Multispectral Image Fusion Based on Laplacian Pyramid Decomposition with Immune Co-evolutionary Optimization

摘要

In this paper, a new multispectral image fusion algorithm is proposed by considering the abundant directional information and complex high frequency of source images. In this algorithm, Laplacian pyramid (LP) is constructed to filter the source images to obtain initial fusion coefficients, then an evolution computation idea, immune co-evolutionary optimization algorithm, is introduced into the image fusion to optimize the fusion coefficients. The optimized fusion coefficients can make better fusion result. Simulation experiments of clearly demonstrate the superiority of this proposed algorithm by comparing with conventional wavelets systems: the information entropy (IE) values keep at a similar level, the average grads (AG) values are higher, the standard deviation (STD) values increases 1.2 averagely, and the time efficiency increases averagely about 51%.