Multifocus Image Fusion Based on Variational Decomposition in Combination with Reconstruction of Vector Fields

An image fusion scheme based on variational decomposition in combination with structure tensor analysis is presented in this paper. In the scheme, each source image is decomposed into a geometrical component and a textured component by applying Rudin-Osher-Fatemi (ROF) model in combination with Chambolle´s projection algorithm. Hereafter, the different components are fused separately. For fusing the geometrical components, a weighted average method is adopted. In order to effectively reconstruct the textured composite which corresponds to the divergence field of a vector field, a smoothing algorithm based on vector field denoise is developed to smooth the composite, then the structure tensor analysis method is adopted to compute the reconstruction result. Finally, the sum of the fused composites is calculated to obtain the fusion result. Experiments show that the proposed algorithm works well in multi focus image fusion.