摘要
我国字符一般由汉字、英文字母以及阿拉伯数字组成,字符图片的类型众多给字符的识别带来了很多困难.参照目前现有人工智能算法的优点,结合了字符特征提取方法设计了一个改进的BP神经网络对归一化后的三类字符进行识别,取得了预期效果.
Chinese characters are generally composed of Chinese characters, English letters and Arabic numerals. Many types of characters in the picture have brought a lot of difficulties to the character recognition~ Referring to the advantages of artificial intelligence algorithms currently available, this paper combined the character feature extraction methods to design an improved BP neural network on three types of characters after normalization to identify and reduce the difficulty of character recognition, and achieved the expected results.
出处
《韶关学院学报》
2014年第2期20-24,共5页
Journal of Shaoguan University
基金
阜阳职业技术学院2013年校级科研项目(2013JKYXM11)
关键词
特征提取
神经网络
字符识别
feature extraction
neural networks
character recognition
