Blind frequency deconvolution : A new approach using mutual information rate
Résumé
In this paper, we purpose a blind frequency deconvolution method using higher order statistics by minimizing the mutual information rate of the deconvolved output. We add to the criterion a regularization term to limit noise amplification. Then, we compare on real data (underwater explosions and seismovolcanic phenomena) our deconvolution algorithm with existing methods.