Hybrid image fusion scheme using self-fractional Fourier functions and multivariate empirical mode decomposition

• Self-fractional Fourier function (SFFF) decomposition components are orthogonal to each other image fusion using empirical mode decomposition of SFFF component may provide better fusion results.
• Empirical mode decomposition (EMD) based image fusion in intermediate of spatial and frequency domains is investigated.
• We generalize the scale based image fusion for multi-channel (multivariate) images in fractional Fourier transform domains.
• Proposed method provide flexibility in number of sensors, type of images (real, complex etc.) and number of fusion rule to improve fusion quality.

Image fusion has emerged as a promising area of research and a bivariate empirical mode decomposition based fusion scheme has recently been proposed in the literature. In this paper, a hybrid fusion scheme combining self-fractional Fourier function (SFFF) decomposition and multivariate empirical mode decomposition is proposed. In the proposed image fusion technique, images to be fused are decomposed into SFFF images. The SFFF images are further decomposed into intrinsic mode functions (IMFs) using multivariate empirical mode decomposition (MEMD). Corresponding IMFs of same decomposition level of SFFF images are fused using local variance based adaptive weight fusion rule to obtain fused IMF images. The fused image is obtained by applying inverse transformation on fused IMF images. The proposed technique provides flexibility in the number of functions in the SFFF decomposition, transform before SFFF decomposition, and the types of source images (real and complex) to be fused. Simulations are performed for fusion of test images with different SFFF decomposition levels and the results are compared with other existing methods. It is seen that the simulation results are comparable to the existing schemes.