Pavement distress image segmentation algorithms are essential and crucialfor developing an automatic pavement distress detection and classification system. There have been many pavement distress segmentation algorithms that have been developed in the past decade; however, there is a lack of good methods to quantitatively evaluate their performance, which hinders thefocused development of better segmentation algorithms. In this paper, a novel method is developed to quantitatively evaluate the performance of different pavement distress segmentation algorithms. This method uses buffered Hausdorff distance to estimate the deviation of the cracks in the automatically segmented image from the ground truth cracks. The proposed methodcaptures the local effectiveness of segmentation methods around the crackregion without compromising its robustness to isolated pixel deviations caused by noise. Besides real pavement images, synthetic images simulating extreme pavement distress conditions are used to evaluate the capability of the proposed method and show its merits. The proposed method outperformsfour other possible quantification methods and demonstrates its superior capability of providing a better score separation to distinguish the performance of different segmentation algorithms.
Abstract