Optimizing image segmentation of pavement defects using graph-based method

Pavement defect detection and classification systems based on machine learning algorithms are already very advanced and are increasingly demonstrating their outstanding advantages. One of the most important steps in the processing is image segmentation. In this paper, some image segmentation algorithms used in practice are presented, compared and evaluated. The advantages and disadvantages of each algorithm are evaluated and compared based on the criteria PA, MPA, F1. We propose a method to optimize the process of image segmentation of pavement defects using a combination of Markov Random Fields and graph theory. Experiments were conducted on 3 datasets from Portugal, Russia and Vietnam. Empirical results show that the segmentation of pavement defects is more accurate and effective when the two methods are combined.