A backlighting method for accurate inspection of woven fabric density

Abstract

To inspect the yarn-dyed fabric density automatically, an efficient backlighting method is proposed in this paper. This method consists of four steps: fabric image acquisition, yarn skew detection, image threshold segmentation and binary image projection. Firstly, it is utilized an imaging acquisition system with two reflective and transmission lights in its top and bottom placements, and the reflective and transmission images of the same region in the fabric are captured respectively by adjusting the illumination condition. Secondly, the skew angles of yarns in transmission image are obtained by using Hough transform to inspect them in reflective image. Then, the transmission fabric image is converted into binary image by Otsu algorithm. Finally, the projection curves produced from the binary image by the projection method are smoothed by the locally weighted regression algorithm. The number of peaks in the projection curve is counted and therefore the number of yarns is detected and the density is calculated precisely. The experimental results proved that the proposed method is effective for woven fabrics by testing eight grey and color fabric samples with different weave patterns.