Abstract
Routine bridge inspections usually consist of visual observations. These inspections are time-consuming
and subjective. There is a need to identify new inspection techniques for infrastructure that reduce
traffic disturbance, and improve the efficiency and reliability of the acquired data. This study compared
the performance of three different imaging technologies for the three-dimensional (3D) geometric
modeling of existing structures: terrestrial laser scanning, close-range photogrammetry, and infrared
scanning. Each technology was used to assess six existing concrete railway bridges. The technologies
were compared in terms of geometric deviations, visualization capabilities, the level of the inspector’s
experience, and degree of automation. The results suggest that all methods investigated can be used
to create 3D models, however, with different level of completeness. Measurements such as span
length, deck widths, etc. can be extracted with good accuracy. Although promising, a full off-site
inspection is currently not feasible as some areas of the bridges were difficult to capture mainly due
to restricted access and narrow spaces. Measurements based on terrestrial laser scanning were closer
to the reality compared to photogrammetry and infrared scanning. The study indicates the no special
training is needed for photogrammetry and infrared scanning to generate a 3D geometric model.
and subjective. There is a need to identify new inspection techniques for infrastructure that reduce
traffic disturbance, and improve the efficiency and reliability of the acquired data. This study compared
the performance of three different imaging technologies for the three-dimensional (3D) geometric
modeling of existing structures: terrestrial laser scanning, close-range photogrammetry, and infrared
scanning. Each technology was used to assess six existing concrete railway bridges. The technologies
were compared in terms of geometric deviations, visualization capabilities, the level of the inspector’s
experience, and degree of automation. The results suggest that all methods investigated can be used
to create 3D models, however, with different level of completeness. Measurements such as span
length, deck widths, etc. can be extracted with good accuracy. Although promising, a full off-site
inspection is currently not feasible as some areas of the bridges were difficult to capture mainly due
to restricted access and narrow spaces. Measurements based on terrestrial laser scanning were closer
to the reality compared to photogrammetry and infrared scanning. The study indicates the no special
training is needed for photogrammetry and infrared scanning to generate a 3D geometric model.