Automated recognition of drunk driving on highways from video sequences

A new method for the detection of abnormal vehicle trajectories is proposed. It couples optical flow extraction of vehicle velocities with a neural network classifier. Abnormal trajectories are indicative of drunk or sleepy drivers. A single feature of the vehicle, e.g., a tail light, is isolated and the optical flow computed only around this feature rather than at each pixel in the image. The velocity fields are accurately extracted using a modification of the basic optical flow method (Horn and Schunck, 1981, and Barron et al., 1994). Trajectories are extracted in the form of direction of motion in each frame. A back-propagation neural network is used to classify the vehicle trajectories as either normal or abnormal. The neural network is shown to perform accurate classification on simulated trajectories