Improved fiber orientation measurement in nonwovens with corner removal

Abstract

The orientation of fibers or filaments in nonwovens is critical in determining their mechanical characteristics. Image processing techniques, prized for their minimal human intervention and rapid processing speed, are widely utilized in nonwoven fiber orientation measurement. However, these techniques often face substantial challenges, such as low accuracy in corner detection, errors in fiber segmentation, and inefficiencies in fiber orientation calculation. Addressing these concerns, this study introduces a novel, enhanced method accompanied by two innovative optimization algorithms to enhance accuracy. The first innovation involves the development of a newly developed fiber corner detection algorithm, dubbed the T-detector, specifically tailored for the unique characteristics of fiber images, enabling efficient corner point detection and removal. Subsequently, we introduce and employ a fiber length restriction algorithm to further segment the processed longer fibers into the remaining fiber fragments and utilize a skeleton projection algorithm to calculate the fiber orientation. These algorithms overcome the existing technology’s inherent shortcomings, thereby heightening measurement accuracy. The results illustrate an improvement in measurement precision over other orientation distribution measurement algorithms, with the fiber information retention (covering ratio) reaching an impressive 95%. Our proposed method not only calculates fiber orientation distribution in nonwovens with remarkable accuracy and efficiency, but its innovative approach also stands to provide a theoretical foundation for the design of three-dimensional filtering models with specific fiber orientation.