Image fusion method can provide a high-quality image by merging the multiple features of different source images, and how to effectively evaluate the quality (informativeness) of image features is an important issue for image fusion. Because a considerable amount of imprecise and uncertain information exists in image fusion processes, this paper proposes a framework based on fuzzy set theory to handle the vague features, and a set of hybrid optimization methods is also designed to improve the performance. First, the two-scale decomposition method is utilized to decompose the source images and obtain a set of corresponding subimages. Second, fuzzy set theory and local spatial frequency are employed to generate preliminary decision maps by evaluating the pixel quality of the subimages. Third, a morphological method and consistency verification are utilized to optimize the decision maps to extract the focused and unfocused regions. Finally, three schemes are designed to generate the fused images according to the optimized decision maps. The experimental results show that the proposed method can achieve competitive performance compared with other methods. (C) 2020 Elsevier Inc. All rights reserved.