Title :
Loopy belief propagation and probabilistic image processing
Author :
Tanaka, Kazuyuki ; Inoue, Jun-ichi ; Titterington, D.M.
Author_Institution :
Graduate Sch. of Inf. Sci., Tohoku Univ., Sendai, Japan
Abstract :
Estimation of hyperparameters by maximization of the marginal likelihood in probabilistic image processing is investigated by using the cluster variation method. The algorithms are substantially equivalent to generalized loopy belief propagation.
Keywords :
belief maintenance; image processing; inference mechanisms; maximum likelihood estimation; optimisation; pattern clustering; cluster variation method; hyperparameter estimation; loopy belief propagation; marginal likelihood maximization; probabilistic image processing; Bayesian methods; Belief propagation; Clustering algorithms; Computer vision; Degradation; Image processing; Markov random fields; Pixel; Systems engineering and theory; Uniform resource locators;
Conference_Titel :
Neural Networks for Signal Processing, 2003. NNSP'03. 2003 IEEE 13th Workshop on
Print_ISBN :
0-7803-8177-7
DOI :
10.1109/NNSP.2003.1318032
