改进粒子群优化的铁谱图像聚类分割-《集美大学学报（自然科学版）》

文章信息/Info

Title:: Clustering Segmentation of Ferrograph Images Based on Improved Particle Swarm Optimization

作者:: 李浩天1; 2; 戴乐阳1; 2; 王永坚1; 2; 宋佳声1; 2; 1.集美大学轮机工程学院，福建厦门361021；2.福建省船舶与海洋工程重点实验室，福建厦门 361021

Author(s):: LI Haotian1; 2; DAI Leyang1; 2; WANG Yongjian1; 2; SONG Jiasheng1; 2; 1.School of Marine Engineering，Jimei University,Xiamen 361021,China;2.Fujian Provincial Key Laboratory of Naval Architecture and Ocean Engineering,Xiamen 361021,China

Keywords:: Ferrograph image segmentation; clustering; particle swarm optimization; simulated annealing

摘要:: 在采用K均值聚类算法对铁谱图像进行分割时，由于初始聚类中心的随机性和铁谱图像的复杂性，常出现误分割现象，影响后续铁谱分析的效果。为了解决这一问题，采用基于归一化RGB颜色模型，引入模拟退火优化的自适应粒子群算法，对K均值聚类方法进行优化，可以有效缓解误分割问题，并取得全局最优搜索能力和收敛速度之间的平衡。实验结果表明，改进算法提高了铁谱图像的分割精度，并保证了运行效率。

Abstract:: When using K means clustering algorithm to segment ferrography images，due to the randomness of initial clustering center and the complexity of ferrography image，missegmentation will occur，which will affect the effectiveness of subsequent ferrography analysis.In order to solve the problem,based on the normalized RGB color model，the K means clustering method is optimized by introducing the simulated annealing optimization adaptive particle swarm optimization algorithm，which can effectively alleviate the problem of missegmentation.At the same time，the algorithm can achieve a balance between the global optimal solution searching ability and the convergence speed.The experimental results show that the improved algorithm improves the segmentation accuracy of ferrograph images and ensures operational efficiency.