Abstract: Most existing person re-identification work focuses on either extracting discriminative features or learning discriminative distance metrics.However,in the complex real-world scenarios,large variation of resolutions between the same person image observed in different cameras is existed,and the person image in the same camera also undergoes large variation of view and illumination,these factors limit the representation ability and robustness performance of extracted features.To address this problem,we propose a cross-view multi-level discriminative dictionary learning method by utilizing the intrinsic correlation of coding coefficients in the feature representation of different views in this paper.First,a feature mapping function is introduced in the dictionary learning model of the image horizontal region-level and image-level to bridge the gaps of the cross-view image.Through the mapping function,the stringent correspondence relation is relaxed between the cross-view images,thereby leaving the coding coefficients more flexibility to maximize the feature representation performance.Then,on the patch-level,we incorporate a local geometry constraint on atoms into the dictionary learning objective function by considering the local manifold structure of the image patch.By learning a graph Laplace matrix adaptively,the local geometry structure of training samples can be mapped to the coding coefficients.Therefore,a more discriminative dictionary pairs can be obtained.Experiments on the two challenging person re-identification datasets demonstrate the proposed method can reduce the influence of large variation of resolution in the different cameras and improve the representative and discriminative abilities of learned dictionaries.The effectiveness of the proposed approach is validated on the VIPeR dataset and the CUHK01 dataset,and the best rank-1 matching rate are reached 68.40%,80.14%respectively.Compared with the state-of-the-art algorithms,the proposed method can improve the performance of person re-identification.