摘要
针对车牌识别算法在手机上应用的需求越来越迫切的问题,提出一种基于OpenCV的车牌号识别方法,并在Android平台下实现.该方法由两部分构成:其一为单一的车牌图像识别,分为图像获取、图像预处理、车牌定位、字符分割和字符识别等过程;其二为自动筛选识别结果,即连续拍摄图像并识别,通过其结果的一致性和匹配度自动选择最佳结果.经实测和分析可知:该方法在主流手机上处理速度可达到20帧/ms;影响识别率的主要因素是车牌污损和拍摄角度偏差;在实际环境中识别率可达90%以上.研究结果表明,该方法具有较高的实时性和识别率,可适用于在运动过程中进行车牌识别的场景.
Aiming at the increasingly urgent problem of the application of license plate recognition algorithm in mobile phones,a license plate recognition method based on OpenCV is proposed and implemented on Android platform.This method consists of two parts:one is license plate image recognition only,which can be divided into image acquisition,image preprocessing,license plate location,character segmentation and character recognition.The second is the automatic filtering of recognition results,that is,continuous shooting images and recognition,the best result is automatically selected by the consistency and matching degree of the results.The measured and analyzed results show that the processing speed of this method on mainstream mobile phones can reach 20 ms per frame,the main factors affecting the recognition rate are license plate fouling and shooting angle deviation,and the recognition rate is above 90%in real environment.The results show that this method has higher real-time performance and recognition rate,which can be applied to the scene of license plate recognition in the motion process.
作者
张军平
徐衍胜
ZHANG Jun-ping;XU Yan-sheng(Institute of Research,Lanzhou Jiaotong University,Lanzhou 730070,China;Shanxi Junyue Maike Information Security Technology Co.,Ltd.,Taiyuan 030006,China)
出处
《兰州交通大学学报》
CAS
2021年第3期57-62,共6页
Journal of Lanzhou Jiaotong University
