A Fast Multi-Threshold Image Segmentation Method using a Bayesian Forecasting Evolutionary Algorithm

Multilevel thresholding is one of the most commonly used techniques in image segmentation. However, it faces a critical challenge in terms of being time-consuming. To reduce computational time and address the issue of dimensionality, this paper proposes using the Bayesian Forecasting Evolutionary Algorithm (BFEA) for natural scenery image segmentation via multilevel thresholding. BFEA, which integrates the fundamental principles of evolutionary computation, relies on the probability distribution of promising solutions and Bayesian theorem. It effectively mitigates the curse of dimensionality. Extensive experiments demonstrate that our proposed algorithm outperforms state-of-the-art population-based thresholding methods in both computational efficiency and dimensionality handling. As a result, complex image processing tasks, such as automatic image or video compression, can also be effectively addressed by our algorithm.