Pavement Cracking Detection Based on Three-Dimensional Data Using Improved Active Contour Model

Abstract

Three-dimensional (3D) pavement distress detection mechanisms have become the trend for pavement performance evaluation, and therefore, a comprehensive study for a high-efficiency pavement distress detection model still calls for more effort. The primary purpose of this paper is to develop an improved active contour model (ACM) using the advantages of 3D pavement data for the effectiveness and efficiency of pavement cracking detection. First, 3D data of the pavement cracking were collected based on the laser triangulation algorithm (LTA), thereby obtaining a 3D display of the pavement texture. Next, linear fitting operation was adopted to build the minimum bounding elliptical model of the crack element fragments, which were intended to be used as the initial evolution curve of the enhanced ACM. Moreover, the penalty function was introduced into the energy function of cracking images to optimize the computing complexity for the central processing unit (CPU). Verification experiments were conducted using data collected from Zhengzhou Road 4 of Xi’an and several segments of National Highway 18, Hanzhong-Mianxian, China. The results indicate that the proposed method has the tendency of giving satisfactory results under both engineering and computing complexity consideration.