| With the rapid development of large-scale scientific projects,higher requirements for the precision and efficiency in the assembly of large-scale scientific projects are put forward.Traditional precision measurement technologies such as industrial surveying total stations,theodolites,and laser trackers can no longer meet the high-precision and high-efficiency requirements in the assembly of large-scale scientific projects.As a high-precision,non-contact measurement method,vision measurement provides a new ideas and measurement methods for this field.In recent years,monocular vision measurement technology is gradually being used in large-scale scientific projects assembly.However,in the assembly by the constraints of a large measurement range,measurement requirements and other conditions,monocular measurement often cannot meet the requirements of measurement.Multinocular vision measurement spatial network is expected to solve the problem.In order to improve the accuracy and efficiency of large-scale multinocular vision measurement spatial network,in this paper,various factors affecting the vision measurement system are analyzed.The study is conducted around camera calibration in large-size scenes,the influence of the structural parameters of the vision measurement system and optimization algorithms of the multinocular vision measurement spatial network,and then optimized design of the spatial network structure for multinocular vision measurement.The content of the research results has been verified by practical project,and finally the optimized design of high precision and efficient vision measurement space network has been realized.The main content of this study is as follows:(1)The imaging model and aberration model of the camera are described.In order to reduce the influence of environmental disturbance on camera calibration accuracy,based on the principle of Zhang Zhengyou calibration method,a new camera calibration optimization method based on total least squares is proposed.The image coordinates of the corner points of the calibration plate are regained using the projection method based on the calibration parameters obtained from the solution,and compared with the theoretical value to obtain the root mean square error of coordinate offset.The experiment is designed to validate the calibration optimization.The results show that after the optimization of the overall least squares method,the coordinate offset error of the corner points is reduced by 0.001 mm on average,and the accuracy of camera calibration is improved.(2)The measurement principles of the models of binocular vision measurement systems and multinocular vision measurement systems are systematically analyzed.Based on the measurement error model of binocular vision measurement system,the error model of structural parameters such as field of view angle,angle between optical axis and baseline,and baseline length are constructed.The simulation analysis of each model is also carried out to investigate the influence law of each structural parameter on the error of the measurement system.The simulation results show that as the structural parameters increase,the measurement error first decreases and then increases,and the increase is gradually accelerated.The optimal range of values for each structural parameter is also obtained.In order to ensure the reliability and accuracy of the simulation results,the actual experiment is designed to verify the simulation results.The results show that the measured data is consistent with the simulation results.Within the optimal values of each structural parameter,the measurement errors are reduced by0.114 mm,0.120 mm,and 0.061 mm respectively,which meet the accuracy requirements and verify the reliability of the simulation results.(3)The constraints of multinocular vision measurement spatial network are analyzed.Then the models of visibility constraint and coverage constraint are constructed,and an optimization method of multinocular vision measurement spatial network based on KNN algorithm is proposed.A network optimization evaluation model was constructed to evaluate the optimized network using the overall network coverage as the criterion.Based on practical projects,the optimization method is validated by simulation experiments with 4-camera network,8-camera network,12-camera network and 16-camera network as examples.The experimental results show that the optimized spatial network has a more uniform station distribution and the coverage reaches 100% under the 12-camera network,which meets the optimization evaluation criteria.In order to ensure the accuracy of the optimized network,an actual measurement experiment is carried out on the simulated network.The results show that with the optimization of the spatial network,the accuracy of the control points in the network is gradually improved and finally meets the accuracy requirements in the assembly of largescale scientific projects. |
