A modified quantum-behaved particle swarm optimization algorithm for image segmentation

Multilevel thresholding is a popular method for image segmentation applications. Traditional methods comprehensively search the optimal thresholds to make optimal the predefined objective function. If the number of thresholds increases, the computational time of these methods grows exponentially. One of the most popular algorithms is Otsu method and operates based on maximization of between classes variance to find the best optimal thresholds. In the recent years, many scientists concentrated on the population based algorithms like PSO (particle swarm optimization) and another PSO family to save the computation time. In this paper, we introduce a modified cooperative method CGQPSO (cooperative-Gaussian-quantum-behaved PSO) based on GQPSO. The method which is proposed in this paper, can reach the best position faster than CQPSO. We use Otsu´s method as fitness function. The experimental results show that, the proposed algorithm gets results more stable than CQPSO algorithm in the small number of population and algorithm iteration. Moreover CGQPSO have computation time less than CQPSO. So we can implement this algorithm for object recognition on the moving targets.