.