Abstract:

To replace time-consuming and lab-intensive of manual inspection, a Fourier transform is proposed to detect the thread densities of woven fabric in this paper. First, theories of the Fourier transform, yarn image reconstruction and the threshold method are introduced. Then the steps of fabric image acquisition, the Fourier transform of the fabric image, feature analysis in the frequency domain, image construction of fabric yarns, and threshold processing are analysed. Lastly, after locating and counting the yarns in the fabric segmentation results, thread densities of the woven fabric are calculated. The experimental results prove that the method proposed can detect the thread densities of woven fabric correctly and it can be used to replace the current manual analysis.