基于空间邻域信息的二维模糊聚类图像分割

传统模糊C均值聚类(FCM)算法进行图像分割时仅利用了像素的灰度信息,并且使用对噪声较敏感的欧氏距离作为像素与聚类中心距离度量的标准,因此抗噪性能较差.为了克服传统FCM算法的局限性,本文提出了一种基于空间邻域信息的二维模糊聚类图像分割方法(2DFCM).该方法利用二维直方图描述的像素邻域关系属性,一方面为聚类提供较准确的初始聚类中心,从而避免聚类中的死点问题;另一方面通过提出聚类中心同时在像素值、像素邻域值二维方向上进行更新的思想,建立了包含邻域信息的新的聚类目标函数,实现了图像的分割.实验结果表明,这种方法抗噪能力强、收敛速度快,是一种有效的模糊聚类图像分割方法.

Image segmentation with two-dimension fuzzy cluster method based on spatial information

With only pixel value information taken into account and non-robust Euclidean distance used as the distance measure standard, the classical Fuzzy C-means Clustering (FCM) algorithm lacks enough robustness in the image segmentation. In order to overcome the limitation of FCM, a novel Two-dimension Fuzzy Cluster Method (2DFCM) was proposed based on the spatial information. Here two-dimensional histograms were used for two purposes. First, more accurate original cluster centers were acquired, which could avoid poor clustering results caused by wrong original cluster centers. Second, a new idea was presented to update the cluster centers in pixel value and pixel neighboring value simultaneously, from which new objective functions were derived to realize the image segmentation. It is shown from the experiments that our proposed algorithm is more robust and faster in convergence.