基于多阶段辅助学习的可见光-红外行人重识别方法
By employing a multi-stage assisted learning strategy, utilizing grayscale histogram equalization and heterogeneous feature compensation learning modules, and combining modal similarity enhancement and distance center alignment loss, the problems of large modal differences and strong illumination sensitivity in visible light-infrared pedestrian re-identification are solved, thereby improving the accuracy and efficiency of pedestrian re-identification.
CN117576729BActive Publication Date: 2026-07-17XINJIANG UNIVERSITY
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- XINJIANG UNIVERSITY
- Filing Date
- 2023-11-27
- Publication Date
- 2026-07-17
Smart Images

Figure CN117576729B_ABST
Abstract
本发明公开了一种基于多阶段辅助学习的可见光‑红外行人重识别方法,属于计算机视觉技术领域,为了减小可见光图像和红外图像之间的异质性并降低训练成本,本发明提出一个新的多阶段的辅助学习策略来解决这个问题,通过将训练过程分为两个阶段,使模型可以逐步提取模态共享特征,并有效减轻由于模态差距过大而引起的负面影响。然后,还提出异质特征补偿学习模块,在可见光和红外特征之间进行信息补偿和融合,分别将其作为辅助分支来学习更多与跨模态相关的信息。此外,为了进一步减少模态之间的差异,提出了跨模态相似性增强模块,通过抑制干扰信息并利用像素相似性概率分布提供监督信息,以提高跨模态特征表示的一致性。最后,设计了距离中心对齐损失来优化网络,降低模态内和模态间的类内差异并增强不同类别之间的可分性。
Need to check novelty before this filing date? Find Prior Art