Automatic Road Pavement Crack Detection using Boosting Classifiers
Rosa, P. R.
Automatic Road Pavement Crack Detection using Boosting Classifiers, Proc European Signal Processing Conference EUSIPCO, Glasgow, United Kingdom, Vol. ., pp. . - ., August, 2009.
Digital Object Identifier:
Detecting cracks on pavement surfaces has a crucial importance for the good maintenance of roads, allowing a high quality of transportation services and supporting the population development. This paper proposes an automatic crack detection and classification system, capable of identifying and retrieving pavement surface images containing cracks from a road pavement survey image database. Crack regions are detected using selected statistical features and learning algorithms based on boosting. The classification step distinguishes longitudinal, transversal and miscellaneous cracks.
Crack detection and classification results are evaluated by comparison with a human labeled test set. The experimental results, obtained using images captured along Portuguese roads, are encouraging.