基于几何特征的车牌识别系统关键技术研究*

摘要

Vehicle license is the unique symbol for vehicle identification, it plays irreplaceable role in traffic management, intelligent monitoring and other fields currently. The study has three contents. Firstly,it uses the mathematical morphology method to preprocess the license plate. Secondly,it utilizes the geometry of the plate’s rectangular structure and fixed area values to achieve the plate’s accurate positioning. Finally, according to the geometry of the large space between the plate characters, no character adhesions, and the fixed length between each character can be set threshold, to obtain accurate segmentation of the plate characters. The design based on the Matlab simulation experimental platform,to achieve the three key aspects of the license plate recognition system. Compared with precious studies,the experimental results show that the mathematical methods used in the project prominent the accuracy of license plate location and character segmentation.%车辆牌照是机动车身份识别的唯一标志，在目前的交通管理、智能监测等领域发挥着无可替代的作用。研究首先用数学形态学的方法对获取的车牌图片进行预处理，其次利用车牌区域矩形结构与面积值固定的几何特征实现车牌准确定位，最后再根据车牌内字符间间隔大，不会出现字符粘连且每个字符长度固定可设定阈值的几何特征得到车牌字符的精确分割。设计在Matlab实验仿真平台上完成，实现了车牌识别系统的三个关键环节。与之前的研究相比，实验结果表明所用的数学方法突出车牌定位的准确性和字符分割的精确性。