Iterative extraction of yarn positional information from three-dimensional image of textile fabric

In this paper, a novel method for analyzing a textile fabric structure is proposed to extract yarn positional information of the textile fabric from three-dimensional (3-d) image made out of its X-ray computed tomography (CT) images. In order to extract the positional information, directions of filaments, which the yarn consists of, on all the voxels are firstly estimated by correlating the voxels of the 3-D image with a filament model. Each filament is secondly reconstructed by clustering the voxels of the filaments using inner product between the estimated filament directions as similarity. Then, each yarn is segmented by clustering the reconstructed filaments using a distance between two filaments as dissimilarity. The each yarn positional information is finally extracted by iteratively calculating representative points of the segmented yarn and segmenting the yarns using them. The effectiveness of the proposed method is discussed by experimentally applying the method to 3-D image of a sample plain woven fabric, which is made of a polyester two folded yarn.