Vision-based portable yarn density measure method and system for basic single color woven fabrics

Abstract

The use of machine vision technologies to replace manual methods of detecting woven fabric density has accelerated significantly in recent decades, but most previous studies focus on desktop devices that are unsuitable for commercial field applications. In this paper, a vision-based portable yarn density measure system and a corresponding measuring method are proposed. The proposed system requires fabric images to be acquired manually using the smartphone. As the distance and the pitch, roll, and yaw angles between the sample plane and the smartphone plane are arbitrary, and the resolution will vary according to the smartphone's properties , image calibration and distortion correction must be carefully considered. So a known size square is marked on the top surface of the background light for dimensional calibration, and dynamic image distortion is handled by calculating the geometric relationships between the four edges of the square. A detailed process for obtaining high-quality fabric images is described, and a discrete Fourier transform is adopted to calculate the density of the woven fabric . Finally, large numbers of samples are tested. Good agreement between the measured and calculated results proves that the system is precise and robust enough to meet the requirements of the fabric market.