基于分水岭和改进的模糊聚类图像分割

Image segmentation based on watershed and improved fuzzy C-means clustering

针对模糊C-均值聚类算法需预先给出初始聚类中心、未考虑邻域信息、计算复杂度高等缺点,提出了一种基于分水岭和改进的模糊聚类图像分割方法。该方法首先利用分水岭分割方法对原图像进行预分割,然后利用粒子群的全局寻优能力从预分割的小区域中搜索出较为准确的初始聚类中心;最后,在对小区域进行模糊聚类时,建立了包含邻域信息的聚类目标函数。实验表明,该方法分割速度快、抗噪能力强,实现了图像的较优分割。

In order to solve the problems in FCM （ fuzzy C-means clustering） such as the original cluster centers to be given in advance, not considering neighbor information and high complexity, this paper proposed an image segmentation based on watershed and improved FCM. Dividing image with the help of watershed algorithm, and gained the primary results. It made full use of the ability of global optimization PSO （ particle swarm optimization） to obtain the accurate original cluster centers of FCM. It had been established a novel objective function which contained neighbor information. The experimental results show that this method has higher segmentation speed and stronger anti-noise property, and it realizes significant image segmentation.