遗传模糊C-均值聚类算法应用于MRI分割

Genetic Fuzzy C-Means Clustering Algorithm for Magnetic Resonance Images Segmentation

通过分析知经典的将图像分割成C类的常用的模糊C-均值聚类算法(FCMA)依赖于初始聚类中心的选择,通常得到的是局部最优解而并非全局最优解,又由于遗传算法能搜索到全局最优解,因此将遗传算法(GA)与FCMA相结合,对MRI直接进行聚类,利用遗传算法搜索全局最优解,从而有效地避免了模糊C-均值聚类算法收敛到局部最优的问题,并在此基础上实现了对MRI的分割,得到了比较满意的效果。

Based on the class fuzzy C-means clustering algorithm （FCMA） is a well-known clustering method to partition an image into homogeneous region.We know FCMA is dependent on the choice of the initial distribution of cluster center, and consequently the algorithm ends up in a local optimum. Because of the genetic algorithm which can achieve the global optimum, we directly unified them in the magnetic resonance images （MRI） segmentation. By applying genetic algorithm, we can achieve the global optimum in MRI segmentation application.