Abstract: The artificial fish swarm algorithm suffers from poor convergence and easy to fall into local optimum. Therefore, a new method is combined with the cloud theory and the artificial fish swarm algorithm and applied to the multi-threshold method for image segmentation effectively. The improved algorithm improves the ability of fish learning. It avoids the fish falling into local optimum and gets the best thresholds for image segmentation. The simulations and experiment results show that the new algorithm can achieve a stable, fast, vivid image segmentation.

Key words: artificial fish swarm algorithm, cloud theory, multi-threshold, image segmentation

摘要： 针对传统鱼群算法在处理图像分割时收敛速度慢，容易陷入局部最优等问题，将云模型和人工鱼群算法结合并有效地应用到多阈值图像分割中。改进后的算法使人工鱼学习能力有所提高，同时满足种群多样性和收敛速度快的特点，避免局部最优得到图像分割的最佳阈值。仿真实验表明，该算法能得到较稳定、快速、清晰的图像分割。

关键词: 鱼群算法, 云模型, 多阈值, 图像分割