The role of image segmentation in comprehending and analyzing images is of paramount importance. One of the key aspects of segmentation is thresholding, which involves classifying pixels based on their intensity values. Determining a single threshold value is simple; however, determining multiple threshold values becomes complex as the number of thresholds increases. Several metaheuristic approaches have been proposed to find the optimal multilevel thresholds. This study proposes an automated design approach for image thresholding. The proposed method is a 2-layered approach that uses two metaheuristics: differential evolution and a genetic algorithm. The differential evolution (DE) algorithm is configured in the first layer and searches in the design space of a genetic algorithm (GA) in the second layer. The GA works in the solution space of multilevel thresholds. Experiments were performed on benchmark test images using various thresholds. The results were analysed using a statistical tool and suggest that the proposed approach can be efficiently utilised for multilevel thresholding. The proposed approach performed better than the standard genetic algorithm approach on tested benchmark images.