School of Transportation, Southeast University, Nanjing 211189, China.
Department of Electrical Engineering and Computer Science, The Catholic University of America, Washington, DC 20064, USA.
Sensors (Basel). 2018 Aug 21;18(9):2754. doi: 10.3390/s18092754.
Since deformation estimation may lead to errors occurring when the camera vibrates, it is necessary to remove the image global motion before computing real bridge deformation. In this study, a combination of image motion correction algorithm and 2D image-based deformation measurement technique was utilized to address the issue of camera motion during the image data acquisition for bridge deformation measurement. Based on the proposed methodology, the image motion parameters were estimated by defining an effective sub-image in the image and using Iterative Affine Motion Estimator. Then the estimated parameters were applied to all pixels of each captured image to remove the motion effect. Finally, the corrected images were used to analyze by a 2D image-based deformation measurement technique in order to extract and measure real bridge deformation by tracking artificial or natural targets. The proposed methodology was validated by two experiments in the lab and field environments. Achieved results show the accuracy and reliability of the proposed methodology.
由于相机振动可能导致变形估计出现误差，因此在计算真实桥梁变形之前，有必要去除图像全局运动。在这项研究中，采用了图像运动校正算法和二维基于图像的变形测量技术的组合，以解决在桥梁变形测量的图像数据采集过程中相机运动的问题。基于所提出的方法，通过在图像中定义有效子图像并使用迭代仿射运动估计器来估计图像运动参数。然后，将估计的参数应用于每个捕获图像的所有像素，以去除运动效果。最后，使用校正后的图像通过二维基于图像的变形测量技术进行分析，以便通过跟踪人工或自然目标来提取和测量真实的桥梁变形。该方法通过实验室和现场环境中的两个实验进行了验证。所获得的结果表明了该方法的准确性和可靠性。
