A ROBUST APPROACH FOR ROAD PAVEMENT DEFECTS DETECTION AND CLASSIFICATION

Abstract

The objective of this paper is to propose a robust approach to building a computer vision system to detect and classify pavement defects based on features, such as the contour of feature (chain code histogram, Hu-moment), the shape of an object (length, width, area). In this paper, we present a method to build an automated system to detect and classify the different types of defects such as rupture of the road edge, potholes, subsidence depressions based on image processing techniques and machine learning methods. That system includes the following steps. First step is to detect defect position (ROI) then the defect is described by its features. Finally, each defect is classified these different defect features. In our approach the following algorithms have been using: Markov Random Fields and Graph cuts method for image segmentation, Random Forests algorithm for data classification.