| Summary |
In order to effectively control the characteristics of locations with high accident risk, this Study uses Traffic Collision Analysis Technology, combined with emerging technologies of Unmanned Aerial Vehicle (UAV) aerial photography and image analysis to automatically identify pedestrian and vehicle collision accidents and hot zones to establish diagnostic analysis tools for high traffic accident risk locations. The analysis results can be used by the Traffic Administration Agencies and Road Designers to evaluate the collisions or risks at the intersections in order to diagnose traffic collision problem before the traffic accidents occur, and serve as the reference information for the development of preventive traffic improvements to reduce the number of traffic accidents and severity.
This Study continues the results of "Development and Application of Road Traffic Flow and Accident Risk Detection Analysis Tools" (hereinafter referred to as the Previous Study) conducted in 2017, aiming at the improvement of software architecture and analysis methods focusing on two tasks. The first task uses the UAV loaded high-resolution digital camera to capture the images of pedestrians and vehicles flow at selected intersections with high accident risk locations, and uses the computer image analysis technology to compute the vehicle flow trajectory information including the vehicle location, vehicle type, traffic volume and vehicle speed, and improve the problems of insufficient sampling for large vehicles, tracking of motorcycle easy to be interfered by the ground markings; as well as the identification of pedestrian and bicycle limited by image pixels and insufficient feature extractions, improve the bottlenecks and issues in previous Study, and improve vehicle identification and tracking works, to improve the accuracy of the vehicle trajectory. The second task is to take the domestic mixed traffic situation into consideration, to plan and analyze the specific functional appearance of the software, uses the vehicle trajectory data to compute the traffic collision accidents, and produces collision hot zone analysis functions, to subsequently deliver to the Traffic Engineers to derive corresponding traffic engineering measures for improvement according to the analysis results of collision type and location, to improve the traffic safety. The traffic flow trajectories obtained from the image analysis can be used to generate traffic flow characteristics of traffic flow speed and turning traffic volume of different types of vehicles, which can be used for traffic analysis or traffic signal system design.
This Study establishes Unmanned Aerial Vehicle (UAV) Traffic Videography technology and tools to collect and identify traffic information and traffic flow route data for subsequent traffic management and safety analysis, and develops automated road traffic safety diagnostic technology to mitigate the cause analysis workload of improving the road section easy to cause accident, to assist the Traffic Administration Agencies and Road Designers as the reference for preventive traffic safety improvements. |
