An intelligent traffic coordination control method and system based on image recognition and deep learning
By constructing dense optical flow fields, encoding behavioral states, and quantizing interaction graphs, the problems of refined and dynamic adaptation of traffic data perception and conflict prediction in intelligent traffic coordinated control are solved, enabling accurate characterization and efficient control of traffic flow states and participant behaviors.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- HEFEI UNIV OF ECONOMICS
- Filing Date
- 2026-05-09
- Publication Date
- 2026-07-17
AI Technical Summary
Existing intelligent traffic coordination and control methods lack refined processing in the traffic data perception stage, failing to accurately capture the pixel-level movement trajectories and dynamic trends of traffic participants. This results in traffic flow state perception remaining at a macro level, making it difficult to reflect individual behavioral differences. Furthermore, conflict prediction and signal control optimization lack systematicity and foresight, and signal control strategies fail to dynamically adapt to real-time traffic bottlenecks, leading to low traffic efficiency.
By constructing dense optical flow fields, encoding behavioral states, constructing and quantifying interaction graphs, predicting conflict situations, and optimizing signal control, pixel-level refined perception and deep interaction logic capture of traffic participants are achieved, and traffic light phase schemes and timing parameters are dynamically adjusted.
It enables precise characterization of traffic flow status and participant behavior, improves the accuracy and systematic nature of conflict prediction, enhances traffic efficiency and safety, and reduces the probability of potential conflicts.
Smart Images
Figure CN122416765A_ABST