A window-based Bayesian estimator for noise removal
R.R. Schultz, R.L. Stevenson · 2002
A window-based Bayesian restoration filter is proposed which smooths noise and preserves edges. The filter estimates a sample value by optimizing a partial signal likelihood function, dependent upon a set of surrounding elements. Simulations confirm that a noisy image restored by the window-based filter compares favorably to an estimate computed using the complete observed image, even for small window sizes.