The fast image fusion based on a lifting wavelet algorithm

Abstract

Image fusion has been used to derive useful information from multimodality image data. This research proposes a novel method for multimodality image fusion based on lifting scheme. In this method, the pixel matrix of input 2-D image are linearized to 1-D pixel sequence by special order on the analysis of lifting wavelet algorithm. Then, the 1-D pixel sequence is decomposed into multi-level subspaces similar to wavelet packet transform. According to the rule of image fusion, coefficients in each wavelet subspace are processed. Lastly, the image is reconstructed with fused coefficients. The whole process of image fusion presents in-place computation and parallel-processing to realize the fast multimodality image fusion. The experiment results demonstrate that the fusion scheme is more effective and suitable to hardware implementation than traditional wavelet-based image fusion.