To effectively improve the dynamic range of visible light imaging devices and reduce the dependency of fused image quality on the number of source images, a dual-exposure image fusion algorithm based on multi-scale decomposition is proposed. This method requires only one set of underexposed and overexposed images as source images to obtain a richly informative image through exposure fusion. Firstly, adaptive exposure adjustment is performed according to the characteristics of the underexposed and overexposed images, fully exploring the potential details in the images. Then, edge intensity, exposure suitability, and color saturation of the image sequence are extracted as evaluation indicators to construct a fusion weight map. Finally, the fused image is obtained through pyramid multi-scale decomposition and weighted fusion. Fifteen groups of image sequences were selected for experiments, and comparisons were made with four representative algorithms from both subjective and objective perspectives. The experimental results show that, compared to other algorithms, the proposed algorithm improves overall image quality by 4.9% and has a stronger ability to retain detailed information.