Abstract:

This paper presents a multiscale saliencybased image fusion model. First, the source images were decomposed into subimage sets by contourlet transform,  then  the saliency of each piece of subimage was calculated  and  the coefficients were selected by maximizing saliency, and finally, the fused image was obtained by inversing contourlet transform. The saliency of the image was  calculated by spectral residual theory. Experimental results  show that the proposed model outperforms traditional models.

Key words: image fusion, saliency, contourlet transform, spectral residual