基于相似类合并FCM在图像分割中的应用

Similar Class Merging Based FCM for Image Segmentation

传统的基于模糊C均值聚类的图像分割算法分割结果中类内数据空间分布离散,无法准确分割出目标物体.针对这一问题,提出一种基于相似类合并模糊C均值聚类算法,并将其应用到图像分割中.首先,提出一种全局空间相似性度量标准和全局灰度相似性度量标准,并将其引入到一种新颖的节点间距离度量公式中来计算图像中任意一点与聚类中心点的差异.其次,算法选取彩色直方图作为区域描述算子,采用巴氏距离计算聚类过程中得到的任意两类间的相似性.最后,应用最大相似类合并策略得到最终的分割结果.实验结果表明,与传统模糊C均值聚类算法和空间约束核模糊C均值聚类算法相比,该算法获得更加精确的图像分割结果.

A similar class merging based FCM algorithm for image segmentation was proposed tosolve the problems that the segmentation results of the traditional FCM based image segmentation algorithm are discrete in the spatial distribution and the object cannot be segmented accurately by the traditional FCM based method. Firstly, a global spatial similarity measure and a global intensity similarity measure were proposed and introduced into a novel distance metric to calculate the difference between the pixels and the cluster centers. Secondly, color histogram was used as a descriptor, and Bhattacharyya distance was used to calculate the similarity between any two classes. Finally, a maximal similarity based class merging strategy was used to obtain the final image segmentation results. The experimental results indicated that the proposed algorithm can obtain more accurate image segmentation results compared with FCM and KFCM methods.