基于特征空间法的旋转多字体文字识别

The Rrecognition of Rotated Multi-font Character Based on Eigen Space

针对计算机识别旋转多字体文字困难问题,提出基于特征空间法的文字识别方法.首先,提取文字特征,建立文字的特征空间,确定文字的运动轨迹;其次,通过空间距离比较待测文字特征与轨迹的相似度,识别出文字及其对应的旋转角度.最后,分别以单一字体旋转字符以及四种字体字符为例分别进行了实验,发现62个字符中54个字符识别率达100%,旋转角度精度在10o以内的字符达94%.实验表明该方法具有较高识别率.

For the problem that it is difficult for computer to recognize rotated multi-font character,a method of character recognition based on the Eigen space approach is proposed.The character features are extracted,the character feature space is established,and then the trajectories of character is determined.The characters and the corresponding rotation angle is identified by comparing similarity of characteristics of the character under test and the trajectory though space distances.Finally,experiments have been performed taking a single font characters and multi-font characters as examples.It is found that54 characters among 62 characters can get 100%recognition rate and the accuracy of rotation angle can get 94%.Experiments indicate that the proposed method has a high recognition rate.