3D Structure Modeling And Vibration Testing Based On Binocular Vision

Dynamic characteristic analysis is an important means to solve mechanical structural dynamics problems,which requires accurate vibration response data through vibration testing.The acceleration sensor is mainly adopted in traditional vibration testing,which is troublesome to install and test,and is easy to cause additional mass for the lightweight and flexible structure,which seriously reducing the test accuracy.A vibration testing technology based on binocular camera is studied in this paper.The 3D testing model of the structure is established through multi-angle photography,and the multimeasurement point tracking based on computer vision is realized to obtain vibration displacement response and realize non-contact measurement.The main work is as follows:The imaging principle of binocular camera and two 3D reconstruction methods,stereo matching and line-line intersection,are studied,and the two methods are compared by space point ranging test.The Iterative Closest Point(ICP)camera pose estimation method is studied.It is used to estimate the motion of the camera,realize the coordinate transformation of multiple measurement results,and solve the problem of secondary measurement in the geometry modeling of objects using 3D reconstruction method.The High-Speed Tracking with Kernelized Correlation Filters(KCF)and Color Names(CN)kernel correlation filtering visual tracking algorithms are studied.Based on this,multi-features and multi-scales are fused to propose MMK tracking algorithms to achieve subpixel and multi-targets tracking.The binocular visual vibration testing system is established,and the test analysis software is developed based on Qt and Matlab.The system is used to conduct 3D modeling and operational modal test to an antenna and a self-built GARTEUR aircraft model.The results show that the test system achieves the expected effect and has obvious application value in engineering practice for the structure with light weight and high flexibility.