基于机器视觉的水表分拣系统研究

Research on Water Meter Sorting System Based on Machine Vision

针对现阶段水表生产分拣主要依靠人工、自动化程度不高、效率低下等问题,对水表定位、机械手自动抓取等问题进行研究,设计了一套基于机器视觉的水表分拣系统。通过颜色特征粗定位水表区域,基于归一化内积相似性精定位水表,得到水表位姿;构建Eye-to-Hand手眼标定模型并求解,将水表位姿转换到机械手坐标系,实现机械手对水表的精确抓取。研究结果表明:该系统的定位误差在1.5 mm以内,机械手能准确抓取水表,极大地提高水表上下料效率,基本避免了人工失误。

In relation to problems like manual sorting,low automation,and low efficiency etc.in present water meter production,water meter positioning and automatic grabbing by manipulators are studied,and a set of automatic water meter sorting system based on machine vision is designed.With the water meter area being roughly located by color features,the water meter is precisely positioned based on the normalized inner product similarity to obtain the water meter pose.The Eye-to-Hand calibration model is constructed and solved to convert the water meter pose to the robot coordinate system,with which the the water meter is accurately captured by manipulator.The research results show that the positioning error of the designed system is within 1.5 mm,and the manipulator can accurately grasp the water meter,which greatly improves the efficiency of water meter loading and unloading by basically avoiding the occurrence of manual errors.