Abstract: Aiming at the segmentation effect of traditional fuzzy c-means algorithm on the image under noise interference, an improved algorithm combining the kernel space and adaptive median is proposed. The algorithm used adaptive median filter to obtain the local spatial information of the pixel, and then generates a new fuzzy factor to add to the objective function. Finally, the objective function is optimized and solved in kernel space to obtain the optimal cluster center and membership degree. Experimental results show that the algorithm has high accuracy and robustness to images polluted by high salt and pepper noise.