一种联机手写哈萨克文字母识别算法  

A RECOGNITION ALGORITHM FOR HANDWRITTEN KAZAK LETTERS USED ON MOBILE PHONE

作　　者：袁保社[1] 

机构地区：[1]新疆大学信息科学与工程学院,新疆乌鲁木齐830046

出　　处：《计算机应用与软件》2010年第9期108-110,121,共4页Computer Applications and Software

基　　金：新疆维吾尔自治区科技攻关重大专项(200732143-2)

摘　　要：在手机中实现哈萨克文手写输入是新疆地区哈萨克族用户的需求。基于手机手写屏尺寸的局限性,哈萨克文在手机上以字母为单位输入。在研究了哈萨克文字母结构特征基础上,提出将哈萨克文字符分成主体笔画与附属笔画两部份。利用字符的笔画数做第一级分类,一笔以上的字符按附件二级分类。提取字符的笔画落笔抬笔象限、笔画x方向最大穿越次数、笔画环、笔画x、y方向变化次数及落笔、抬笔走势等20个特征为特征集。通过采集当前手写输入字符的xy坐标序列,提取相应的特征组成用二进制表示的特征数据,与样本库中的样本逐个异或运算比对,取运算后1的个数最少的样本为识别样本,经进一步检验后得到识别的字符及相近的侯选字符。It is the demand of Kazak ethnic users in Xinjiang Autonomous Region to realize Kazak handwriting input on mobile phone. Lim- ited in the size of the touchscreen on mobile phone,the input of Kazak on mobile phone is through its letter. Based on the research on Kazak letters＇ structural features,we put forward that to separate the Kazak characters to main strokes and affiliated strokes. The letter＇ s stroke num- ber of a character is considered as the first classification,the characters with two or more strokes are of the second classification according to their appendages. 20 features of the characters strokes are extracted to form the feature set,including strokes quadrants of penpoint pressing and penpoint lifting,maximum traversing times of stroke on x direction,stroke ring, number of change times of strokes on x and y direction, penpoint lifting trend, etc. By collecting current handwritten character＇s x and y coordinate sequence to extract corresponding characteristic and compose the characteristic data expressed with binary system,then these data are made XOR operation and alignment one by one with the sam- ples in sample library, the operated sample with least number of 1 is to be granted as the recognition sample, the recognized character and close candidate character are obtained after further verifications.

关 键 词：哈萨克文 手机 特征编码 手写输入 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]