Joint map image restoration and ML parameter estimation using MRF models with explicit lines (Contributo in atti di convegno)

Type
Label
  • Joint map image restoration and ML parameter estimation using MRF models with explicit lines (Contributo in atti di convegno) (literal)
Anno
  • 1997-01-01T00:00:00+01:00 (literal)
Alternative label
  • Tonazzini Anna; Bedini Luigi; Minutoli Salvatore (1997)
    Joint map image restoration and ML parameter estimation using MRF models with explicit lines
    in IASTED International Conference Signal and Image Processings: SIP'97., New Orleans, 4-6 December 1997
    (literal)
Http://www.cnr.it/ontology/cnr/pubblicazioni.owl#autori
  • Tonazzini Anna; Bedini Luigi; Minutoli Salvatore (literal)
Pagina inizio
  • 215 (literal)
Pagina fine
  • 220 (literal)
Http://www.cnr.it/ontology/cnr/pubblicazioni.owl#titoloVolume
  • International Conference Signal and Image Processings: SIP'97. Proceedings (literal)
Http://www.cnr.it/ontology/cnr/pubblicazioni.owl#pagineTotali
  • 6 (literal)
Http://www.cnr.it/ontology/cnr/pubblicazioni.owl#affiliazioni
  • CNR (literal)
Titolo
  • Joint map image restoration and ML parameter estimation using MRF models with explicit lines (literal)
Http://www.cnr.it/ontology/cnr/pubblicazioni.owl#isbn
  • 0889862427 (literal)
Abstract
  • This paper deals with Che unsupervised edge-preserving restoration of images. when MRF models with explicit lines are considered. The approach adopted is based on thè maximization of the posterior distribution with respect lo the line fleld and ihe parameters. while the intensity field is assumed clamped to the maximizer of thè posterior itself. given thè current values of the lines and thè parameters. This enables thè application of a mixed-annealing algorithm for the MAP estimation of the image field. and of MCMC techniques, over binary variables only, for the simultaneous ML estimation of the parameters. The whole procedure is nearly as fast as a single supervised MAP restoration. (literal)
Prodotto di
Autore CNR

Incoming links:


Prodotto
Autore CNR di
data.CNR.it