基于空间邻域加权的模糊C-均值聚类及其应用研究*

针对模糊C-均值聚类法用于图像聚类时仅利用了像素的灰度信息，而忽视空间位置信息，导致在噪声区域和边界处有误分类现象，提出一种新的基于空间邻域加权的模糊C-均值图像聚类法。首先，定义了一个空间邻域信息函数，该函数能够有力抑制噪声点，同时能够很好保留边界的特性；其次，设计了具有空间约束的样本邻域信息加权隶属度矩阵；最后，将该方法应用于人工合成图像和模拟MR脑图像的聚类。实验结果表明，该方法能够获得较好的聚类效果，同时具有较强的抑制噪声的能力。

Research on fuzzy C-means clustering algorithm based on spatial weighted and its application

bstract:The application of C-means algorithm to image clustering is not taking into account spatial information apart from intensity values, which will lead a misclassification on the boundaries and inhomogeneous regions with noises. This paper proposed a new image clustering method using fuzzy C-means algorithm based on the spatial weighted. Firstly, defined a spatial information function to exploit the spatial information, which was not only effective to deal with noisy, but also reserve well edge property. Secondly, it designed the neighbourhood information weighted membership matrix with spatial contraints. Finally, applied this algorithm to synthetic image and simulated MR data clustering. The experimental results show that the proposed clustering scheme is effective for noisy image.