A Vision-based Method for On-Road Truck Height Measurement in Proactive Prevention of 2 Collision with Overpasses and Tunnels
Automation in Construction
MetadataShow full item record
Dai, F., Park, M., Sandidge, M., & Brilakis, I. (2014). A Vision-based Method for On-Road Truck Height Measurement in Proactive Prevention of 2 Collision with Overpasses and Tunnels. Automation in Construction, 50 29-39. https://doi.org/10.1016/j.autcon.2014.10.005
This is the accepted manuscript. The final version is available from Elsevier at http://www.sciencedirect.com/science/article/pii/S0926580514002167.
Over-height trucks are continuously striking low clearance overpasses and tunnels. This has led to significant damage, fatalities, and inconvenience to the public. Smart systems can automatically detect and warn oversize trucks, and have been introduced to provide the trucks with the opportunity to avoid a collision. However, high cost of implementing these systems remains a bottleneck for their wide adoption. This paper evaluates the feasibility of using computer vision to detect over-height trucks. In the proposed method, video streams are collected from a surveillance camera attached on the overpass/tunnel, and processed to measure truck heights. The height is measured using line detection and blob tracking which locate upper and lower points of a truck in pixel coordinates. The pixel coordinates are then translated into 3D world coordinates. Proof-of concept experiment results signify the high performance of the proposed method and its potential in achieving cost-effective monitoring of over-height trucks in the transportation system. The limitations and considerations of the method for field implementation are also discussed.
Accident avoidance, Computer vision, Height measurement, Low clearance, Over-height truck
This material is based upon work supported by West Virginia University, Myongji University, and University of Cambridge.
External DOI: https://doi.org/10.1016/j.autcon.2014.10.005
This record's URL: https://www.repository.cam.ac.uk/handle/1810/246887