Medical image fusion using non subsampled contourlet transform and iterative joint filter

This study proposes an improved medical image fusion scheme based on components of non subsampled contourlet transform (NSCT) and iterative joint filter. Multimodal images are split into approximation and detail components using NSCT. The former are subsequently normalized and further smoothed using box filter. The underlying morphological structure of the smoothened components is obtained with the help of gradient operator using Wiener filter. The filtered structures are then used to compute decision map. Iterative joint filter is finally applied on the decision map along with input guidance image to compute the resultant image. Eight performance metrics as well as qualitative visual evaluation shows the efficacy of the proposed fusion scheme.