Research on Cross-Camera Trajectory Prediction Algorithm Based on Big Data Signal Representation and Graph Neural Networks
Year: 2026
Type: conference-paper
Abstract: This paper proposes an Adaptive Spatiotemporal Signal Enhancement Graph Attention Network (AST-GAT), which integrates big data signal representation and graph neural networks to construct an integrated framework of “signal encoding-dynamic mapping-spatiotemporal prediction,” solving the core challenge of cross-camera trajectory prediction. An attention-driven heterogeneous signal enhancement encoding module is designed, which improves the robustness of multi-dimensional signals and alleviates the signal attenuation problem across cameras through weight allocation and temporal differential enhancement. An innovative dynamic graph topology based on signal similarity and spatiotemporal layout is used to update the association weights of camera nodes in real time, replacing the traditional fixed topology. The graph attention mechanism is improved by incorporating temporal weights to achieve collaborative modeling of spatial association and temporal motion continuity, and a composite loss function is used to optimize model training. Experiments on the Market-1501, DukeMTMC-ReID, and self-made SBHD datasets validated that the algorithm’s average positional error is as low as 2.3 pixels, with trajectory prediction accuracy and cross-camera matching success rate reaching $89.7 \%$ and $86.3 \%$, respectively. This significantly improves performance compared to mainstream algorithms such as ST-GCN and ReID+KF, and demonstrates excellent adaptability in complex scenarios such as occlusion and sudden changes in lighting. It can provide technical support for intelligent monitoring and traffic management.
Authors: Xiaoting Yu, Xi Wang, Nannan Jiang, Xiujuan Li
Institution: Harbin Huade University
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7165668917