A hybrid swarm intelligence algorithm for region-based image fusion.

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Metaheuristics

Background:

• Optimization problems often suffer from local optima and require efficient exploration/exploitation balance.
• Existing metaheuristic algorithms have limitations in handling complex, high-dimensional search spaces.

Purpose of the Study:

• To introduce a novel multi-hybrid optimization algorithm, DHPN, for enhanced performance.
• To evaluate DHPN's effectiveness on benchmark datasets and a practical image fusion problem.

Main Methods:

• DHPN integrates Dwarf Mongoose Algorithm (DMA), Honey Badger Algorithm (HBA), Prairie Dog Optimizer (PDO), Cuckoo Search (CS), Grey Wolf Optimizer (GWO), and Naked Mole Rat Algorithm (NMRA).
• Incorporates iterative division for exploration, parametric enhancements for exploitation, and a stagnation phase using CS and GWO to avoid local optima.
• Analyzes six inertia weight operators for parameter adaptation and assesses DHPN's suitability for population variations and high dimensions.

Main Results:

• DHPN demonstrated superior performance on CEC 2005 and CEC 2019 benchmark datasets compared to algorithms like JADE, SaDE, SHADE, and GWO-E.
• In image fusion tasks, DHPN achieved optimal results across metrics such as edge-based similarity index, SCD, SSIM, and artifact measure.
• Statistical tests (Wilcoxon's, Friedman's) confirmed DHPN's significant performance improvement over other tested algorithms.

Conclusions:

• The proposed DHPN algorithm offers a robust and effective approach to complex optimization problems.
• DHPN shows significant advantages in both general optimization benchmarks and specific applications like image fusion.
• The hybrid nature and adaptive parameter strategies contribute to DHPN's superior performance and robustness.