Title :
Parameter estimation in 2D fields
Author :
Balram, Nikhil ; Moura, José M G
Author_Institution :
Dept. of Electr. & Comput. Eng., Carnegie Mellon Univ., Pittsburgh, PA, USA
Abstract :
The problem of estimating the parameters of noncausal finite lattice Gauss Markov random fields is addressed. It is shown how the structure of the potential matrix (the inverse of the field covariance matrix) can be used to specify the valid parameter space and formulate a computationally practical maximum likelihood estimation procedure. A modification that enables this to generate accurate parameter estimates from noisy data is provided
Keywords :
Markov processes; maximum likelihood estimation; parameter estimation; 2D fields; Gauss Markov random fields; MLE; maximum likelihood estimation; noisy data; noncausal finite lattice fields; parameter estimation; potential matrix; Covariance matrix; Eigenvalues and eigenfunctions; Gaussian processes; Image coding; Lattices; Markov random fields; Maximum likelihood estimation; Noise generators; Parameter estimation; Symmetric matrices;
Conference_Titel :
Acoustics, Speech, and Signal Processing, 1992. ICASSP-92., 1992 IEEE International Conference on
Conference_Location :
San Francisco, CA
Print_ISBN :
0-7803-0532-9
DOI :
10.1109/ICASSP.1992.226230
