A License Plate Tilt Correction Algorithm Based on the Character Median Line Algorithme de correction d’inclinaison de plaque d’immatriculation basé sur la ligne médiane du caractère

Abstract

License plates intelligent identification systems must be able to correct the tilt of a license plate in an image. Aiming at improving on the low tilt accuracy, complex algorithms, and weak robustness against noise of existing tilt correction methods, we proposed an algorithm based on the character median line. The license plate image is first preprocessed, and a projection method is applied to find and segment the character region, resulting in a license plate with no border. For the no border license plate image, we then fix x-coordinates, and find the maximum and minimum values of y-coordinates, and put them into a matrix. The next step is to obtain the mean value of the maximum and minimum values of y, obtain the point sets on the character median line of the license plate, and remove the singular points using a threshold. Finally, a straight line is fitted using the least-squares method, and the tilt angle is obtained by applying a formula for the slope and the angle. For a tilted and damaged license plate, experiments show that the proposed algorithm is simple, has a low error ratio, and has good robustness against noise and deformation.