一种通用的机织物密度图像自动识别方法

A General Image Automatic Recognition Method of Woven Fabric Density

探讨一种通用的机织物经密纬密图像自动识别方法。采用数字图像处理技术采集机织物彩色图像,经图像灰度化、经纱纬纱方向纠偏、经纱纬纱亮度投影曲线、漏检与去异常点、单根经纱纬纱分割等步骤,最终实现机织物经密纬密的自动识别,根据相邻纱线间隙位置差与均值的比值实现纱线颜色突变漏检与去异常点。结果表明:颜色突变系数、异常点突变系数能准确实现漏检和去异常点,实现单根纱线正确分割,图像法测量准确度高、速度快、效率高。认为:该图像自动识别方法对素色机织物、单色机织物、多色机织物及印花织物均有较好的适应性。

A general image automatic recognition method of warp density and weft density was discussed.Color image of woven fabric was collected by digital image processing technology,automatic recognition of warp density and weft density was finally realized through the steps of image graying,warp and weft direction correction,warp and weft brightness projection curve,missed detection and removal of abnormal points,and single warp and weft segmentation.According to the ratio of position difference of adjacent yarn gap to the mean value,the yarn color mutation missed detection and abnormal points removal were realized.The results showed that color mutation coefficient and abnormal point mutation coefficient could accurately realize the missed detection and removal of abnormal points,and realize the correct segmentation of single yarn.The image method had the advantages of higher accuracy,higher speed and higher efficiency.It is considered that the image automatic recognition method have better adaptability to plain woven fabric,monochrome woven fabric,multicolor woven fabric and printed fabric.