Research Information System
Arabian Journal for Science and Engineering, 2025 (SCI-Expanded, Scopus)
Multi-exposure image fusion (MEF) is a technique employed to generate a singular image exhibiting an extended dynamic range by integrating multiple source images, each possessing a limited dynamic range, yet all representing the same scene. Traditional block-based methods (TBB) utilize non-overlapping blocks that are regularly distributed across the source images. Subsequently, the blocks with the highest exposure values are selected for the fused image. However, this approach is prone to generating fused blocks that may harbor both overexposed and underexposed regions, even when the blocks with superior exposure values are chosen. In this paper, we propose a novel optimal block-based MEF methodology leveraging the Genetic Algorithm (GA). Unlike conventional methods, our approach imposes no constraints on the placement of the blocks. Rather, the block positions are determined optimally via the GA, thus allowing for the identification of regions with superior exposure characteristics. The GA fitness function, designed to maximize the disparity between the criteria function (CF) values corresponding to regions covered by the blocks in the source images, is employed to evaluate the optimality of the block placements. The blocks with higher fitness values are subsequently marked on the fusion map (FMap). The FMap is then reconstructed through the application of morphological operators, culminating in the synthesis of the final fused image. The resulting fused images demonstrate enhanced utility for both human and computational perception. Empirical results substantiate that the proposed method outperforms traditional approaches in terms of fusion quality.