<p>Cloth-Changing Person Re-identification faces significant challenges in identifying individuals across different scenarios, viewpoints, and clothing variations. Existing methods mainly focus on body information extraction or clothing-irrelevant feature extraction, but are limited by single features or simple feature combinations, making it difficult to achieve higher recognition accuracy. To address this issue, we propose an innovative Tri-Stream Network with Dynamically Weighted Three-Way Decision that requires only image input for efficient identification. Our proposed method refines the cognitive granularity of recognition targets by introducing three-way decision and shadow set theory from granular computing, constructing parallel feature streams for facial features, head-limb features, global features, and a dynamic weight three-way decision module. Our dynamic weighting mechanism adaptively combines features based on their reliability in different scenarios, significantly enhancing model robustness when certain features become unreliable. Extensive experiments on multiple benchmark datasets (e.g., PRCC, Celeb-reID, VC-Clothes) demonstrate that our method significantly outperforms existing state-of-the-art techniques across all evaluation metrics.</p>