Superresolution of License Plates in Real Traffic Videos

This paper presents a method of enhancing license plate text using superresolution in Intelligent Transportation Systems (ITS) type real traffic video monitoring. First a license plate candidate is determined using a high-resolution (HR) image that is gathered by fusing multiple, noisy, subpixel shifted low-resolution (LR) images. Superresolution is acquired by using a Markov random field (MRF) as determined from a procedure of graduated nonconvexity optimization (GNC). Some difficulties with such enhancement are discussed. Often the distance between vehicle and camera is large, LR images are very noisy, and superresolution algorithms are mostly effective only when blur estimates are known accurately. The method presented is called a discontinuity adaptive MRF (DAMRF). In addition, the paper describes the use of a Gibbs distribution (GD).

  • Availability:
  • Authors:
    • Suresh, K V
    • Kumar, G Mahesh
    • Rajagopalan, A N
  • Publication Date: 2007-6

Language

  • English

Media Info

Subject/Index Terms

Filing Info

  • Accession Number: 01054556
  • Record Type: Publication
  • Source Agency: UC Berkeley Transportation Library
  • Files: BTRIS, TRIS
  • Created Date: Jun 18 2007 8:22PM