朝鲜语联机手写识别建模研究

Research of Korean Online Handwriting Recognition's Modeling

为了解决朝鲜语(韩语)手写识别,对韩语文字提取独特几何特征定义方式。提出六组基本元定义,用它们区分韩语的各个字母。使用二次编码矫正,通过实验,使得重码率降到极低。能够较为准确的识别韩语中各种风格的手写正楷与部分行书。对于韩语草书的识别,是否成功,取决于建立大而全的草书字典。这一独特方法,也可以移植到韩语印刷体识别。

In order to solve the Korean handwriting recognition, extract the unique geometry definition according to the characteristics of Korean characters. Put forward six basic definitions; use them to distinguish between each letter in Korean. Using secondary coding correction, make the rate of repeated signs down to very low level through the experiment. It can accurately identify the Korean with various styles of handwriting in block letters and partly of the running script. For Korean cursive script identification, the success depends on establishing large cursive dictionary. This unique method can also be transplanted to Korean printed letter＇s identification.