An Edge-Based Approach to Image Fusion of Images via Reconstruction Estimation

Image fusion algorithms attempt to combine multiple registered images into a single image in a way which retains the most pertinent information from each of the images to be fused. In this paper, a new edge-based image fusion algorithm using the Parameterized Logarithm Image Processing (PLIP) model is presented. Coarse approximation and edge information from the images to be fused are extracted using PLIP primitives and separately processed. The results are fused to yield the reconstructed fusion result. The Logarithmic Michelson contrast measure by entropy (log AMEE) is used to quantitatively access the quality of image fusion results. Experimental results via computer simulations show that the proposed algorithm can outperform current image fusion techniques both by qualitative and quantitative means.