Automated Characterization of Bridge Deck Distress Using Pattern Recognition Analysis of Ground Penetrating Radar Data

Detecting distress in concrete bridge decks in the early stages of deterioration is a priority for highway engineers. However, tools for accomplishing this detection task have been unreliable and qualitative in the past. Two new ground penetrating radar (GPR) system prototypes have been built by Lawrence Livermore National Laboratory for the Federal Highway Administration to address this problem. To handle systematically the large volumes of data that are produced by these GPR systems, pattern recognition algorithms are being used along with image processing techniques. The initial results show that pattern recognition methods provide an effective means of handling this data and merit further development efforts. This work was the result of the ASNT Fellowship Award awarded to Virginia Polytechnic Institute and State University in 1996.