An Investigation of Structural-Mechanical Properties of Spun-Bonded Non-Woven Using Computer Vision Method

This paper aims at the measurement of surface uniformity, thermally-bonded points, distribution of fibers orientation and local displacement in tensile testing for spun-bonded nonwoven polypropylene fabrics. For this purpose, an image processing method was used to produce clustered images based on the k-means clustering algorithm along with Davies-Bouldin index and the PSNR image quality evaluation method. Then, the quadrant method for surface uniformity, an image processing method based on morphological operators for uniformity of thermally-bonded points, the regionprops function (RF) method for distribution of fiber orientation and the digital image correlation (DIC) method for local displacement were used to calculate the parameters of nonwoven samples. Also, the relationships between image processing and the experimental results of tensile tests were studied. The results indicated that the structural properties of a fabric, such as surface uniformity, bonded structure, distribution of fiber orientation and critical points, have great impacts on its tensile properties at the selected weights and non-uniformity levels. Hence, a sample with a higher level of uniformity and, consequently, more regular bonding points with a higher bonding percentage, better distribution of fiber orientation and less critical points offers the best tensile properties.