查看文章

Maintenance of good condition of roads are very essential to the economy and everyday life of people in a every country. Road cracks are one of the important indicators that show degradations of road surfaces. Inspection of roads that have been done manually took a very long time and tedious. Hence, an automatic road crack segmentation using thresholding methods have been proposed in this study. In this study, ten road crack images have been pre-processed as an initial step. Then, normalization techniques, L1-Sqrt norm have been applied onto images to reduce the variation of intensities that skewed to the right. Then, thresholding methods, Otsu and Sauvola methods have been used to binarize the images. From the experiment of ten road crack images that have been done, normalization technique, L1-Sqrt norm can help to increase performance of road crack segmentation for Otsu and Sauvola methods. The results also show that Sauvola method outperform Otsu method in detecting road cracks.