A New Fast Algorithm to Rectify Tilt Image of Vehicle License Plates

In the vehicle license plate recognition system (LPR), the license plate image would inevitably be tilted to an extent. A tilt image brings difficulties for further accurate segmentation and identification of characters, eventually affects the recognition accuracy greatly. The traditional method which is based on Hough transform, in many situation, often can not accurately and promptly rectify the plates for it needs a great amount of calculation, so the overall result is unsatisfactory. In this paper, a new algorithm based on geometric region is used to collect the plates fast and accurate by rotating license plates, seeking for the minimum geometry region. The experimental results show that this method can detect the angle of tilt plates with simple calculation and high precision, which is better than the traditional algorithm based on Hough transform.