一种车牌识别软件系统研究  

Study of License plate recognition software system

作　　者：楚岩[1] 邵严[1] 陈亮[1] 巨永锋[1] 

机构地区：[1]长安大学电子与控制工程学院,陕西西安710064

出　　处：《电子设计工程》2013年第4期149-151,155,共4页Electronic Design Engineering

摘　　要：针对现有车牌识别算法中的车牌定位、字符分割和字符识别三个核心模块存在的不足,提出了一种基于边缘两侧颜色检测的车牌定位方法;通过采用动态阈值调整方法,很好的实现了字符分割;对神经网络在字符识别技术中的应用进行了大量的研究和实验,根据汉字和数字、字母特征提取的不同,在对字符信息初识别时将汉字和数字、字母采用不同结构参数设置的神经网络进行识别,并对识别结果中包含的具有形体相近的字符提出了一种"不等权值"的方法。结果证明识别率有了明显提高。For the shortcomings of the existing license plate recognition algorithm in three core modules of license plate location, character segmentation and character recognition, proposing a license plate location method based on the edge of both sides of the color detection; The method of dynamic threshold adjustment achieved the character segmentation well; Much research and experiment for neural network applications in character recognition have been done. According to the different of feature extraction of characters, numbers and letters, the different network parameters are set to recognition. For similar character of recognition results, proposing a method of differ-weights. The results show that the recognition rate has been markedly improved.

关 键 词：车牌识别 边缘检测 动态阈值调整 神经网络 不等权值 

分 类 号：TP31[自动化与计算机技术—计算机软件与理论]