A Study of Image-based Element Condition Index for Bridge Inspection

This paper presents an innovative computer vision method for condition assessments of bridges with multiple defects in bridge elements using digital images. This work utilizes 3D model of existing bridges and overlays digital images on 3D model to simulate on-site visual inspection. The analysis of element condition index (ECI) of bridges requires information about the severity and extent of defects in elements. In general, ECI is evaluated manually during routine bridge inspection considering the severity of dominant defects. The evaluation of ECI with multiple defects needs to be addressed with consideration of dominant defect as well as the interaction among defects. However, Image-based quantification techniques largely depend on geometry of objects (i.e. shapes). Shape vectors of a given object change as they are translated, rotated, and scaled with different magnitudes. This work considers shape preserving algorithms such as, affine and projective transformation for proper image alignment. Semi-automated approach for detection and quantification of concrete distress such as cracks and spalling are considered for the defects analysis. The proposed methodology ensures the consistency in reporting ECI and eliminates the shortcoming of traditional approaches.