Colony predation algorithm-based variational mode decomposition for multilevel color image segmentation with chaotic and simulated annealing techniques
Tirumalasetti Supraja, Kankanala Srinivas · Results in Engineering · 2025
For in-depth analysis of digital images, it is essential to divide an image into regions of interest. Image segmentation is the process used to accomplish this task and is an essential stage in digital image processing. Among the various established techniques, thresholding-based methods are widely favored due to their ease of implementation and low computational cost. However, traditional histogram-based thresholding approaches often rely on basic variations, resulting in high variability, which leads to artifacts and loss of detail. To address these limitations, a novel multilevel color image segmentation method based on Partitioned Variational Mode Decomposition (VMD) is proposed. In this approach, VMD is employed to decompose the histogram into intrinsic mode functions (IMFs), enabling more effective threshold selection. By balancing exploration and exploitation, the Colony Predation Algorithm (CPA) optimizes these segmentation and enhances convergence behavior. Additionally, Tent and Logistic chaotic maps are integrated to enhance population diversity and prevent premature convergence, while Simulated Annealing (SA) refines threshold values through probabilistic local search. Minimum Cross Entropy (MCEM) is adopted as the fitness function to further improve segmentation accuracy. Experimental results on standard benchmark images, evaluated using PSNR, SSIM, and FSIM. The suggested approach performs better than current metaheuristic based frameworks in terms of both quantitative performance and segmentation quality.