基于NW小世界邻居的粒子群多阈值分割算法

Image Multilevel Thresholding Based on PSO with NW Small World Neighborhood

针对多阈值分割问题,提出了一种新的多阈值分割算法。此算法采用相对类内方差代替传统Otsu算法中的绝对类内方差,改善了传统Otsu对小对象分割不理想的弱点;采用NW小世界模型作为粒子群优化的社会认知拓扑结构,具有较好的全局寻优能力和较快的收敛速度。实验结果显示此算法具有好的性能。

As for image multilevel thresholding, a novel algorithm was proposed. Adopting relative intra-class variance instead of absolute intra-elass variance, the algorithm improved the segmentation result for small objects with Otsu method. Introducing NW small world model into particle swarm optimization, the algorithm gained better optimization performance. Experimental results show that the proposed algorithm was promising and outperformed some existing techniques.