Abstract: A method of image segmentation based on 3-D threshold and Predator Prey Particle Swarm Optimization（PPPSO） is proposed. The method not only retains the traditional 2-D Otsu method’s advantages, but also makes full use of the fuzzy information of the image-fuzzy median value, which with gray value and average gray value constitute a 3-D feature vector. In addition, PPPSO is used to search for the best threshold, which can reduce the search time greatly and converge to global optimum rapidly. Experimental results show that the improved method has a good anti-noise and real-time performance.

Key words: image segmentation, 3-D threshold, predator-prey, particle swarm optimization

摘要： 提出一种基于三维阈值和捕食-被捕食粒子群（PPPSO）的图像分割方法。该方法在保留传统二维最大类间方差（Otsu）分割算法优点的基础上，充分利用图像自身模糊信息-模糊中值，该特征维与像素灰度值、邻域均值组成一个三维矢量集；另外，通过采用捕食-被捕食的粒子群优化方法搜索最佳分割阈值，大大缩短了搜索时间，且能快速收敛到全局最优。实验结果表明，该方法具有较好的抗噪性和实时性。

关键词: 图像分割, 三维阈值, 捕食-被捕食, 粒子群优化