Investigating the accuracy and reliability of UAV photogrammetry on railway infrastructure

3D scanning systems have become more commonly used for geometrical data capture, especially with the increased uptake of new technologies within the industry, coupled with the advances in computer systems. 3D scanning is a wide, encompassing term that relates to the non-contact, non-destructive process of digitally capturing the shape and form of physical elements. Within the industry, this is most commonly carried out using a laser scanner, which uses a line of laser light to scan an element and return a set of points, or coordinates. However, with advances in technology and consumer drones, UAV photogrammetry is quickly becoming more popular. Photogrammetry can be defined as the use of photography to obtain measurements between objects to define points in space. When combined with drones, this technique allows rapid, inexpensive scanning of the built environment. Critical to the use of photogrammetry is the establishment of accuracy and reliability. As laser scanners are generally accepted as being accurate for this work it is necessary to compare the two technologies to establish how accurate UAV photogrammetry is and if it can be used as a suitable tool for 3D data capture.

For this project, drone technology has been used to photograph railway infrastructure to provide a basis for comparison to other survey methods and to test the technology on a varied landscape. The accuracy of models was evaluated by determining measurements within the scene and comparing these results to measurements obtained from a laser scan survey. Results shows a minimum difference in measurements of 0.33mm and a maximum average of 37.5mm.