基于双目视觉的堆叠零件识别和定位研究  

Research on recognition and positioning of stacked parts based on binocular vision

作　　者：刘祚时[1] 范雨婷 陈飞 LIU Zuoshi;FAN Yuting;CHEN Fei(School of Mechanical and Electrical Engineering,Jiangxi University of Science and Technology,Ganzhou 341000,China)

机构地区：[1]江西理工大学机电工程学院,江西赣州341000

出　　处：《传感器与微系统》2023年第2期41-44,共4页Transducer and Microsystem Technologies

基　　金：国家自然科学基金资助项目(71361014);江西省研究生创新专项资金资助项目(YC2019—S313)。

摘　　要：针对堆叠零件识别与定位精度不高与效率低下等问题,设计了一种基于双目视觉的堆叠零件识别与定位方法。利用双目视觉中的立体匹配,结合前景分割、孔洞填充、随机抽样一致性(RANSAC)迭代法剔除误匹配点,对半全局块匹配(SGBM)算法进行改进,获取更优的视差效果图。最后,对圆形线圈骨架堆叠零件进行了测量精度和定位精度实验。实验结果表明:改进算法能够很好地解决弱纹理和遮挡区的匹配问题,系统测量精度实际偏差控制在1 mm以下,在实际应用方面具有一定参考价值。Aiming at the problems of low precision and low efficiency in identification and positioning of stacked parts,a method of identifying and positioning stacked parts based on binocular vision is designed.Using stereo matching in binocular vision,combining foreground segmentation,hole filling,and random sample consensus(RANSAC)iterative method to eliminate mismatched points,the semi-global block matching(SGBM)algorithm is improved to obtain a better parallax effect map.Finally,the measurement precision and positioning precision experiments are carried out on the stacked parts of the circular coil bobbin.The experimental results show that the improved algorithm can well solve the matching problem of weak texture and occlusion area,and the actual deviation of the system measurement precision is controlled below 1 mm,which has certain reference value in practical applications.

关 键 词：堆叠零件 双目视觉 半全局块匹配算法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]