Measurement Of Dynamical Displacement Of Bridge Based On Digital Image Processing Techniques

Bridge health monitoring has become an important thrust in the current structural engineering research. As a significant part for bridge detection, the measurement of dynamic displacement is a vital indicator for the evaluation of structural safety. However, the contact-type sensors may require temporary support platforms to fix their sensor bodies or targets, so stationary platform usually has to be close to the structure because the size of sensor is relatively small compared with the size of the structure. In addition, most bridges are constructed overpass the sea or the river. As a result, measuring the displacement of bridge is extremely difficult. Therefore, an alternative system, which is able to reduce the installation expenses, enhance the reliability of measured data, and measure the displacement in a remote distance without attaching sensors, is needed. In this dissertation, the main purpose is to explore the implementation of a real-time dynamic displacement measuring system for bridge based on digital image processing technique, laboratory and real bridge tests were carried out. The research has important theoretical significance and practical value.By comparing the results of the different methods of image de-noising, it is concluded that when difference between target and background is obvious, wavelet transformation de-noising method can effectively remain edge information of original images and improve the accuracy of measurement as well. Meanwhile, after study on several image segmentation algorithms based on the threshold method, an improved maximal between-cluster variance-based algorithm was proposed, which can decrease the detrimental effect caused by lighting on system measurement precision.According to the ideal of pin-hole image model of optical system, and also taking the axisymmetric and non-axisymmetric distortion into consideration, cubic polynomial is adopted for the surface fitting aiming at modeling the relationship between the object space and pixel coordinate system. Through the experiments of different angles between visual axis of camera and observation point, we drew a concluded that the precision of this system is 0.03mm. To obtain the coordinates of the multiple spots in a picture, each spot as a connected region must be labeled and recognized first, and then the coordinate of the spot center can be calculated. This dissertation proposed a method based on region growing for the identification of connected region.Under the windows 7 platform, Lab VIEW is used to develop the dynamic displacement measurement system for bridge based on digital image processing techniques. To verify the feasibility and the accuracy of the measurement system, an accuracy test, a series of static loading and dynamic loading experiments were carried out in laboratory and a field dynamic experiment was conducted on real bridges. The results measured using the displacement meter and the values measured using the communicating pipe methods were employed to compare the results of static loading and dynamic loading experiments in laboratory and dynamic test on a real bridge, respectively. Meanwhile, spectrum analyses were conducted on the measured data of this system.The results show the displacements obtained by image processing have a good agreement with the values measured by the compared methods, and the accuracy of this measurement system maintains millimeter-level and the frequency of bridges calculated using image processing data were accurate, which satisfies the requirement of displacement measurement for bridge and the basic analysis of dynamic performance.