Visible-infrared person re-identification aims to achieve individual retrieval in both daytime and nighttime, which is of great significance for the implementation of intelligent pedestrian recognition systems. This paper proposed a Progressive Feature Interaction and Enhancement Network (PFIENet) to enhance cross-modality feature extraction capabilities. We propose a Feature Interaction Module (FIM) to achieve different perceptual feature interactions at different stages of feature extraction. Secondly, we propose a Progressive Dependency Enhancement Module (PDEM) that enhances correlated features by progressively capturing the dependency relationships of cross modal features. Additionally, we introduce the intra-identity multi-constraint center loss (IMCL) to further narrow intra-modal differences by leveraging mutual constraints between different perceptual features. Finally, the proposed method achieves 67.94% mAP and 71.37% Rank-1 accuracy on SYSU-MM01.