Creating and sharing knowledge for telecommunications

Automatic Crack Pavement Detection Using a Bayesian Stochastic Pattern Recognition System

Oliveira, H.O. ; Correia, P.L.

Automatic Crack Pavement Detection Using a Bayesian Stochastic Pattern Recognition System, Proc Portuguese Conf. on Pattern Recognition - RecPad, Lisbon, Portugal, Vol. -, pp. - - -, October, 2007.

Digital Object Identifier:

Abstract
A good road maintenance management policy requires the implementation of an adequate maintenance management information system. One of the most important topics to include in such a system is related with the pavement condition, supported by the visual inspection of the pavement surface, notably to allow the identification of cracks. This paper proposes a stochastic pattern recognition system following a Bayesian approach, for the automatic identification of cracks in flexible pavement surfaces. The implemented approach is fully automatic, not requiring any type of input from the system operator. The proposed system has been tested on a database composed of gray images acquired by a personal digital camera with its optical axis oriented perpendicular to the pavement surface, at a constant vertical distance.