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

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 ${H} _{p}$ 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 H-p and the H 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.