01624786

Component

Pavement crack is one of the important indicators of pavement conditions that are necessary for the correct and timely maintenance operations in transportation agencies. Video log pavement images are collected to identify the cracks automatically or semi-automatically. However, due to the device noise, improper lighting, pavement textural mixture, surface debris, etc, imaging conditions of pavement crack images vary a lot. Automatic detection of pavement crack from those images remains a challenge. This paper is motivated to develop a new perspective of characterizing pavement crack. Unlike the traditional edge based crack detection, the proposed algorithm locates crack pixels in the pavement image using the directional curvature along the 3-D valley of pavement crack. Continuous cracking is composed of the connection of neighboring profile of 3-D crack valley along the direction of cracking. Curvature at different point of the profile varies with its different depth from the pavement surface. A normalized pavement surface is obtained by Gaussian smoothing of different imaging conditions. Multi-scale directional curvature of the crack valley profile is then calculated to best accommodate with its actual scale. With the strategy of edge drawing, false detection of short crack-like valley structure is controlled by the minimal level of significance of the interconnection of the optimal directional crack curvature. Centerline of the pavement crack valley is finally delineated using the morphological and thinning operations. The proposed algorithm is tested on a diverse set of actual pavement images taken on interstate highway G-4 near Beijing, China under varying lighting conditions and shadows. A buffered hit percentage is used to quantitatively evaluate the accuracy of the proposed algorithm. Experimental results show the proposed algorithm detects well the crack pixels with the hit percentage of 100\%. It is promising for the automatic classification and severity evaluation of pavement crack in the intensity and range images.

This paper was sponsored by TRB committee AFD20 Standing Committee on Pavement Condition Evaluation.

01618707

17-04154

English

15p

2017

Transportation Research Board 96th Annual Meeting

Digital/other

Figures; References; Tables

Algorithms; Pavement cracking

Automatic detection; Cracking severity; Pixels; Video images

Beijing (China)

Highways; Pavements

Transportation Research Board Annual Meeting 2017 Paper #17-04154

TRIS, TRB, ATRI

Dec 8 2016 11:35AM