Skip to Main Content
Automation in structural health monitoring has generated a lot of interest in recent years, especially with the introduction of cheap digital cameras. This paper presents fuzzy logic and artificial neural network based models for accurate crack detection on concrete. Features are extracted from digital images of concrete surfaces using image processing which incorporates the edge detection technique. The properties of extracted features are fed into the models for detecting cracks. Two kinds of approaches have been implemented in this study: the image approach which classifies an image as a whole, and the object approach which classifies each component or object in an image into cracks and noise. The models have been tested on 205 images and evaluated on the basis of five measures of performance.