Risk assessment of on-ramp confluence area traffic based on image information

Based on introduction of image identification, the method of non-linear grid conversion is used to convert the image coordinates into road plan coordinates. Vehicle speed and vehicle distance are adopted to judge free-driving, car-following and lane-changing state of vehicle behavior. According to the body's endurance for acceleration, driving risk is divided into seven levels. Critical acceleration is used as assessment standard to evaluate driving risk in different traffic state. For entire section risk assessment, the concept of average risk is brought in and risk assessment model of on-ramp area based on image information is established. In addition, a video of a freeway on-ramp area is selected to calculate driving risk in twenty minutes, and the result shows that vehicles confluence on ramp can bring sudden change of vehicle speed and vehicle distance in main line, which will lead to car-following risk and lane-changing risk.