<p>This paper introduces a new method for multiple image fusion that minimizes a nonlocal isotropic osmosis regularizer combined with a similarity term applied to a specific subregion. The proposed model captures nonlocal pixel interactions through nonlocal differential operators while also accounting for contrast variations. Using the semi-group theory, we demonstrate the existence and uniqueness of a solution for the corresponding evolution partial differential equation, and establish several properties that make the model well-suited for image processing. Experimental results show that this new method outperforms other existing state-of-the-art techniques in both visual quality and quantitative evaluation for two and multiple-image fusion, including multi-focus image fusion.</p>