Pilling Grade Evaluation Of Fabric Based On Digital Image Technology

Pilling grade evaluation of fabric is an important index to assess fabric quality. Now the grade evaluation method based on comparison of standard images is subjective and depended on inspectors'visual. The method based on image analysis is more objective and accurate. But it's still staying in the thesis research, so indices extracting, grade evaluation standard establishing and the automate grade evaluation of pilling fabric samples are discussed in this text based on the image analysis of standard pilling images.First, the image sampling method is given for standard and pilling fabric images. The standard pilling images of woven fabric are scanned by a flat scanner with a proper resolution, and the largest square within it is cropped for the pilling grade evaluation object. For fabric samples, YG401G automatic plain grinding machine is used to carry out the pilling experiment under the GB/T 4802.1-1997, also fabric images are captured by scanner.Then, theory of pilling image analysis is given in this text. It's difficult to distinguish the cyclical component representing the texture from the non-cyclical component representing the pills in space domain, in order to in favor of pilling separating, the texture in the fabric should be effectively filtered. So the Fourier Transform is used to translate the pilling image into frequency domain to get the amplitude spectrum, and filtering template is established by threshold which can filter the cyclical component representing the texture, but in the filtered image it has a little contrast between pills and background. So partial low frequency needs to be remained in the above-mentioned filtering processing which makes the pills prominent.In order to extract pills from the filtered image, enhanced local OTSU threshold is adopted. Pilling image after filtering is blocked by OTSU threshold after contrast enhancing. At last, the pilling binary image is got after threshold processing and noise eliminating. After then, grade evaluation standard is established based on the standard pilling images analysis. Through the analysis of standard images, pill number, pill total area proportion and pill max area are extracted, then the relations between pilling grades and these parameters are analyzed, pill total area proportion is selected as the grade evaluation standard.Last, texture filtering, pills extracting, pill total area proportion calculating, pilling grade evaluation are carried out according to the text above-mentioned for pilling fabric samples. And 25 fabric samples are taken into pilling experiment, image analysis and manually comparison of standard images method are used to evaluate pilling grade, we can see that it has a good consistency and the rate reaches at 84% between subjective and objective evaluation results, also the grade difference to be 1 grade. The result shows that the pilling grade evolution method based on image analysis proposed in this paper is more objective and rapidly, which can substitute current manually comparison of standard images method.