Visible and Infrared Image Fusion Using Convolutional Dictionary Learning with Consensus Auxiliary-Auxiliary Coupling

Abstract

To preserve details of source visible images and target of source infrared images, infrared and visible image fusion method based on convolutional dictionary learning with common auxiliary coupling is proposed in our paper. First, dictionary filters are obtained using convolutional dictionary learning with ADMM consensus. Then, each test visible-infrared images are decomposed into low-pass and high-pass components, and the sparse coefficient is structured using a convolutional sparse representation. Finally, image reconstruction is performed to obtain the fusion results. In comparison with the ASR-based algorithm, our method produces an improvement of 2.10% and 0.05% in QTE and QNCIE, respectively. The experimental results show that our fusion algorithm offers the advantages of higher clarity, contrast and information entropy than that of conventional algorithms.