Traffic conflict automatic discrimination at non-signalized intersection based on video processing

For the defects of traditional artificial recognition and record method of traffic conflict,traffic conflict automatic discrimination at non-signalized intersection based on video processing is presented in this paper.Firstly,foreground moving objects in the image and their spatial and temporal trajectories are extracted by motion detection and motion tracking;and then,the motion behaviors are obtained based on objects types and moving data.Traffic conflict can be discriminated in terms of two traffic conflict discrimination rules which are concluded combined with classic traffic conflict definition.Based on video processing captured from non-signalized intersection,the method is validated that it can be well applied in real-world.