A new Fuzzy C-Means algorithm based on multi-feature fusion FMFCM was proposed in this paper, aiming at the problem of poor anti-noise performance and low segmentation accuracy of Fuzzy C-Mean Clustering (FCM) algorithm. The brain tumor MR image was taken by the algorithm as the experimental object, extracted the gray value, spatial information and local binary pattern of the brain image as multi-dimensional features to segment, and tested the anti-noise performance of the image by adding different intensity of Gaussian noise and salt and pepper noise. Experiments show that compared with FLICM algorithm FRFCM and FCM-S2 algorithm, The proposed algorithm improves the partition coefficient (Vpc) by 6.9%, 10.3% and 13.2% respectively, and reduces the partition entropy (Vpe) by 16.2%, 11.8% and 15.2%, respectively. The segmentation accuracy is improved by 2.2%, 1.6% and 3% respectively. The proposed algorithm has better robustness to noise and can achieve better segmentation results.