中文摘要：

针对传统FCM聚类算法在图像分割时对噪声敏感的问题，提出一种结合空间邻域信息的核FCM图像分割算法。该算法在FCM算法目标函数中增加了空间约束函数，并引入考虑邻域信息的局部隶属度函数，同时引入核函数，用内核诱导距离替换原来的欧式距离，优化分割图像的特征。最后通过将全局模糊隶属度函数与局部隶属度函数结合在一起，得到新的加权隶属度函数，实现图像的分割。通过对人工合成图像和自然图像进行分割实验，结果表明，在分割质量和效果上该算法明显优于标准FCM算法及KFCM等改进算法，同时对噪声更具鲁棒性。

英文摘要：

Aiming at the noise sensitive problem of traditional FCM clustering algorithm in image segmentation, a kernel FCM image segmentation algorithm based on spatial neighborhood information is proposed. The algorithm adds the spatial constraint function to the objective function of FCM algorithm and introduces the local membership function which considers the neighborhood information, and then the kernel function is introduced and the original Euclidean distance is replaced by the kernel-induced distance to optimize the features of the segmented image. Finally, by combining the global membership function and the local membership function, a new weighted membership function is obtained, and the image segmentation is realized. Through the segmentation experiments of synthetic images and natural images, the results show that the proposed algorithm is superior to standard FCM and KFCM algorithm in segmentation quality and effectiveness, and is more robust to noise.