Geometrical Constraints in Bayesian Wavelet Filtering of Images

Geometrical Constraints in Bayesian Wavelet Filtering of Images

Paper in proceedings (conference paper)

This paper describes a new method for the suppression of noise in images based on wavelet transform [3]. The method relies on two criteria. The first is a traditional criterion of smoothness of the image based on an approximation of the local Hőlder exponent via the wavelet coefficients. The second, novel criterion takes into account geometrical constraints, which are generally valid for natural and also simulated images. The smoothness measure and the geometrical constraints are combined in the described method in Bayesian probabilistic formulation, and are implemented as a Markov random field (MRF) image model. The manipulation of the wavelet coefficients is consequently based on the obtained probabilities. This method is proposed to quantitatively improve noise suppression comparing to classical methods based on wavelet transform. Qualitative improvement of images is also required (subjective sensation of sharpness and contrast).

This paper describes a new method for the suppression of noise in images based on wavelet transform [3]. The method relies on two criteria. The first is a traditional criterion of smoothness of the image based on an approximation of the local Hőlder exponent via the wavelet coefficients. The second, novel criterion takes into account geometrical constraints, which are generally valid for natural and also simulated images. The smoothness measure and the geometrical constraints are combined in the described method in Bayesian probabilistic formulation, and are implemented as a Markov random field (MRF) image model. The manipulation of the wavelet coefficients is consequently based on the obtained probabilities. This method is proposed to quantitatively improve noise suppression comparing to classical methods based on wavelet transform. Qualitative improvement of images is also required (subjective sensation of sharpness and contrast).

image processing, noise suppression, Markov Random Fields, applied probability, Hőlder regularity, random number

image processing, noise suppression, Markov Random Fields, applied probability, Hőlder regularity, random number

ČERVINKA, T., PROVAZNÍK, I.

2011

01.01.2004

Brno

80-214-2635-7

Proceeedings of the 10th Conference STUDENT EEICT 2004

26

5

http://www.feec.vutbr.cz/EEICT/

http://hdl.handle.net/