摘要 针对工程应用中视觉测量网络站位布局规划不完善,各站位测量相对全局测量结果具有较大偏差,导致全局测量点云数据分层问题,提出了双目视觉测量网络站位布局规划方法。首先,建立双目视觉测量模型,结合理论分析构建参数误差模型;其次,根据参数作用机制将其划分为内部结构参数和外部测量参数两类,并进行精度仿真分析;然后,结合仿真结果,建立全局测量场视觉测量多约束区域划分模型,求解各测量站位初值,构建测量站位优化模型,基于外部测量参数的遗传算法求解各站位位置参数,完成大尺寸视觉测量网络的构建。最后,通过规划实验表明,各站位测量结果之间偏差均在0.04 mm以内,实现了点云数据的高精度拼接,满足工业测量精度要求。 The visual measurement of the network station planning is not consummate in the engineering application.The measurement of each station has a large deviation from the global measurement result,which leads to the layering problem of global measurement point cloud data.In this study,a method is proposed to measure the network station planning using binocular vision.First,the measurement model of binocular vision is established,and the parameter error model is establishedbased on theoretical analysis.Secondly,according to the action mechanism of the parameters,it is divided into internal structure parameters and external measurement parameters,and the accuracy simulation analysis is performed.Then,combined with the simulation results,a multi-constrained region division model of global measurement field vision measurement is established.The initial value of each measurement station is solved,and the optimization model of measurement station is constructed.The genetic algorithm based on external measurement parameters is used to calculate the position parameters of each station,so as to complete the construction of large-scale vision measurement network.Finally,the planning experiment shows that the deviation between the measurement results of each station is within 0.04 mm,which realizes the high-precision splicing of point cloud data and suffices the requirements of industrial measurement accuracy.
机构地区 长春理工大学光电工程学院
出处 《长春理工大学学报:自然科学版》 2021年第2期27-36,42,共11页 Journal of Changchun University of Science and Technology(Natural Science Edition)
关键词 工业测量 视觉测量 网络规划 优化 大尺寸 industrial measurement visual measurement network planning optimization large size
