Image-based Method for Concrete Bridge Crack Detection

摘要

This paper presents a modified image processing method for detecting cracks in surface images of concrete bridges. Clip, filling and rotation transformation are applied on concrete images for precise extracting cracks. Modified C-V model, adaptive threshold, morphology, C-V model and Canny are compared on segmenting crack images which are gathered in different illumination. Analysis results indicated that the misclassification rate of the modified C-V model is 3.56% and the operation time is 96 ms. Therefore, the proposed method can be used for the accurate detection of cracks in surface images recorded under various conditions. Moreover, the estimated crack widths are in good agreement with those measured manually.