Functional Evaluation of Pavement Condition Using a Complete Vision System

Abstract

The ability of a pavement to withstand traffic loads and serve the motoring public in a safe and efficient manner is adversely affected by the observable pavement surface distresses. Therefore, monitoring the performance of pavement will help to objectively determine the current condition of the pavements and, consequently, a management plan for maintenance, rehabilitation, or reconstruction. The current study builds a complete vision system and integrates its output into geographic information systems (GIS) for automated pavement monitoring and management. The system consists of an image retrieval algorithm which increases the sampling of pavement image data and reduces the computational time for processing acquired data. Also, a novel model method for edge detection, called active contours or snakes, is introduced. The active contour models show promise in detecting pavement distresses in very noisy environments. Their ability to split and match different topologies of the image data is essential for accurate crack location and shape detection. Finally, the output of the vision system is successfully integrated into GIS through concepts such as automated georeferencing and vectorization. Overall, the system is designed to overcome challenges associated with platform differences, complete automation, and processing of massive data sets.