The Vibration Measurement Research Of Flexible Structure Based On Machine Vision

Large flexible structures are widely used in our industrial and life. Due to the inherent elasticity of the flexible structure,vibration phenomenon often appears in the flexible body,which may cause damage to safety performance, strength and service life to the flexible body.Therefore, this paper proposes a non-contact measurement method based on machine vision,which is a quick and convenient means of vibration measurement. It can achieve multi-point simultaneous and long-distance measurement of vibration and ensure the safety for researchers.It is a quick and convenient means of vibration measurement and there is a good practical.First of all, based on the current research status and existing research problems, in this paper, the solution of flexible body vibration measurement based on monocular machine vision was presented. The test system of the machine vision vibration was built and Zhang Zhengyou's camera calibration method was selected to extract camera's interior and external parameters.In addition, according to the actual situation on-site testing, the paper designed a calibration method based on laser ranging and direct reference size.In the image preprocessing part, the fast median filtering algorithm was used to filter the image noise, and ROI technology was used to reduce the image operation redundancy.Before the measurement, Hough algorithm was adopted to extract straight lines as the auxiliary line to establish reference coordinate. In the image processing algorithm part of this paper, target vibration tracking was divided into feature detection and dynamic target tracking algorithm for a more in-depth study. Considering the complexity of environment and target tracking, OSTU threshold, SURF descriptors and Meanshift were used for feature detection and tracking algorithm. In the part of dynamic tracking algorithm,in order to solve the problem of target tracking loss, interference of camera platform vibration and multi object synchronization measurement, Kalman dynamic filtering was added in the tracking algorithm, and a multi feature fusion target tracking algorithm was designed.The Opencv open source vision framework is used to complete the design of the camera calibration algorithm, image processing algorithm and vibration analysis software system. In the VS2013 programming platform, the QT framework is used to complete the design of the host computer vision processing algorithm and the UI software.Finally, the measurement experiment of flexible manipulator vibration, derrick hoist vibration and hoist vibration were designed to compare the electrical measurements and visualmeasurements. The results verified the accuracy of the algorithm and analyzed the system error and the measurement accuracy.