Multi-modal image fusion using window-based ICA and fractal dimension

The goal of multi-modal image fusion is to combine complementary information from multisensory data such that the fused image is more suitable for the purpose of human visual perception, computer-processing tasks and detection applications. In this paper, we first use independent component analysis (ICA) as the primary transformation to obtain adaptive analysis bases trained by similar reference images; we then utilize Fractal Dimension (FD) to detect textural information and an intensity histogram to sense objects of interest as a new additional fusion rule. Substantiating examples show an improved performance in perception and in an enhancement of textural information and objects of interest.