基于机器视觉的筒子纱线头识别方法  

An identification method of cheese yarn ends based on machine vision

作　　者：金鹏翔 刘宜胜[1] JIN Pengxiang;LIU Yisheng(School of Mechanical Engineering,Zhejiang Sci-Tech University,Hangzhou 310018,China)

机构地区：[1]浙江理工大学机械工程学院,杭州310018

出　　处：《浙江理工大学学报（自然科学版）》2024年第1期112-119,共8页Journal of Zhejiang Sci-Tech University(Natural Sciences)

摘　　要：为了在纺织织造的整经工序中高效吸取筒子纱线头,克服人工和机器寻线工作效率低、装置复杂等问题,提出了一种基于机器视觉的筒子纱线头识别方法。首先,对采集的筒子纱图像进行畸变校正,根据透视投影和局部透视变换建立筒子纱校正模型,得到理想的筒子纱侧面展开图像;其次,采用4邻域连通规则对校正后的图像进行连通域标记,得到含曲线部分最多的连通域;最后,计算该连通域的曲折度,达到阈值的即为所需识别的线头。对含线头图像和不含线头图像进行识别,结果表明:局部透视变换的校正方法比重映射速度提升近30%,表现出较高的校正精度;曲折度阈值判别方法能有效滤除非线头图像,提高了线头图像的识别精度。使用机器视觉进行筒子纱线头识别能有效检测出每个筒子纱中的线头部分,减少生产线上的人工干预,为构建高效节能的智能化生产模式提供参考。To efficiently extract cheese yarn ends in the warping process of textile weaving and solve the problems of low work efficiency of manual and machine thread hunting,and complex devices,a method for identifying cheese yarn ends based on machine vision was proposed.Firstly,distortion correction was performed on the collected cheese yarn images,and a cheese yarn correction model was established based on perspective projection and local perspective transformation to obtain an ideal cheese side expansion image.Secondly,the 4-neighbor connection rule was used to mark the connected domain of the corrected image,and the connected domain containing the most curve parts was taken out.Finally,the tortuosity of the connected domain was calculated,and the one reaching the threshold was the thread end to be identified.The results of identifying thread end-containing images and thread end-free images show that the local perspective transformation correction method is nearly 30%faster than heavy mapping,showing high correction accuracy;the tortuosity threshold discrimination method can effectively filter out non-thread end images and improve the recognition accuracy of thread end images.Using machine vision to identify the ends of the cheese yarn can effectively detect the thread end part of each cheese yarn,reduce manual intervention on the production line,and provide reference for building an efficient and energysaving intelligent production model.

关 键 词：机器视觉 筒子纱线头识别 校正模型 局部透视变换 连通域标记 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]