The majority of road accidents occur as a result of driver irresponsibility. Utilizing deep learning models to analyze traffic CCTV can effectively mitigate traffic accidents. This paper aims to identify and monitor vehicles and assess traffic conditions by comparing and categorizing the similarities between vehicle trajectories. Three similarity measurement methods are employed: cosine similarity, Jensen-Shannon divergence, and Euclidean distance similarity. The results show that lanes and traffic patterns can be effectively identified using the presented approach. Through our approach, one can further develop traffic monitoring and analysis advancements, thereby enhancing road safety and traffic management
Real-Time Traffic Analysis Using Vehicle Trajectory Similarity in Edge Computing
2025-02-18
5215310 byte
Conference paper
Electronic Resource
English
Self‐Similarity of Real Time Traffic
Wiley | 2007
|Advance Real-time Detection of Traffic Incidents in Highways using Vehicle Trajectory Data
ArXiv | 2024
|Vehicle trajectory calculation method based on space-time similarity
European Patent Office | 2021
|Real-Time Vehicle Trajectory Prediction for Traffic Conflict Detection at Unsignalized Intersections
DOAJ | 2021
|