A New Method of Traffic Conflict Data Collection and Processing for Traffic Safety Evaluation

摘要

At present, the most common method of collecting traffic conflict data was the site observation method. Data processing methods were mainly manually observed by site or video. To overcome the shortcomings of traditional data collection and processing methods, video recognition was collected by UAV (Unmanned Aerial Vehicle) flying at high altitude and conflict were automatically detect by video recognition software in this paper. Compared with traditional data collection methods, UAV has a good high-altitude field of view, does not have shooting angle and occlusion problems, and has a large shooting range, which has obvious advantages. The video recognition program consists of the following parts: image reading and calibration, vehicle identification, vehicle tracking. Video recognition got continuous high-precision vehicle trajectory data, subsequent processing to obtain continuous vehicle conflict data. Finally, we can use these data to evaluate road traffic safety. The results show that this method can effectively evaluate road traffic safety.