<p>Achieving high precision in image segmentation within complex and noisy environments remains an ongoing challenge in image processing. This paper introduces the differential evolution fuzzy multilevel thresholding (DEFMT) model, a ground-breaking algorithm designed to overcome these challenges by combining the strengths of fuzzy multilevel thresholding (FMT) and differential evolution (DE). The DEFMT model addresses inherent segmentation ambiguities through fuzzy logic while leveraging DE for adaptive and dynamic threshold optimization. Using the Euro SAT Dataset (ES), the DEFMT model is meticulously evaluated to demonstrate its ability to achieve superior segmentation accuracy. Key contributions include an innovative integration of fuzzy logic for capturing intricate image details and DE’s optimization capabilities, which enable precise parameter tuning tailored to the dataset’s complexity. The methodology incorporates rigorous preprocessing for noise reduction and data uniformity, setting the foundation for reliable segmentation. Additionally, the model’s performance is validated using the ResNet152V2 architecture, achieving a remarkable 95% classification accuracy. Benchmark comparisons underscore the DEFMT model’s adaptability and superiority over traditional and state-of-the-art techniques. By tackling segmentation challenges head-on, this work establishes DEFMT as a transformative approach with broad applicability, particularly in precision-critical domains such as medical imaging and remote sensing.</p>