Real Time Multiple License Plates Localization and Distortion Normalization for Moving Vehicles

Detecting the accurate location of a license plate from a vehicle image is considered to be the most important stage of vehicle license plate recognition (VLPR) system, which greatly influences the overall recognition accuracy and processing speed of the whole system. In this paper, we propose a system to locate the multiple vehicle license plates of moving vehicles from a video traffic image sequence. We also analyze the shifting of the license plate in the image, and then develop a transformation based on relative position vector to correct the distorted plate image into a normalized standard image. By means of the distortion normalization techniques, the license plate which is initially in distortion state can also be identified more correctly. The experiment results show that our method can correctly localize the license plates even in overlapped vehicles situation and can effectively adjust the distortion license caused by the shifting of relative position between car and the camera.