An improved multilevel thresholding approach based modified bacterial foraging optimization

In this work, a multilevel thresholding approach that uses modified bacterial foraging optimization (MBFO) is presented for enhancing the applicability and practicality of optimal thresholding techniques. First, the diversity of solutions is considered during the reproduction step. Each weak bacterium randomly selects a strong bacterium from the healthiest bacteria, attempts to reach a location near the chosen strong bacterium, and maintains the same direction. Particle swarm optimization is subsequently incorporated into each chemotactic step to strengthen the global searching capability and quicken the convergence rate of the bacterial foraging algorithm. Finally, the optimal thresholds are obtained by maximizing the Tsallis thresholding functions using the proposed MBFO algorithm. The performance of the proposed algorithm in solving complex stochastic optimization problems is compared with other popular approaches such as a bacterial foraging algorithm, particle swarm optimization algorithm, and genetic algorithm. Experimental results show that the optimal thresholds produced using MBFO require less computation time. The devised algorithm generates more stable results, and the proposed method performs better than the other algorithms in terms of multilevel thresholding. In addition, MBFO method can achieve significantly better results than other compare algorithms on a set of benchmark functions.