Multilevel Thresholding Method Based on Aggressive Particle Swarm Optimization

The multilevel thresholding problem based on Otsu criterion is examined in this paper. The problem is the computational time of Otsu method increases exponentially according to number of thresholds dimension. A new variant of PSO namely AgPSO is proposed. The contribution to the original PSO is highlighted by introducing an aggressive behavior to the movement rule. The proposed algorithm performance was evaluated based on three natural images. The experimental results showed the proposed PSO variant was more efficient according to other PSO variants.