Dynamic mechanism-assisted artificial bee colony optimization for image segmentation of COVID-19 chest X-ray

The artificial bee colony optimization (ABC) algorithm operates efficiently and converges well but still suffers from the problem of easily falling into local optimum, and there is room for improving the convergence speed. For this reason, this paper proposes a dynamic mechanism-assisted ABC algorithm (EABC), which contains a dynamic approximation strategy for the optimal solution and a periodic variable food source number strategy. The dynamic approximation of the optimal solution strategy improves the swarm position update formulation and increases the pre-convergence speed of the ABC algorithm. Utilizing a periodic variable food source number scheme allows for more rapid algorithm convergence while simultaneously producing higher variability and diminishing the chances of the algorithm becoming trapped in local optima. In addition, this paper proposes a multi-threshold image segmentation (MTIS) model for COVID-19 X-ray chest images based on EABC. In this paper, the optimization performance of EABC is verified on the benchmark function of IEEE CEC 2017. The effectiveness of the EABC-based MTIS model is also validated on COVID-19 X-ray chest images.