Automated Real-Time Pavement Crack Detection and Classification

This Innovations Deserving Exploratory Analysis (IDEA) project was aimed at refining and field testing a computer vision technique using multiple cameras for automated condition survey of highway pavements. Work in the initial stage involved components acquisition, evaluation and integration into the automated pavement survey system. A new vehicle for collecting digital highway data was procured and four cameras were mounted in the rear of the vehicle to collect pavement surface images across a 4-m wide pavement. However, calibration work to correct camera distortion for 3-D surface reconstruction showed the inadequacy of the Direct Linear Transformation (DLT) method for the purpose and further work indicated that the Tsai method provided better accuracy than the DLT method. The space relationship between the two cameras also affected the calibration accuracy. While efforts were directed at improving the accuracy by adjusting each camera's angle and the space between the cameras, the research team also experimented with a new laser-based illumination imaging system for image capture and developed an automated algorithm for surveying pavement cracking with promising results. The system allows image acquisition without the influence of sunlight or shadows providing a 1-mm resolution of both longitudinal and transverse cracks at speeds up to 60 mph. However, with the line-scan camera, the stereovision technology is not directly applicable and additional work is needed to establish the 1-mm level resolution of 3-D pavement surface models with multiple laser imaging devices.

  • Record URL:
  • Supplemental Notes:
    • This NCHRP-IDEA investigation was conducted by the University of Arkansas, Fayetteville. Distribution, posting, or copying of this PDF is strictly prohibited without written permission of the Transportation Research Board of the National Academy of Sciences. Unless otherwise indicated, all materials in this PDF are copyrighted by the National Academy of Sciences. Copyright © National Academy of Sciences. All rights reserved
  • Authors:
    • Wang, Kelvin C P
    • Gong, Weiguo
  • Publication Date: 2007-5

Language

  • English

Media Info

  • Media Type: Print
  • Edition: Final Report
  • Features: Figures; Photos; References; Tables;
  • Pagination: 31p
  • Serial:
  • Publication flags:

    Open Access (libre)

Subject/Index Terms

Filing Info

  • Accession Number: 01054249
  • Record Type: Publication
  • Report/Paper Numbers: TRB-NCHRP-111
  • Files: TRIS, TRB
  • Created Date: Jul 23 2007 9:32AM