Proceedings Article | 4 September 2009
Proc. SPIE. 7447, Videometrics, Range Imaging, and Applications X
KEYWORDS: Calibration, Inspection, Roads, 3D image processing, Cameras, Imaging systems, 3D scanning, 3D metrology, 3D modeling, Digital filtering
Rutting and pothole are the common pavement distress problems that need to be timely inspected and
repaired to ensure ride quality and safe traffic. This paper introduces a real-time, automated inspection system
devoted for detecting these distress features using high-speed transverse scanning. The detection principle is based
on the dynamic generation and characterization of 3D pavement profiles obtained from structured light
measurements. The system implementation mainly involves three tasks: multi-view coplanar calibration, sub-pixel
laser stripe location, and pavement distress recognition. The multi-view coplanar scheme was employed in the
calibration procedure to increase the feature points and to make the points distributed across the field of view of the
camera, which greatly improves the calibration precision. The laser stripe locating method was implemented in four
steps: median filtering, coarse edge detection, fine edge adjusting, stripe curve mending and interpolation by cubic
splines. The pavement distress recognition algorithms include line segment approximation of the profile, searching
for the feature points, and parameters calculations. The parameter data of a curve segment between two feature
points, such as width, depth and length, were used to differentiate rutting, pothole, and pothole under different
constraints. The preliminary experiment results show that the system is capable of locating these pavement
distresses, and meets the needs for real-time and accurate pavement inspection.
