Vehicle Speed Estimation from Single Still Images Based on Motion Blur Analysis

Motion blur arises when the object motion is fast relative to the shutter time of the camera. In this work, we p ropose a method to estimate the speed of moving vehicles from sing le still images based on motion blur analy sis for the p urpose of traffi c law enforcement. T he blur parameters are estimated from a sing le motion blurred image and then used to calculate the speed of the moving vehicle in a scene. F irst, the motion d irection of the vehicle is estimated using image grad ient information and the orientation of the motion blur is simp lifi ed to a one-d imensional case by image rotation and rectifi cation. T he length of the motion blur is then estimated by ed ge d etection and blurred ed ge analy sis. F inally , the speed of the moving vehicle is calculated accord ing to the imaging geometry , camera parameters, and the estimated blur parameters. E xperimental results are p resented for both synthetic and real images.