Abstract: An adaptive median filter is applied in the FCM algorithm. Firstly, adaptive weighted window is used to calculate the weighted coefficient of neighborhood pixels, adaptive weighted average image of the original image is recalculated, and the new objective function, the clustering center function and membership function are used to complete the image segmentation.Due to pixels of space information and the different influence of the neighborhood pixels to center pixel is fully considered, the distinguishments between normal data and abnormal data are given out, thus the influence of the noise data to the center pixel is reduced. Finally, the improved algorithm is applied to extraction of rail gap.The experimental results show that the improved FCM algorithm can relatively complete rail gap image segmented with noise interference, overcome the disadvantage of the FCM algorithm sensitive to noise, and improve the robustness of FCM algorithm.