Metaheuristic Algorithms based Multi-objective Optimization for Image Segmentation

In this paper a multi-threshold image segmentation procedure based on nature-inspired multi-objective optimization is proposed. The Particle Swarm Optimization, Black Hole and Gravitational Search algorithms were adapted for multi-objective optimization. The Root Mean Square Error and the number of segmented regions were used as optimization criteria. The three procedures were applied for human silhouettes detection in video sequences and the obtained results are compared. Concerning the algorithms performances, the experiments revealed that the results of multi-objective Black Hole algorithm based segmentation are better than those of the other two algorithms, at least for the test images used in this experiment.