赵鹏翔, 冯培, 欧建鑫,等.基于双目立体视觉的纱线运动参数识别[J].河南理工大学学报（自然科学版）,2023,42(3):103-110.

ZHAO P X, FENG P, OU J X,et al.Motion parameter identification of yarn based on binocular stereo vision[J].Journal of Henan Polytechnic University(Natural Science) ,2023,42(3):103-110.

基于双目立体视觉的纱线运动参数识别

赵鹏翔1, 冯培1,2, 欧建鑫1, 李姝佳1,2, 王永兴1,2

1.东华大学 机械工程学院，上海  201620;2.东华大学 纺织装备教育部工程研究中心，上海  201620

摘要:针对目前运用传统接触方法难以准确、无干扰测量纱线运动参数的问题，提出基于双目立体视觉的纱线运动参数识别方法。采用双目高速相机采集运动纱线图像，开发双目高速同步抓帧存图程序；利用高精度标定板进行双目标定；经滑动窗口和线面交点方法定位纱线测试点；对极线校正后严格行对准的像对使用块匹配、原始和改进半全局块匹配等立体匹配算法计算深度图，并采用视差原理求得纱线测试点空间位置坐标；结合存图时刻求得纱线运动参数。实验表明，本文方法得到的纱线运动参数中位移误差均值为0.132 mm，与实际情况吻合良好。

关键词:双目立体视觉;纱线运动参数;非接触测定;双目标定;极线校正;改进SGBM算法

doi:10.16186/j.cnki.1673-9787.2021110028

基金项目:国家重点研发计划项目（2017YFB1304000）

收稿日期:2021/11/05

修回日期:2021/12/13

出版日期:2023/05/25

Motion parameter identification of yarn based on binocular stereo vision

ZHAO Pengxiang1, FENG Pei1,2, OU Jianxin1, LI Shujia1,2, WANG Yongxing1,2

1.College of Mechanical Engineering，Donghua University，Shanghai  201620，China;2.Engineering Research Center of Advanced Textile Machinery of Ministry of Education，Donghua University，Shanghai  201620，China

Abstract:Aiming at the problems of low accuracy and no interference of measuring yarn motion parameters for traditional contact method，a binocular stereo vision method was proposed：Images were collected by binocular high-speed camera，stored by the developed software of binocular high-speed synchronous frame capture；The binocular camera was calibrated by high precision calibration board；The yarn test point was detected by sliding window and intersection；After polar correction，the depth map was calculated by stereo matching algorithms of Block Matching，Semi-Global Block Matching and improved SGBM algorithm.The spatial location information of yarn test point was obtained by parallax principle.The yarn motion parameters were obtained with the three-dimensional coordinate and image storage time，the average error of displacement was 0.132 mm.The experimental results showed that the yarn motion parameters were in good agreement with the actual situation.

Key words:binocular stereo vision;yarn motion parameter;non-contact measurement;binocular calibration;polar correction;improved SGBM algorithm

基于双目立体视觉的纱线运动参数识别_赵鹏翔.pdf