Aiming at the problem of non-linear separability in single sample face recognition, this paper proposed a multi manifold discriminant analysis based on kernel sparse representation(KSRMMDA) algorithm. Firstly, it divided the data image into blocks and established the multi-manifold model. Secondly, it used the method of kernel sparse representation to depict the relationship among data points of manifolds, and learned the intra-manifold graphs and inter-manifold graphs. Thirdly, it found the best projections in each manifold space to maintain the characteristics of the intra-manifold graph while suppressed the characteristics of the inter-manifold graph. Finally, it calculated the distance from the test sample manifold for classification and identification. Experiments on Extended Yale B and CMU PIE datasets show that the proposed algorithm is more robust to illumination and occlusion changes than other algorithms.

[2020-10-05] Printed Article /doc/2019.04.0135/v2