Pothole Detection and Classification Using 3D Technology and Watershed Method

Abstract

Potholes are one of the roadway distresses that negatively impact roadway safety. With emerging sensing technology, three-dimensional (3D) pavement data, derived using 3D laser technology, have become available for detecting cracking and rutting. This paper presents a pothole detection method using 3D pavement data and a watershed method. Tests using the 3D data collected on 10th Street, Atlanta, Georgia and 6 mi of roadway on U.S. 80, Savannah, Georgia, has shown a 94.97% accuracy, 90.80% precision, and 98.75% recall. It has been demonstrated that the proposed method is promising for pothole detection and can provide a reliable method for pothole detection, especially when 3D pavement data have been collected for crack detection and already available.