摘要
在核模糊聚类算法(KFCM)的基础上,提出了一种新的PSO_KFCM聚类算法。新算法利用高斯核函数,把输入空间的样本映射到高维特征空间,利用微粒群算法的全局搜索、快速收敛的特点,代替KFCM算法逐次迭代的过程,在特征空间中进行聚类,克服了KFCM对初始值和噪声数据敏感、易陷入局部最优的缺点。通过对医学图像进行分割,仿真实验结果表明,新算法在性能上比KFCM聚类算法有较大改进,具有更好的聚类效果,且算法能够很快地收敛。
A novel kernel fuzzy C_ means clustering algorithm which uses the merits of the global optimizing and higher convergent speed of particle swarm optimization （PSO） algorithm and combines with kernel fuzzy C_ Means （KFCM） is proposed with application to medical image segmentation. The algorithm eliminates FCM trapped local optimum, being sensitive to initial data and the noise data. The performance of this modified KFCM is compared with KFCM. Numerical results of this comparative study are performed on medical images segmentation.
出处
《计算机工程与设计》
CSCD
北大核心
2008年第9期2295-2296,2299,共3页
Computer Engineering and Design
关键词
微粒群算法
核函数
图像分割
模糊C_均值聚类
特征空间
particle swarm optimization
kernel function
image segmentation
kernel fuzzy C means clustering
feature space
