基于机器视觉的细纱管分类检测  

Bobbin inspection and classification based on machine vision

作　　者：高畅[1] 刘浩克 刘基宏[1] GAO Chang;LIU Haoke;LIU Jihong(Key Laboratory of Eco-Textiles,Ministry of Education,Jiangnan University,Wuxi 214122,China)

机构地区：[1]江南大学生态纺织教育部重点实验室,江苏无锡214122

出　　处：《丝绸》2019年第1期28-32,共5页Journal of Silk

基　　金：江苏高校优势学科建设工程资助项目(苏政办发[2014]37号);江南大学自主科研项目(JUSRP51417B)

摘　　要：为实现自动络筒机中纱管输送装置的余纱量检测与纱管分类,文章提出一种基于机器视觉的纱管分类方法。将采集的纱管图像分割为若干区域,以各分区前景、背景和凸包面积为基础构建反映目标形态和对称性的几何特征;利用Gabor滤波器组增强目标纹理信息,随后通过主色提取和色差计算构建各分区的纹理特征。采用多分类支持向量机利用提取特征进行分类,将输入样本归为空管、残纱管和有纱管三类。分类算法交叉验证结果表明,在多种参数水平下,分类器对各种管壁颜色的棉纱纱管的分类准确率达到96%以上。多品种纱线试验表明,分类器对不同细度和颜色纱线的纱管分类真阳性率达到92%以上。To achieve residual yarn detection and bobbin classification of bobbin conveying device in automatic winder,a method of bobbin classification based on machine vision was proposed.The image of bobbin was captured and then divided into several areas.In each partition,geometric features representing the shape and symmetry of corresponding partition were constructed based on the sizes of foreground,background and convex hull area.The Gabor filter bank was used to enhance the target texture information,and the texture features of each partition were constructed by the dominant color extraction and color difference calculation.The classification was conducted by utilizing features extracted with multi-classification support vector machine.The samples were classified into three types:empty bobbins,yarn bobbins and residual-yarn bobbins.The cross-validation results of the classification algorithm showed that the classification accuracy of the cotton yarn bobbins with various tube wall colors by the classifier reached more than 96%under various parameter levels;the tests of multi-species yarns showed that the classification true positive rate of yarn bobbins with yarns of different fineness and color reached more than 92%.

关 键 词：细络联 纱管分拣 GABOR滤波 主色提取 支持向量机 

分 类 号：TS103.7[轻工技术与工程—纺织工程]