Modular traffic signs recognition applied to on-vehicle real-time visual detection of american and european speed limit signs

A new modular traffic signs recognition system is presented, successfully applied to both American and European speed limit signs. The sign detection step is based only on shape-detection (rectangles or circles). This enables it to work on grayscale images, contrary to most European competitors, which eases robustness to illumination conditions (notably night operation). Speed sign candidates are classified (or rejected) by segmenting potential digits inside them (which has several advantages), and then applying a neural digit recognition. The global detection rate is about 90% for both (standard) USA and EU speed signs, with a misclassification rate <1%, and no validated false alarm in >150 minutes of video. The system processes in real-time about 20 frames/s on a standard high-end laptop. For the covering abstract see ITRD E140665.

• Authors:
  • MOUTARDE, F
  • BARGETON, A
  • HERBIN, A
  • Chanussot, Lowik
• Publication Date: 2007

Language

• English

Media Info

• Serial:
  • PROCEEDINGS OF THE 14TH WORLD CONGRESS ON INTELLIGENT TRANSPORT SYSTEMS (ITS), HELD BEIJING, OCTOBER 2007
  • Publisher: ITS America

Subject/Index Terms

• TRT Terms: Conferences; Detection and identification; Image processing; Intelligent transportation systems; Numbers; Real time data processing; Software; Speed limits; Traffic signs
• Geographic Terms: Europe; United States
• Subject Areas: Operations and Traffic Management; I73: Traffic Control; I91: Vehicle Design and Safety;

Filing Info

• Accession Number: 01121462
• Record Type: Publication
• Source Agency: TRL
• Files: ITRD
• Created Date: Feb 5 2009 8:59AM