Image-based pavement crack detection by percolation theory

Asphalt pavement performance may deteriorate so rapidly because the desired service level of local governments is reduced. Thus, there are a lot of road damages like cracks, potholes, and ruts on some pavements. If they remain in the condition of being inconvenient, road-traffic accidents such as vehicle crash and fall of motorcycle would be occurred. Moreover, if they get worse further, there are possibilities that the pavement safety performance cannot be maintained and the structures may collapse. Hence, it is necessary to diagnose the pavements for safety and to carry out proper maintenance on a regular basis. In the present study, we present the approach for highly accurate crack detection based percolation theory. The shadow reduction approach is used to detect cracks more accurately. Experimental results show that many kinds of cracks can be detected correctly and fast with the adaptive search window method.