The collection and analysis of pavement distress data are a primary component of any pavement management system. Different approaches for automatic interpretation of asphalt pavement distresses, recorded on video or photographic film, with emphasis on segmentation and classification of digitised distress pavement images are examined. Segmentation deals with the problem of extracting the objects of interest from the background, whereas classification assigns distresses to corresponding distress types. Results from the application of the different methods on a data set of asphalt pavement images are presented. Alternative segmentation and classification approaches and the effectiveness of global geometrical descriptors characterising the various distress classes are evaluated. Issues associated with the accuracy and validity of the proposed methods are discussed and possible sources of error examined. Directions for further research are also identified.
Abstract