Title :
Two-stage non-linear algorithm of the piecewise constant image filtration
Author :
Gruzman, I.S. ; Kurilin, I.V.
Author_Institution :
Novosibirsk State Tech. Univ., Russia
Abstract :
The task of stepwise random field filtering which is observed on a background of additive Gaussian white noise is considered. For the assumed model, recurrent relationships for calculation of two-dimensional a posteriori distribution of useful and control signal probabilities are found. It is shown that the filtering mean-square error for the obtained non-causal algorithm is much less than that for a causal one
Keywords :
AWGN; data compression; image coding; image segmentation; mean square error methods; piecewise constant techniques; additive Gaussian white noise; control signal probabilities; filtering mean-square error; noncausal algorithm; piecewise constant image filtration; recurrent relationships; stepwise random field filtering; two-dimensional a posteriori distribution; two-stage nonlinear algorithm; Additive white noise; Brightness; Filtering algorithms; Filtration; Image coding; Image segmentation; Probability; Random processes; Signal processing; White noise;
Conference_Titel :
Microwave Electronics: Measurements, Identification, Application Conference, 2001. MEMIA 2001
Conference_Location :
Novosibirsk
Print_ISBN :
0-7803-6743-X
DOI :
10.1109/MEMIA.2001.982343
