Bio-inspired algorithms for multilevel image thresholding

Bi-level image thresholding methods can be easily extended to multilevel cases. However, extended versions are computationally expensive. In this paper, we propose first a differential evolution (DE) algorithm using Tsallis entropy as objective function. Second, we conduct a comprehensive comparative study by investigating the potential of the proposed algorithm to find the optimal threshold values along with two other bio-inspired algorithms namely artificial bees colony (ABC) and particle swarm optimisation (PSO). Two entropy-based measures have been considered as objective functions. Real images with different complexities have been used to evaluate the performance of the three algorithms. Experimental results demonstrated that DE and ABC achieve the same quality of solutions in terms of peak signal to noise ratio values and Uniformity values. They are more robust than PSO. Furthermore, DE has shown to be the most stable and ABC the fastest with the advantage of employing few control parameters.