An Improved Vehicle-license Plate Recognition Based on Color Clues and Coding Rules

An effective vehicle-license plate detection and recognition method based on the deep learning was presented that aim to address the limitations of traditional techniques failed in dealing with images from vehicle monitoring system under complex scene. Firstly, the coordinate information of license plate is located by the YOLOv2 algorithm with color clues. Then, the Radon transformation is used to rectify the license plate region which maybe tilted. Finally, the license plate recognition results are obtained by virtue of CRNN based on license plate coding rule. Experiments show the effectiveness and efficiency of our proposed license plate detection and recognition method.