摘要
提出了一种汽车牌照的自动识别算法：先对汽车图象进行直方图均衡、平滑去噪、二值化预处理，再根据车牌上文字变化特点快速、准确地从复杂背景中分割车牌，最后采用投影－变换系数法提取汉字、英文和阿拉伯数字的不同维数的特征，送到相应的ＢＰ神经网络中进行识别，较好地解决了汽车牌照的自动识别问题。
Automatic recognition algorithm of a car license plate (CLP) is proposed.First the image with a car must be preprocessed.Then based on the steady of the characters on the CLP,a new method to locate CLP in a complex background is presented accurately and rapidly.Finally,the project changing method is used to extract vectors of different dimension for Chinese character, English and Arab alphabet.The recognition result by BP network shows that owing to the high speed and strong recognition ability of BP network, the automatic recognition problem of the CLP is solved very well.
出处
《南京航空航天大学学报》
EI
CAS
CSCD
北大核心
1998年第5期573-577,共5页
Journal of Nanjing University of Aeronautics & Astronautics
关键词
特征识别
模式识别
车牌定位
车牌识别
character recognition
pattern recognition
CLP location
CLP recognition
BP network
