Standard fuzzy c-means (FCM) algorithm does not consider spatial neighborhood information in the process of clustering, which makes it sensitive to noise and cannot achieve satisfying results for image segmentation. In order to reduce the noise effect during segmentation, we present an improved anti-noise FCM algorithm by incorporating a neighborhood term into the standard FCM algorithm. The neighborhood term combines the relative location information and gray level information of the neighboring pixels. It can control the influence of the neighboring pixels on its central pixel effectively and enhance the robustness to noise. Experiments on images with different noisy levels demonstrate that the proposed algorithm is effective and more robust to noise than standard FCM and some other extended FCM algorithms.