Research Of License Plate Location Based On Subtraction Images And Camera Calibration Method

The automatic recognition of vehicle license plate is an important step in Intelligent Transportation System. It is mainly applied in freeway charging without parking, management of parking lot, and etc. Even though the LPR has been developed for a long time, the complexity of the manage environment makes the LPR a system which can not get ideal result. The location of vehicle license plate is the most pivotal technology in the recognition system, which plays an important role in the performance of whole system. This thesis focuses on the key problem, and proposes the new license plate locating algorithm based on the subtraction images and the camera calibration after studying and analyzing the existing research fruits.The license plate location algorithm of this thesis is not that it picks up the license directly from the image. And it gets the license from the vehicle area .The algorithm is more complex, but in the complexity environment, because of the outer disturbing, such as moving human, billboard and etc., picking up the vehicle area first will improve the accuracy effectly. The developed subtraction images is raised in the thesis fully considering the influence of weather change and camera dither. It makes the picture more suitable for the license picking-up environment through many algorithms, such as mathematical morphology, connect area following etc.Aiming at the segmented car image, the thesis proposed the position method of using multiplex color information template. The method locates the license area in the vehicle area. It takes the advantages of gray and color location, which can improve the accuracy of license location. The license color can be obtained by the effective of location method.In the follow-up tilt correction algorithm, fully considering the affection of plate imaging caused by the camera relative position, studying the imaging theory, and considering the apriori knowledge of camera calibration, the thesis brings forward the tilt correction algorithm of reducing tilt angle checking range based on the camera calibration parameter. The algorithm can reduce the checking time, and avoid the disturbing of non-aim area effect.The performance of the method was investigated on many real vehicle images under the complex conditions with various illuminations and weather conditions. The experimental results show that the proposed algorithms have high accuracy and robustness.