1. Home
  2. TRID
  3. View Record
https://trb-annual-meeting.nationalacademies.org/registration

Automated detection and measurement of potholes from road surface video clips

The assessment of potholes is an essential part of a road maintenance system for developing repair and maintenance strategies. Nowadays, road condition assessment is gradually being automated by using various imaging systems. However, the processing of the collected raw visual data is still being done manually or semi-automatically in India. Existing methods of data processing are very costly, time consuming and mostly focused on the detection and classification of cracks only. This paper presents an efficient and accurate algorithm for automated detection and measurement of potholes out of a large volume of road surface video clips of Indian Highways. In the proposed algorithm, potholes are detected and their features are measured automatically out of a video clip in one go, using various image processing techniques supported by user defined decision logic. This algorithm is implemented in a Windows environment with the help of Visual Studio 2008 and OpenCV, an open source computer vision library, and tested on nine different road video clips. The results show that this algorithm can detect and measure potholes automatically with accuracy up to 94%. The information extracted using the proposed algorithm can be used for determining maintenance levels of Indian roads and taking further appropriate actions for repair and maintenance.

  • Supplemental Notes:
    • Abstract used with permission from the publisher.
  • Authors:
    • Lokeshwor, Huidrom
    • Das, Lalit K
    • Sud, S K
  • Publication Date: 2012-7

Language

  • English

Media Info

  • Media Type: Print
  • Features: Figures; Photos; References; Tables;
  • Pagination: pp 9-18
  • Serial:

Subject/Index Terms

Filing Info

  • Accession Number: 01473328
  • Record Type: Publication
  • Files: TRIS
  • Created Date: Feb 20 2013 9:40AM