基于字符连通域特征的车牌定位方法  

License Plate Localization Algorithm Based on Features of Character Connected Component

作　　者：张金林[1] 胡宏健[1] 方虎生[1] 孙磊[1] 

机构地区：[1]中国人民解放军理工大学工程兵工程学院,江苏南京210007

出　　处：《电视技术》2012年第7期140-142,共3页Video Engineering

摘　　要：车牌定位是整个车牌识别系统的关键技术,传统方法普遍存在实时性差、适应性不强等问题。强调硬件系统构架,固定部分图像输入环境,引入大量先验知识,并最终利用车牌字符连通域的显著特征,将车牌定位与车牌字符的显著性特征相联系,缩短了车牌识别的流程,实现了可适应性强、定位准确的车牌定位。实验结果证实,该算法准确率高,效果明显且具有较好的稳健性。license plate location is the key technology in license plate recognition system. Most prevalent traditional methods are poor of real-time arid flexible. Hardware system architecture are emphasised on. The part of input environment and priori knowledge are fixed. The use of the salient features of the license plate characters are focused on. At last, the license plate license plate location is connected with the features of the characters. It shortens the process of license plate recognition system, strengthens the adaptability of the license plate location. The experimental proves that the effect of this algorithm is high accurate and obvious.

关 键 词：车牌定位 连通域 二值化 全局阈值 行扫描 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]