Statistical Mechanics of Image Restoration by the Plane Rotator Model

Yôhei Saika, Hidetoshi Nishimori

Research output: Contribution to journalArticlepeer-review

17 Citations (Scopus)


On the basis of statistical mechanics formulation for problems of image restoration and error-correcting codes, we propose a new technique of image restoration for a binary image using the plane rotator model. In our formulation, the restored image is obtained from the equilibrium state of a ferromagnetic plane rotator model under a random field which consists of the corrupted image at finite temperature. The validity of our technique is evaluated by the dependence of overlap on the hyperparameters using the replica symmetric theory for the infinite-range model. The theory shows that our technique achieves the same optimal performance with that by the Ising spins. This statement is qualitatively confirmed by Monte Carlo simulations for two-dimensional images. Furthermore we estimate the dynamics of our technique by using Monte Carlo simulations. The simulations reveal that the convergence to the restored image is faster than that by the Ising model at low temperature.

Original languageEnglish
Pages (from-to)1052-1058
Number of pages7
Journaljournal of the physical society of japan
Issue number4
Publication statusPublished - 2002 Apr


  • Image restoration
  • Monte Carlo simulation
  • Plane rotator model
  • Replica theory
  • Statistical mechanics

ASJC Scopus subject areas

  • Physics and Astronomy(all)


Dive into the research topics of 'Statistical Mechanics of Image Restoration by the Plane Rotator Model'. Together they form a unique fingerprint.

Cite this