An efficient image fusion method based on dual tree complex wavelet transform

Image fusion is a process which combinesinformation from two or more images of the same scene into asingle image preserving important features from each. Theobjective of image fusion is to combine complementaryinformation from multiple images into a single resultant imagewhich is more informative, comprehensive, reliable and precisecompared to each source images. Fusion is an effective tool withinvarious fields such as remote sensing, robotics and medicalapplications. In the proposed method the source images were firstdecomposed using the shift invariant and directionally selectivedual tree complex wavelet transform (DT-CWT) and then fusionrules namely, max and local energy were applied to combine lowand high frequency coefficients respectively. The final fused imagewas obtained by applying inverse DT-CWT to the fused low andhigh frequency components. The obtained fused images wereanalyzed qualitatively as well as with various quantitative metricsnamely mutual information (MI), structural similarity index(SSIM), entropy, standard deviation (STD) and average gradient(AVG) where both analysis show that the proposed scheme canprovide more information and better results compared to theother methods.

Keywords:

Image fusion , DT-CWT , WT , Quantitative metrics , fusion rules