Multi-modality image fusion combining sparse representation with guidance filtering

Multi-modality image fusion technique is essential for target description. The complementary information can not only compensate the limitations of each image effectively, but also enhance visual effect to human eyes. To preserve structure information and perform detailed information of each source multi-modality image, a novel fusion framework with two-scale image reconstruction is proposed. In the proposal, an improved guided image filtering (GIF)-based weighted average via Gabor energy is put forward for the fusion of base layers contained large scale structure information, and a sparse representation-based separable dictionary learning is recommended to capture small scale detailed information of detail layers. Finally, according to the texture enhancement fusion rule, the fused base and detail layers are integrated to obtain the fusion image. Experimental results demonstrate that the proposed method exhibits significant performance than the basic GIF algorithm, and also outperforms the existing state-of-the-art methods in terms of better edge texture clarity. Moreover, the fusion results show abundant information and better visual effect.