The framework is presented of Bayesian image restoration for multi-valued images by means of the Q-Ising model. Hyperparameters in the probabilistic model are determined so as to maximize the marginal likelihood. Practical algorithms are described based the conventional mean-field approximation and loopy belief propagation. We compare the results empirically with those provided by conventional filters and the new methods are found to be superior.
|Number of pages||4|
|Journal||Proceedings - International Conference on Pattern Recognition|
|Publication status||Published - 2004 Dec 17|
|Event||Proceedings of the 17th International Conference on Pattern Recognition, ICPR 2004 - Cambridge, United Kingdom|
Duration: 2004 Aug 23 → 2004 Aug 26
ASJC Scopus subject areas
- Computer Vision and Pattern Recognition