Fabric defect detection using wavelet decomposition

Abstract

Fabric defect detection is done by humans to assure textile quality traditionally. This paper proposes an automatic detection scheme using wavelet decomposition for the warp knitting machine. The following aspects are considered in dealing with broken-end defects caused by a single yarn. First, a wavelet transform is used to decompose fabric image, and an improved direct thresholding method based on high frequency coefficients is proposed. Second, a proper template is chosen in mathematical morphology filter to remove noise. Experimental results show that the scheme is effective and can satisfy the detection requirements for the warp knitting machine.