Abstract—Recently, oversampled graph filter banks (OSGFBs) have been constructed on graph theory for signal processing in [1]-[7]. This paper introduces a new image fusion for medical images in OSGFBs. Images are decomposed to subband images by spectral graph wavelet transform in OSGFBs, revised by a simple fusion rule, and synthesized back to make a fused image. Since the OSGFBs have good capability to decompose regular/irregular signals, the proposed algorithm shows better performance than traditional fusion algorithms even with a simple fusion rule. The proposed algorithm is effective particularly for medical images. Visual and numerical performance comparisons of the proposed algorithm with traditional image fusion algorithms are included for medical, multifocus, and infrared images. 
 

Index Terms—image fusion, graph filter banks, medical images, oversampled, graph signal processing

Cite: Min-Sung Koh, "Medical Image Fusion in Oversampled Graph Filter Banks," International Journal of Signal Processing Systems, Vol. 4, No. 4, pp. 318-322, August 2016. doi: 10.18178/ijsps.4.4.318-322