双目视觉测量系统位姿解算方法研究  

Research on Pose Calculation Method of Binocular Vision Measurement System

作　　者：李勤文 王志乾[1] 杨文昌 沈铖武[1,2] LI Qin-wen;WANG Zhi-qian;YANG Wen-chang;SHEN Cheng-wu(Changchun Institute of Optics,Fine Mechanics and Physics,Chinese Academy of Sciences,Changchun Jilin 130033,China;University of Chinese Academy of Sciences,Beijing 100049,China)

机构地区：[1]中国科学院长春光学精密机械与物理研究所,吉林长春130033 [2]中国科学院大学,北京100049

出　　处：《计算机仿真》2021年第5期214-218,共5页Computer Simulation

基　　金：吉林省重点科技攻关项目(20170204050GX);吉林省科技发展计划项目(20190302086GX)。

摘　　要：在大部件自动对准系统中,为了解决近距离观测无法获得被观测目标全部特征的问题,使用双目视觉位姿测量解算方法,设置其中一个相机为基准相机,利用两相机之间已知的位姿关系,将非基准相机拍摄的特征点信息转换到基准相机的相机坐标系下,然后利用位姿解算的迭代算法思想求解两部件之间的位姿关系。当特征点数目较少时,迭代算法的误差函数会陷入局部极小值,针对上述问题,设计了一种利用各坐标系之间的转换关系求解迭代初始值的方法,并通过仿真证明了算法的有效性。In the automatic alignment system of large parts, the binocular vision pose measurement and calculation method is used to solve the problem that all the features of the observed object cannot be obtained by close observation. By utilizing the known pose relationship between the two cameras, the feature point information captured by one of the cameras can be converted to another camera coordinate system, which is set as the reference camera. Then, the iterative algorithm of pose calculation is used to solve the pose relationship between the two components. When the number of feature points is small, the error function of the iterative algorithm will fall into the local minimum. To solve this problem, a method was designed to solve the initial value of iteration by using the transformation relationship between coordinate systems, and the effectiveness of the algorithm was proved by simulation experiments.

关 键 词：自动对准 双目视觉 位姿解算 迭代算法 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]