A multi-faceted adaptive image fusion algorithm using a multi-wavelet-based matching measure in the PCNN domain

Abstract Pulse coupled neural network has been applied in image fusion in recent years. In image fusion, determining how to make full use of its characteristics and guarantee the scientific value of the key parameters in the pulse coupled neural network model is particularly important. This paper proposes a novel multi-faceted adaptive image fusion algorithm based on a pulse coupled neural network. This algorithm comprehensively utilizes the advantages of multi-wavelet and pulse coupled neural networks. First, registered original images that need to be fused are decomposed using the multi-wavelet system. Second, pulse couple neural network models of the low- and high-frequency parts are established respectively. The values of several key parameters of pulse coupled neural network model are set in accordance with the information contained in the images to be fused. Third, a new image matching measure, named the pulse number matching measure, is defined. Then, image fusion is performed according to a set of fusion rules, which are constructed based on matching measure of pulse number, and a new image can be obtained after the reconstruction of the multi-wavelet system. The experimental results illustrate that the algorithm can effectively improve the entropy, standard deviation, and quality measure of the fused image. The improvement in performance is demonstrated through detailed theoretical analyses and simulations.