Answers are grounded in this paper and cite their sources.
Computational modeling reveals improved visual coherence and multi-scale stability in digital textile patterns, highlighting reliable feature transmission for intelligent visual communication.
With the rapid development of digital media and intelligent information transmission technologies, maintaining semantic consistency and stable multi-scale feature propagation has become increasingly important for visual communication systems and data-driven engineering applications, including future electromagnetic sensing and communication scenarios. However, textile pattern communication often suffers from fragmented information representation and disrupted transmission pathways, resulting in reduced visual coherence and communication effectiveness. To address these issues, this study proposes a visual communication optimization framework based on Swin Transformer (Swin-T) by integrating a Multi-Scale Attention (MSA) module and a path regularization strategy. The MSA module reorganizes hierarchical feature representations to achieve semantic alignment across different scales, while the path regularization constraint stabilizes similarity distributions during cross-layer propagation and suppresses feature dispersion. Experimental results demonstrate that the proposed framework achieves a maximum cosine similarity of 0.918 and a minimum Jensen–Shannon divergence of 0.183 during cross-scale transfer, while reaching an SSIM of 0.911 in geometric pattern scenarios and 0.871 in complex composite scenarios. The collaborative optimization strategy significantly improves visual consistency, information aggregation, attention concentration, and cross-scale transmission stability compared with existing approaches. The proposed framework provides an effective solution for intelligent textile pattern communication and offers valuable insights for multi-scale information representation and reliable feature propagation in future electromagnetic information processing and communication-oriented engineering systems.
No takes yet. Share an insight, caveat, or question.
X. L. Zhang (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: