高级检索

    中国印刷体汉字的识别

    Recognition of Printed Chinese Characters

    • 摘要: 本文提出了一种印刷体汉字计算机自动识别的方案.从EIT-PS简易图文扫描器输入一页印刷体汉字,经前处理程序切割出每一个字并将其正规化;利用网格单元特征进出该输入文字的候补文字;再计算输入文字与候补文字的复合类似度,复合类似度最高的候补文字被认为是输入文字.本识别方法具有抗噪声干扰,抗连笔、断笔等变形的能力.对任选的一万字进行识别试验的结果,取得了积累识别率98.6%的良好成绩.

       

      Abstract: An automatic method for recognizing printed Chinese characters is proposed in this paper,Every Chinese character on a page read through the EIT-PS scanner is cut out with the pre-processing program and then normalized,The candidates for these Chinese characters are selected by their background features.After that,the analogy degree of inputted Chinese characters to their candidates is computed.Those candidates having the highest analogy degree will be identified as the Chinese characters inputted. The method given in this paper has the ability of high noise immunity and ability of anti-distortion caused by stroke linking and stroke breaking.An experimental test has been conducted to recognize 10000 Chinese characters selected in a random way.The result shows that the accumulative recognizing rate reaches 98.6%.

       

    /

    返回文章
    返回