Inspection On The Cover Factors Of Staple Woven Fabric Based On Image Processing

The cover factor is a basic structural parameter closely related to final use and performance requirements of fabric. The cover factor of fabric has effects on fabric quality, handle, air permeability, dyeing ability and anti-ultraviolet radiation property. The uniformity of cover factors reflects the uniformity of yarn arrangement in fabric. And it is related to the stability of fabric structure, protective effect or permeability. Traditionally, the test of cover factors is based on theoretical formula and are calculated by yarn diameter and fabric counts. However, the result of this method is not in accordance with the reality. The uniformity of cover factors can’t be described intuitively and be quantified by theoretical formula.At present, computer digital image processing technology is widely used in textile testing. In this paper, this technology was used to deal with staple fabric images to extract features such as cover factors, degree of pore rectangular. Using this technology to detect cover factors and its uniformity could decrease the influence of subjective factors, increase date accuracy, reduce the complexity of computation and provide possibility of automatic detection. The application of this technology in batch testing could not only improve labor productivity but also guarantee the quality. In addition, the application of this technology could provide technical support for the research of products with excellent protective performance.Staple fabric images were obtained by the scanner and fabric counts and cover factors were extracted after tilt correction, local image segmentation, morphological processing, pore recognition and image fusion. Then, the image test results and theoretical value were compared analyzed to verify the feasibility of the image processing technology. By processing fabric images of different fabric count, the relationship between features such as cover factors of staple fabric,the uniformity of cover factors, degree of pore rectangular, and fabric count and structures was analyzed. The main contents and conclusions of this paper were as follows:(1) Image was obtained by scanning a kind of plain weave fabric., The tilt degree of warp and filling yarns was analyzed by Radon transform and tilt correction was made in order to facilitate the subsequent processing of image.(2) Fabric images were segmented by threshold of iterative method, Otsu method and maximum entropy method. By comparing these three methods, maximum entropy method was chose to convert gray images into binary images. The centerline of yarn was calibrated according to the projection of binary images in both horizontal and vertical direction and then the fabric was divided into many local images. Each local image was segmented by threshold of Canny edge gray mean value and maximum entropy threshold respectively. By comparing the segmentation effect of three different types of local images, the local threshold segmentation method based on maximum entropy was finally selected.(3) After image segmentation, the binary image was processed by closing operation to fill the black pores. Then false pores were removed by area comparison method and the binary image with true pores was obtained.(4)The complete fabric binary images were obtained by mixing a number of local binary images and cutting off the areas outside of lateral yarns. The edge of pores were marked by contour line. The result showed that the pores contour lines of binary image were well matched with the edge of real pores.(5) The fabric counts, total cover factors, the unevenness of cover factors and the average degree of pore rectangle were extracted from the binary images. The image test results and theory calculating results were compared and the conclusion was that the method proposed in this paper had feasibility and superiority.(6) Nine different plain, twill and stain fabrics with different fabric counts were prepared with the same cotton yarn. The images of fabrics were collected and then images were treated as follows: tilt correction, local image segmentation, closing operation, pore recognition and image fusion. Cover factors, unevenness cover factors and degree of pore rectangle were extracted. The relationship between the fabric structure and the fabric counts and the cover factors was studied and the reason was analyzed.