Road Surface Crack Identification by Using Different Classifiers on Digital Images Dr HEYDAR TOOSSIAN SHANDIZ, HOSEIN GHASEMZADEH TEHRANI, HADI HADIZADEH Shahrood University of Technology, Electrical, Civil Engineering Faculty 7 Th Tir Square, P.o.Box 36155-316, Shahrood, IRAN Abstract: In this paper different classifier are used to identifying different type of cracks on road surface. As our experience shows Region Growing Classifier (RGC) method can be used to divide all surface road images in two main groups. First group covers alligator and block cracks. Longitudinal, transverse cracks and other kind of distress are put in second group. In first group, wavelet Statistic Feature Classifier (WSFC), vertical and horizontal histogram and proximity are used for classification. They help to judge about the kind of crack based on digital image from road surface. Histogram, RGC and proximity are classifiers which are used in second group. Multi layer Perceptron neural network is used to judge about the cracks. Keywords: Road Crack, Region Growing Classifier (RGC), Statistic Feature Classifier (WSFC), Multi Layer Perceptron, Pattern recognition 1 Introduction Highways and roads are a major public asset in all countries. To efficiently manage these assets road authorities need accurate, up-to-date information on the condition of their highway and road networks. For example the maintenance and rehabilitation of highway pavements in the united state requires over 17 billion dollars a year. Conventional visual and manual pavement cracking analysis methods are very costly, time consuming, dangerous, labor intensive and subjective. Automatic monitoring of some aspects of road condition, for example roughness and skid resistance, has been carried out for a number of years. However, one of the most important road quality indicators, the extent and type of cracking, has up until now been measured only by visual inspection. The result is that only very sparse sampling has been carried out, at a very high cost per kilometer, and very little information has therefore been available in this important aspect of road condition. The main idea of digital image processing methods is based on the fact that the crack pixels in pavement images are darker than the surroundings and continuous [1], [2]. Based on researchers work the distress on pavement can be categorized as follow [3-5]: A longitudinal crack which is appears along the highway. Transverses crack is a crack perpendicular to the pavement centerline. Alligator crack which is a series of interconnected cracks with many sides and sharp angled pieces. Block crack as a pattern of rectangular pieces of road surface from transverse cracks. Proceedings of the 5th WSEAS Int. Conf. on SIGNAL, SPEECH and IMAGE PROCESSING, Corfu, Greece, August 17-19, 2005 (pp220-226)