Automatic pavement defect detection using 3D laser profiling technology

Abstract Asphalt pavement defects e.g. cracks, potholes, rutting, often cause significant safety and economic problems, thus, to automatic detect these defects is vital for pavement maintaining and management. The fact that 3D defect detection methods is superior to traditional 2D methods and manual survey methods in terms of accuracy and comprehensiveness has been widely recognized. Based on 3D laser scanning pavement data, an automatic defect detection method is proposed to detect pavement cracks and pavement deformation defects information simultaneously in this paper. Specifically, a sparse processing algorithm for 3D pavement profiles is first designed to extract crack candidate points and deformations support points, these processing is based on the assumption that the cracks are microscopic local defects while deformations are macroscopic defects in profiles. Then, the crack information can be detected by combining the extracted candidate points and an improved minimum cost spanning tree algorithm. On the other hand, the deformation depth information is acquired based on the profile standard contours which are constructed by profile envelopes and deformation support points, the accurate location and classification information of deformation defects can be obtained by the regional depth property. Experimental tests were conducted using real measured 3D pavement data containing two categories of defects. The experimental results showed that, based on the 3D laser scanning data, the proposed method can effectively detect typical cracks under different road conditions and environments, with the detection accuracy above 98%. Furthermore, different types of deformation defects including potholes, rutting, shoving, subsidence, can also be accurately detected with location error less than 8.7%.