Research On Establishing Method Of Yarn Coordinate System For Flexible Fabric Structure

In the automated processing and inspection applications based on machine vision technology in the textile industry,there are some problems such as large pattern recognition errors which are caused by illumination interference and fabric deformation,low dimensional measurement accuracy and low template matching accuracy.Aiming at such issues,this paper proposes a binarization method based on illumination pre-compensation in order to solve the problem of poor image segmentation caused by uneven illumination for the first time.At the same time,this paper proposes a method of establishing flexible coordinate system by analyzing the structure of fabrics,which can restore the deformation graph of fabric and solve the problems.The binarization algorithm based on illumination pre-compensation needs to obtain the whiteboard image under the current illumination condition firstly.Then the method needs to calculate the illumination distribution intensity of different regions in the camera field through the whiteboard image and uses a weight value to represent light intensity in each small region.Then,the method uses the calculated light intensity weight to process the corresponding area on the image after the camera acquires the image in order to realize "light compensation" for the image.Finally,it calculates the threshold using OTSU algorithm in order to split the image.Practice has proved that the method can adapt to a variety of harsh lighting conditions to achieve adequate segmentation of image targets and backgrounds.The yarn coordinate system is a flexible coordinate system established by the fact that the warp and weft yarns are interlaced and the relative positional relationship is unchanged in the fabric image.The abscissa of a point on the coordinate system indicates the warp number of the point,and the ordinate indicates the weft number of the point,and the origin of the coordinate is the intersection of the first warp and the first weft of the upper left corner of the image.In this paper,the image refinement technique is used to obtain the fabric skeleton diagram.Then,according to the characteristics of the unit pixel width of the yarn in the skeleton diagram and the judgment conditions of adjacent pixel connectivity,a yarn extraction algorithm is designed and implemented,which can extract the information of the warp and weft yarns from the skeleton diagram.Finally,the pixel coordinate system of the warp and weft yarns can be transformed into the yarn coordinate system using the information of the warp and weft yarns in the pixel coordinate system,and the yarn coordinate system is established.The yarn coordinate system proposed by this paper has a flexible character that can change as the fabric deforms,and it can adapt to various harsh and uneven illumination conditions.No matter how the fabric is stretched and deformed under external forces,the coordinate values of specific points on the fabric are invariable,and the original information of the deformed pattern can be restored.The problems of large error in machine vision recognition,low accuracy in dimension measurement and low accuracy in template matching of flexible fabrics can be solved.So far,the system has been used in the automatic fabric laser cutting system in Golden Laser company.