تحلیل مقایسه‌ای روش‌های بخش بندی تصویر برای سنجش عرض ترک در سطوح بتنی

Cracks in concrete structures are considered to be one of the most important indicators of structural degradation and durability, and accurate measurement of their width plays a key role in structural health monitoring. In this study, three common image processing approaches for calculating crack width in segmented images, including the skeletonization and perpendicular distance method, the intensity profile method, and the connected component analysis method, are evaluated and compared. For all three methods, a set of 50 real images of cracked concrete surfaces along with real crack width values ​​are used, and the criteria of mean absolute error, percentage relative error, and noise stability are investigated. The results show that the skeletonization method provides the most accurate performance in narrow cracks, while the intensity profile method performs better in wider cracks, and the connected component method shows higher noise stability. Considering that according to the ACI224R-01 standard, a structural crack has a minimum width of 0.1 mm, the supplementary evaluation showed that the skeletonization method has the most ability to accurately detect and measure structural cracks. Based on the findings, the combination of skeletonization and connected component analysis can be proposed as an efficient and stable approach for automatic crack width measurement.