Maximum a Posteriori Maximum Entropy Signal Denoising
Abd‐Krim Seghouane, Luc F. Knockaert, Kevin H. Knuth, Ariel Caticha, Adom Giffin, Carlos C. Rodríguez · AIP conference proceedings · 2007
When fitting wavelet based models, shrinkage of the empirical wavelet coefficients is an effective tool for signal denoising. Based on different approaches, different shrinkage functions have been proposed in the literature. The shrinkage functions derived using Bayesian estimation theory depend on the prior used on the wavelet coefficients. However, no simple and direct method exists for the choice of the prior. In this paper a new method based on maximum entropy considerations is proposed for the construction of the prior on the wavelet coefficients. The new shrinkage function is obtained by coupling this prior to maximum a posteriori arguments. A comparison with classical shrinkage functions is given in a simulation example of image denoising in order to illustrate the effectiveness of the proposed thresholding method.