Automatic real-time identification of road surface type from the texture profile

Experience has shown that up to date information regarding the surface type is of benefit when interpreting the pavement condition from routine survey data. In particular, this information is valuable in the assessment of cracking data and the estimation of the noise levels. This work has been carried out to develop a method of deriving the pavement surface type from the data collected routinely for the Trunk Road network in England. The method is based on characteristic differences between the distributions of parameters derived from the texture profiles for different surface types. These characteristics are combined in a probabilistic manner to derive the most likely surface type. The luminosity of the pavement obtained from images is also included to improve the distinction between bituminous and concrete surfaces. At present, the surface type prediction algorithms can identify Hot Rolled Asphalt, thin surfacings, and brushed and grooved concrete surfaces. Tests have shown that overall the correct surface type is assigned in 86% of cases. However, this success rate is higher on HRA and concrete surfaces where over 95% of cases are correctly identified. On thin surfacings the success rate is lower, but this stems from the requirement that, to improve the assessment of cracking, the method has been developed to have a low level of false positive reports of thin surfacings.

Language

  • English

Media Info

  • Pagination: 11p

Subject/Index Terms

Filing Info

  • Accession Number: 01389561
  • Record Type: Publication
  • Source Agency: ARRB
  • Files: ITRD, ATRI
  • Created Date: Aug 23 2012 3:35AM