Low-Cost Framework for Textile Yarn Characterization Using Image Processing

The textile industry increasingly demands innovative and cost-effective solutions for yarn quality assessment, as conventional equipment is costly and occupies substantial space. This work presents a compact, low-cost image processing framework to characterize key yarn parameters, providing a foundation for future automated quality control systems. The framework employs classical image processing techniques—smoothing, thresholding, segmentation, and morphological operations—implemented with open-source tools such as Visual Studio and OpenCV. An experimental setup using low-cost hardware enabled the acquisition of high-quality images under controlled conditions. The system extracted parameters including linear mass, average diameter, specific volume, defect quantification, hairiness coefficient, and twist direction and pitch. Tests on three yarn types (cotton and polyester) produced results comparable to the industrial reference Uster Tester 3, with error rates below 7%. The proposed solution offers an affordable alternative for small industries and research laboratories, with potential for future integration of advanced computer vision and artificial intelligence to enhance defect detection and classification.