Research On The Fast Intelligent Analysis System Of Woven Fabric Structure Parameters

The fast intelligent analysis system of woven fabric structure parameters is a computer system which is developed by digital image processing,pattern recognition and artificial intelligence technology. Using this system can improve the analysis efficiency and result accuracy of textile fabrics, reducing labor intensity. In this paper, the research on the structure parameters of the textile fabric is carried out in-depth analysis and research, the main work and innovation are as follows:(1) The skew rectification of fabric image has an important influence on the location and recognition of the tissue points. Firstly, the paper uses the vertical and horizontal edge detection to detect the edge of fabric image, and then uses the Hough transform to detect deflection. In order to improve the efficiency of the Hough transform, the edge extraction is used.(2)Warp yarn cannot reach vertical status or weft yarn cannot reach level status,because fabric images were acquired by artificially placed or the level of manufacturing. After the skew rectification for fabric image, we can get warp yarn vertical but weft yarn not be the level state or the others.Set warp yarn vertical an example. At many times, weft yarns are not present at the level status when warp yarns are vertical. We can not directly use gray projection method to automatically locate weft yarn. Firstly The Hough transform is used to detect weft skew angle, then the gray projection is used to obtain weft locations.(3)A new method for obtaining the position of warp and weft yarns was proposed based on vertical and horizontal projection, which is based on the vertical and horizontal projection curves.(4)After the yarn segmentation, some eigenvalues can be obtained to recognize the fabric pattern correctly. K-means algorithm is used to cluster the type of woven pattern.(5)In the processing of yarn dyed recognition, based on RGB color space to Lab color space, K means algorithm is used to cluster the fabric image color and get the coordinate of color clustering center. Combined with the fabric image pixel coordinates, color unit can be gotten on the basis of clustering of the dominant law amendment. Finally, the layout of yarn dyed is recognized based on the result of colorunit and fabric pattern.According to the color yarn type and the recognition of the organization chart to complete the identification of the fabric.The method proposed in this paper can be applied to different size, resolution,color, structure parameter of fabric image. The result of recognition has high accuracy,reliability and algorithm efficiency, so it has considerable application value and theoretical significance.