Precise Yarn Segmentation of Fabric Images via Active Grid

Wenya Feng, Li Liu, Yilin Guo, Zhixuan Zhou, and Huiping Sun


Image Processing, Segmentation and Representation, Textile Fabric, Yarn


In this paper, a novel method which is based on a variational model for precise yarn segmentation is proposed to segment each yarn of the textile fabric by an active grid. The first part of our method allows automatic weft yarns and warp yarns crossed-area segmentation through a spatial domain integral projection approach as previous research did. The second part uses an active grid to adjust the segmentation lines to accurately approximate the real contour of yarns. Each lattice of the active grid is a relative separate unit and every lattice can move separately. After all lattices stop moving, the final segmentation lines are indicated by the lattices\textquoteright edges. In the real world, most yarns in a fabric image are not absolute straight because of the fabric quality,production process and user actions, etc. So the segmentation lines should also have corresponding bend. Most of the state-of-the-art segmentation algorithms have difficulties in this challenging task. But based on an active grid, our algorithm achieves precise twisty segmentation lines and experimental results confirm its effectiveness.

Important Links:

Go Back