基于单摄像机视觉测量系统的网络化数据融合  

Networked data fusion based on visual measure system with single camera

作　　者：周婧[1,2] 高印寒[3] 陈小林[4] 刘长英[1] 刘静[5] 

机构地区：[1]吉林大学仪器科学与电气工程学院,长春130022 [2]吉林农业大学信息技术学院,长春130118 [3]吉林大学汽车仿真与控制国家重点实验室,长春130022 [4]中国科学院长春光学精密机械与物理研究所,长春130033 [5]吉大小天鹅仪器有限公司,长春130012

出　　处：《吉林大学学报（工学版）》2013年第1期92-97,共6页Journal of Jilin University:Engineering and Technology Edition

基　　金：高等学校博士学科点专项科研基金项目(20110061110059);吉林省科技发展计划项目(20070535)

摘　　要：针对单摄像机视觉测量系统无法对被测物体的部件间相互遮挡部位进行测量的问题,提出一种基于单摄像机视觉坐标测量网络化技术。通过单摄像机视觉测量系统分别在每个网络控制点位置对光学特征点进行测量,再将测量的坐标归一化到同一世界坐标系下,进而得到全局数据,实现整体测量。最后应用加权数据融合算法,依据特征向量的稳定理论分配融合权重,将全局归一化的空间坐标进行融合,得到融合估计的最终表达式。应用三坐标测量机进行仿真对比实验,实验结果表明,应用四元数法及欧拉角法归一化的空间坐标,经过本文算法融合后,最大绝对误差为0.035mm,比点集矩中心融合算法精度高、稳定性好。同时,在测量系统的网络归一化应用中四元数法优于欧拉角法。In view of the fact that the visual measure system with single camera can not measure the shield between the measured objects each other, a networked technique was proposed for this kind of measure system. The optical characteristic spots on every network control point were measured by the single camera measure system, and the measured coordinates were normalized into the same world coordinate system to get the global data, and the overall measurement was achieved. Applying the weighted data fusion algorithm, distributing the weight factors according to the stability theory of eigenvector, fusing the global normalized space coordinates, and the ultimate expression of fusion estimate was derived. The experiment in comparison to the simulation was performed in a CMM and the results showed that the maximal absolute error of the space coordinates is 0. 035 mm afternormalization by quaternion and Euler angles algorithms. The proposed algorithm apppears more accurate and more stable than the point set center fusion algorithm, and in the application of networked normalization of measure system, the quaternion algorithm is better than the Euler angles algorithm.

关 键 词：仪器仪表技术 视觉测量 网络控制点 加权数据融合 

分 类 号：TH741[机械工程—光学工程]