Automatic measurements of partial cover factors and yarn diameters in fabrics using image processing

Abstract

Warp and weft cover factors and their mean yarn diameters were automatically estimated from microscope digital images of woven fabric samples. Regarding the periodicity of weaves, images were processed in the frequency domain. Frequencies were carefully analyzed to suitably crop the image to an integer number of complete yarn crossings in warp and weft directions, so as to eliminate possible bias due to fragmented yarn crossings, if any, in the original image. Warp and weft frequencies were identified in from the power spectrum of the cropped image by means of the Radon transform. Image filtering in the Fourier domain and binarization yielded partial cover factor estimates. The method was tested using a set of 81 diverse woven fabric samples and the results were consistent with their manufacturing specifications: yarn linear densities and thread counts. Finally, the mean yarn diameters were derived from partial cover factor estimates. To validate the proposed method the results were compared with the mean yarn diameters directly measured from images.