Abstract: To address the problem of low image recognition accuracy in facial recognition software, a fuzzy face image recognition algorithm based on the improved Grassmannian manifold is proposed. Under the conditions of zero noise and the maximum size of the fuzzy basis being known, the subspace (obtained from images and complete orthogonal basis convolution, where the maximum size has been preset) is constructed, and the subspaces obtained from clear images and fuzzy images are demonstrated to be the same. Flurry image recognition by adopting this subspace presentation is tested by taking the constructed subspace as a point in the Grassmannian manifold. Moreover, the flurry image recognition problem under the condition that the fuzzy quantity exhibits both homogeneity and mutability is studied. The simulation results show that the method proposed in this paper will improve facial recognition accuracy and achieve a high recognition rate.