IMAGE ANALYSIS OF ROAD SURFACES

This thesis examines an image analysis technique for describing bituminous road surface conditions. Here an automated road surfac inspection focus is on the aggregate parameters as size, shape an orientation and to study whether this information can be used to assess the road surface condition. Sudden changes in the behaviou different measurement variables, indicating an accelerating wear an increasing need of repair, may be observed by regular image capturing and analysis. Digital still images of a number of sampl road surfaces were recorded at regular time intervals during a three-year period. Steel nails were hammered into the pavement an served as fixpoints, which ensured that exactly the same areas we identified on each recording occasion. The images were processed the IBAS image analysis system, where a new application was devel specially for the analysis of aggregates in road surfaces. The segmentation was performed by the use of a highly accurate edge-b segmentation technique, Bergholm's edge focusing algorithm, in combination with median thresholding. The results show that image analysis can be used to measure various properties of aggregates road surfaces, provided that the surface is dry and that the aggregates are not entirely covered by bitumen binder. The result indicate that aggregates in the wheel track differ significantly the rest of the road surface with respect to several parameters. main objective was the analyses of images of exactly the same are captured on different occasions, which showed significant differe between the samples, although no clear trends or breakpoints coul identified. The accuracy of the program was carefully evaluated a the tests showed notable deviations between manually and automatically processed images. Moreover, there was an unexpected co-variation between the data of all test road surfaces, possibly the lighting conditions and, thus, that the desired constant conditions were not obtained. (A)