摘要
螺栓在工程结构中起到固连作用,其工作状态直接关乎到设备的工作状态甚至人员的生命安危,而多数处于真实开放环境中的螺栓难以排查.为满足螺栓自动检修设备对处于开放环境中的螺栓进行安全检修,更加准确快速地对其进行识别定位,以保障其发挥正常功效,并提高设备自动化率,提出一套针对处于开放环境中螺栓的位置与姿态测算方法.通过自主搭建的双目系统,在完成相机参数标定校准后对真实环境中的六角螺栓进行图像采集,通过归一化相关匹配(normalized cross correlation,NCC)方法对2幅图像进行一致性检测,完成特征点匹配以及立体计算,精确地获取目标螺栓的深度信息;同时利用虚拟相机技术将3D CAD模型生成2D模型数据库,通过高斯金字塔分层方法与贪婪算法对处于开放环境中的螺栓与2D模型数据库进行搜索与模型匹配,得到最优匹配结果,最终完成对开放环境中螺栓的姿态信息测量.实验结果表明:此方法准确度高,且具有很好的实用性,效果良好.
Bolts play a role of fixation in engineering structure,and their working state is directly related to the working state of equipment and even the life safety of personnel.However,most bolts in the real open environment are difficult to be checked.To meet the requirements of bolt automatic maintenance equipment for safe maintenance of bolts in the open environment,more accurate and rapid identification and positioning,to ensure its normal function,and improve the rate of equipment automation,a set of bolt position and attitude measurement method in the open environment was proposed.In this paper,an independent binocular system was built to collect images of hexagon bolts in the real environment after the calibration of camera parameters.The consistency detection of the two images was carried out by NCC method.Feature point matching and three-dimensional calculation were completed to accurately obtain the depth information of the target bolts.At the same time,a 3D CAD model was generated into 2D model database by using virtual camera technology.The bolt and 2D model database in the open environment were searched and matched by Gaussian Pyramid hierarchical method and greedy algorithm,and the optimal matching results were obtained.Finally,the attitude information of bolts in the open environment was measured.The experimental results show that the method is accurate,practical and effective.
作者
刘志峰
王子涵
赵永胜
程强
LIU Zhifeng;WANG Zihan;ZHAO Yongsheng;CHENG Qiang(College of Mechanical Engineering and Applied Electronics Technology,Beijing University of Technology,Beijing 100124,China)
出处
《北京工业大学学报》
CAS
CSCD
北大核心
2020年第7期734-742,共9页
Journal of Beijing University of Technology
基金
北京市科技重大专项资助项目(Z18110400310000)。
关键词
计算机视觉
螺栓
特征提取
三角测距
位姿测量
高斯金字塔
computer vision
bolt
feature extraction
trigonometric ranging
position measurement
Gaussian pyramid
