Person re-identification (ReID) focuses on retrieving pedestrians in scenarios where camera views do not overlap. In order to alleviate the interference of factors such as pedestrian postures, obstruction by obstacles, and the lack of a frontal appearance, this paper proposes a Global-local Feature Mutual Enhancement for Person ReID with View Mixup. Firstly, we propose a view-mixup adversarial learning mechanism, which enhances the discrimination ability of the global and local view discriminator by view mixup, thus the network can learn view-robust pedestrian discriminative features. On this basis, a global-local feature mutual enhancement training mechanism is designed to make the model more fully extract view-invariant features from both local and global perspectives under the joint operation of dual discriminators. The proposed method has been validated on DukeMTMC and Market1501 datasets, demonstrating its effectiveness and superior performance.