Recalage non-linéaire des images médicales par maximisation de l'information mutuelle quadratique
Jamal Atif, Xavier Ripoche, Cedric Coussinet, A. Osorio · 2003
We present a new highly-accurate similarity criterion for medical image registration, inspired from mutual information. The novelty of our approach lies first, in the use of quadratic mutual information, and second, in a new adaptive kernel density estimation which does not require any prior information on the image. Practically, our method exhibits a drastic decrease of the computation cost compared to conventional adaptive kernels.