摘要
文章提出了一种基于整体特征的小写英文字母识别方法。首先根据字母图像的赋值背景提取其整体特征,然后构建7个模板进行模板匹配。该方法不需要对图像作复杂的细化处理、轮廓提取等,减少了可能带来的误识和拒识,也不需要现有神经网络方法的长期训练,因而简单快速。同时,不同字体的字母图像其整体特性基本相同,因此识别率较高。
This paper presents a fast lowercase english letter recognition based on global feature. First the global feature of letter image is extracted through its value associated background counting, then it is template matched by seven templates. This method needs no complex thinning procedure, feature analysis which reduce the possibility of error and rejection, and it also no needs long-time train procedure such as neural network, so it is simple and fast. At the same time, since global feature of different typeface is the same basically, the method has high recognition rate as well.
出处
《计算机与数字工程》
2012年第9期117-118,136,共3页
Computer & Digital Engineering
关键词
整体特征
赋值背景
神经网络
global feature
value-associated background
neural network
