Computer Vision Techniques for Bridge Bearing Condition Assessment using Visual Inspection Photographs

This paper is proposed a computer vision technique for measuring the static displacement of bridges with changes in temperature using a general photograph, which is stored in the database of the bridge management system (BMS) for visual inspection. Also, the proposed methodology is developed using open source computer vision (OpenCV) language program and validated both in the laboratory using a bearing model and on an actual bridge. The proposed numerical procedure shows that the nontarget imagebased technique using Gaussian highpass filters in the frequency domain, projective linear transformation (PIT) and digital image correlation (DIC) algorithm can extract the displacement of a bearing. It is concluded that this technique is able to obtain the bridge bearing condition assessment.