Title :
The convergence of mean field procedures for MRFs [image processing]
Author_Institution :
Dept. of Electr. Eng. & Comput. Sci., Wisconsin Univ., Milwaukee, WI, USA
fDate :
12/1/1996 12:00:00 AM
Abstract :
Convergence of the mean field equations is analyzed for the Ising model with a nonhomogeneous external field. When β is small, the equations are a contraction mapping and thereby convergent. When β is large, however, they can be noncontracting and in this case, contraction is restored by a slight modification of the equations
Keywords :
Ising model; Markov processes; convergence of numerical methods; image processing; Ising model; MRF; Markov random fields; contraction mapping; convergence; image processing; mean field equations; mean field theory; nonhomogeneous external field; Annealing; Application software; Computational modeling; Computer vision; Convergence; Equations; Image converters; Image processing; Image restoration; Markov random fields;
Journal_Title :
Image Processing, IEEE Transactions on
