Feature-Based Pothole Detection in Two-Dimensional Images

Abstract

Accurately detecting potholes is an important task in determining the proper strategies for pavement maintenance and rehabilitation. However, manually detecting and evaluating methods are expensive and time-consuming. A pothole detection method is proposed in this study; the method uses various features in two-dimensional (2-D) images that improve the existing method and can accurately detect a pothole. The proposed method can be divided into three steps: ( a) segmentation, ( b) candidate region extraction, and ( c) decision. First, a histogram and the closing operation of a morphology filter are used; the process extracts dark regions for pothole detection. Next, candidate regions of a pothole are extracted with the use of features such as size and compactness. Finally, a decision is made on whether candidate regions are potholes with a comparison of pothole and background features. The 2-D asphalt images with a pothole and without a pothole extracted from a pothole database collected by a survey vehicle on national highways in South Korea were used to compare the performance of the proposed method with that of the existing method. Experimental results show that the proposed pothole detection method has better results than the existing method and that it performs well in distinguishing between a pothole and similar patterns.