Texture and noise features for road pavement identification and classification

Abstract

Road planners and road administrators frequently face the problem of insufficient knowledge of the correlation between the type of road surface and the resulting noise emission. The aim of this research is to establish the relationship between road surface characteristics, such as macrotexture, and tire/pavement noise emission, in order to identify and classify road surfaces by using statistical learning methods, which is a non-destructive approach. For this purpose, several road sections with different pavement types were selected and tested. The Close-Proximity (CPX) method was adopted, as it is commonly used to register the traffic noise in near field conditions. In its turn, macrotexture of each surface was provided by a high speed profilometer, which is one of the parameters required for the assessment of the performance and conformity of road pavements. The set of features extracted from the noise emission profile and from the surface texture was applied to a statistical classifier for evaluation. A correct identification of the road pavement leads to better data, thus enhancing the accuracy of road noise predictions. Results are presented and discussed.