摘要
提出一种机器人"手-眼"视觉分步动态目标定位方法,实验中保持摄像机图像平面与工作台平面平行并对摄像机的拍摄高度进行设定,最终将目标定位在图像的中心位置。在目标类型识别过程中,采用BP神经网络技术识别目标表面的字符,实现对目标类别的判定。这两种方法相结合,用于活塞的在线识别与分选,由活塞表面的字符信息及活塞定位信息引导末端执行器完成对活塞的准确分选。
A step dynamic object positioning method of robot eye-in-hand vision was proposed. An experimental platform was developed in which the CCD plane was in parallel with worktable plane, and images were acquired by camera at several fixed high positions, and target was positioned at the center of the image finally. During the recognition procession of target type, BP neural network was used to recognize the characters on target's surfaces, and classification of target type was achieved. By application of combination of these two methods on piston's online recognition and selection, the end-executor was led to select the piston accurately by character information on piston's surface and position information.
出处
《机床与液压》
北大核心
2011年第1期27-30,共4页
Machine Tool & Hydraulics
基金
吉林省教育厅科技合作项目字(2004)第22号
东北电力大学博士基金资助项目(BSJXM-200803)