Abstract:Current methods for yarn hairiness evaluation generally adopt off-line destructive sampling, with unwinding the yarn from the package. In this article, we introduce a hairiness measurement system based on computer vision, which can be deployed on a cone transportation line and evaluate the hairiness of each cone nondestructively, showing significant advantages in statistical ability compared with current methods. The system consists of a yarn cone transportation module, an image acquisition module, and a data processing module. With the cooperation of the yarn cone transportation and image acquisition modules, the cones can be grabbed to a specific location and then rotated in multiple directions for tangential image capturing. The data processing module implies a hairiness detection algorithm and calculates the hairiness index. The hairiness detection algorithm consists of a filtering algorithm for fuzzy hairs based on the Frangi filter, and a double-threshold hair segmentation algorithm, with self-adaptive parameter optimization strategies for both parts to ensure adaptability. The scale invariant hairiness index is defined as the ratio of total hair length to the cone edge length. In the experiment, the superior precision of the proposed algorithm is demonstrated by a quantitative comparison. The correlation coefficient between the proposed and the index tested by a commercial instrument achieves 0.92, verifying the effectiveness of the proposed system. The stability experiment confirms that the system can adapt to a wide range of yarn varieties, package sizes, and image acquisition conditions. In addition, in the application scenario simulation experiments on spinning and winding process parameters, the system can provide hairiness measurement results with sufficient distinguishability of different parameters, further demonstrating the industrial application value of the system. In conclusion, the proposed system shows potential significance for process management, quality control, and system diagnosis in intelligent spinning manufacturing.

Using Machine Vision to Determine Dyeing Capability of Nylon

An Unsupervised Method for Dominant Colour Region Segmentation in Yarn-Dyed Fabrics

A Novel Method for Weft and Warp Yarn Segmentation in Multicolour Yarn‐dyed Fabric Images

A Machine Vision-based Fabric Defect Detection Solution for Textile Production Industry Using Object Detection

Design of Yarn Quality Detection System Based on Machine Vision

A defect detection system of glass tube yarn based on machine vision

Advances in Dyeing Technologies for Polyacrylnitrile

A novel method based on near-infrared imaging spectroscopy and graph-learning to evaluate the dyeing uniformity of polyester yarn

Dyeing Creation: a Textile Pattern Discovery and Fabric Image Generation Method

A Novel Feature‐based Network with Sequential Information for Textile Defect Detection

Research and Realization of Dobby CAD System

A Deep Multi-Instance Neural Network for Dyeing-Free Inspection of Yarn Dyeing Uniformity

Machine-Learning-Based System for the Detection of Entanglement in Dyeing and Finishing Processes

Quality Prediction and Abnormal Processing Parameter Identification in Polypropylene Fiber Melt Spinning Using Artificial Intelligence Machine Learning and Deep Learning Algorithms

Automated vision system for fabric defect inspection using Gabor filters and PCNN

Detecting dyeing machine entanglement anomalies by using time series image analysis and deep learning techniques for dyeing-finishing process

Online Measurement of Sizing Yarn Hairiness Based on Computer Vision

Online fabric defect inspection using smart visual sensors

Machine learning assisted mechanical properties prediction of fine denier polyester fiber

Dyeing of advanced denim fabrics (blend of cotton/bicomponent polyester filaments) using different processes and artificial intelligence method

A Computer-Vision-Based Nondestructive Hairiness Measurement System on Yarn Cone Transportation Line