Research On Pose Measurement And Optimization Of Moving Objects Based On Monocular Vision

The measurement of moving object pose in high-speed complex flow field is an important means to determine the aerodynamic performance of moving object trajectory and motion mode.The object is affected by flow separation,flow shock and so on in high-speed complex flow field.Accuracy measurement is difficult.In recent years,visual measurement has a wide measurement range,does not change the aerodynamic shape of the model,and has high measurement accuracy.It is widely used in wind tunnel experiments at home and abroad.To this end,this paper based on monocular vision for the measurement of moving object poses,focusing on the identification and matching of feature marker points,monocular pose measurement and optimization,designed a monocular vision measurement system experiment,the specific research content and results are as follows:(1)The identification and matching of marker points in monocular visual pose measurement were studied.Firstly,in the marker point identification positioning,according to the closed contour feature of the circular marker point,an eight-neighbor boundary tracking method for the closed contour is designed.Through experimental comparison,this method can directly eliminate the background interference that is not closed contour,which is beneficial to the identification of feature points.Then,according to the constraint criterion,the circular marker points are extracted,and the gray center of gravity method is used for central positioning.For small rolling bodies,it is measured.The target's circular marker points match,and a wraparound marker point encoding and decoding method is set up,which decodes the distance between the head and tail markers and the vector relationship to achieve the same name matching.The experimental results show that the accuracy and rapidity of decoding are improved under long-distance large field of view,complex background and high blur.(2)In order to meet the high-precision measurement requirements of the placed position and attitude in high-speed complex flow field,the influence of image noise on pose estimation error is analyzed.Based on the traditional nonlinear optimization pose algorithm,a monocular pose estimation optimization method based on adaptive re-projection error is proposed.The method sets the constraint interval with the initial value of pose estimation,establishes a new penalty function,transforms the constrained nonlinear optimization into unconstrained nonlinear optimization,analyzes the relationship between image plane re-projection error and constraint interval and establishes the corresponding mathematical model.The model automatically adjusts the constraint interval and performs adaptive reprojection error constrained nonlinear optimization pose estimation parameters.The simulation results show that the proposed method has the optimal solution for the reprojection error and the pose estimation parameters under different image noises,and it is superior to the traditional nonlinear optimization algorithm and has higher pose estimation accuracy.The relative pose measurement schemes of different moments are given to solve the relative pose of moving objects.The validity and superiority of the monocular pose measurement method are verified by experiments.It has certain application value for realizing high-precision monocular pose measurement.(3)The monocular visual pose measurement system was designed,and the hardware construction and software development were carried out respectively.In the software design,the functional modularization was developed,and a good human-computer interaction interface and visualization were presented.Distortion correction is performed on the camera using multiple resections,and distortion parameters are obtained.Finally,the motion object pose measurement experiment based on the plane target and the roll target is carried out.The experimental results show that the displacement measurement accuracy of the rolling model along the X direction is 0.26 mm,and the angle measurement accuracy is 0.10°.The displacement accuracy of the planar target in the X direction is 0.18 mm,and the angle measurement accuracy is 0.37°.