基于安卓平台的哈萨克斯拉夫文识别系统

Android platform based Kazakh Slavonic recognition system

为发展Android手机等移动终端在模式识别的应用,提出一种在Android移动终端识别哈萨克斯拉夫文的系统。对移动终端里的文档图像进行预处理,将整个文档分割成行对象和字符对象,进一步过切分粘连字符;将分割字符归一化,提取对称性特征进行粗分类;提取穿越特征和八方向码特征进行二级分类筛选;将分割字符进一步归一化,提取其81维的网格特征,进行训练和识别,输出识别结果。运用树形XML文件建立了特征库。实验结果表明,与以前提出的识别方法相比,该系统运行速度高,识别率有明显提高。

To develop the application of Android mobile phones in pattern recognition,Android platform based Kazakh Slavic optical character recognition system was proposed.Preprocessing was done to the document images in the mobile phone.Image objects were segmented into words and characters.Connected characters were segmented by over segmentation.Divided character images were normalized,and rough classification was made based on symmetric characteristic.Three different through features were extracted,based on these features,second step classification was done.The character images were normalized,81 dimension grid features were extracted.By training and classifying,satisfied recognition results were showed.A tree XML file feature set was established.Compared with the earlier proposed methods,the method is more efficient on the running speed and recognition rate.