Developing Corrosion Maps of RC Bridge Elements Based on Automated Visual Image Analysis

Reinforced Concrete (RC) bridge elements scanned by Ground Penetrating Radar (GPR) are widely analyzed by utilizing numerical analysis approach. Typically, this involves picking ampli-tude values at rebar location manually or automatically after applying depth correction and sub-sequently, utilizing these values to develop corrosion maps. However, the results obtained from this approach have proven to be unreliable when it is compared with actual bridge element con-dition at the construction or destruction phase. Based on another new visual image analysis ap-proach, accurate condition maps can be developed with less noise as this approach involves an expert analyst to scan through each GPR profile of bridge element while considering all surface and structural anomalies. The current visual image method is highly subjective because it involves manual condition zoning of each profile and it is also prone to human error. The authors of this paper aim to automate this approach through image processing tools in MATLAB® and develop corrosion maps which will be unique and more objective in nature. Image filtering will be utilized to detect hyperbolic regions in a GPR profile and furthermore, corrosion maps will be developed based on intensity of pixels present in the detected hyperbolic regions. The developed maps are expected to be reliable and will be useful in conditional assessment of bridge elements.