基于密度特征的维吾尔文离线签名识别  

Off-line Uyghur signature recognition technology based on density feature

作　　者：库尔班.吾布力 热依买.阿不力克木 努尔毕亚.亚地卡尔 阿力木江.艾沙[2] 吐尔根.依布拉音 

机构地区：[1]新疆大学信息科学工程学院,新疆乌鲁木齐830046 [2]新疆大学网络与信息中心,新疆乌鲁木齐830046

出　　处：《计算机工程与设计》2016年第8期2200-2205,共6页Computer Engineering and Design

基　　金：国家自然科学基金项目(61163028;61363064;61563052);新疆维吾尔自治区高校科研计划重点基金项目(XJEDU2013I11);模式识别国家重点实验室2014年度开放课题基金项目(201306321)

摘　　要：对维吾尔文签名图像进行预处理(含灰度化、平滑、二值化、归一化、细化等)的基础上,提出基于密度特征的维吾尔文离线签名识别方法,结合签名样本的空间地位信息,对离线签名进行有效处理和识别。从维吾尔文签名样本库中选取75个签名者的(20个样本/人)1500个签名样本进行分类实验,得到最高为96%的识别率,实验结果表明,密度特征是维吾尔文手写签名的一种有效特征,能较全面地描述与捕捉维吾尔文手写签名的书写特点。On the basis of preprocessing Uyghur signatures through gray scaling,noise removing,binarization,normalization and thinning etc.,a type of signature recognition method based on density features was proposed,and Uyghur signatures were effectively identified by combining the position information of signature samples.The classification experiments were carried out using 1500 Uyghur signatures selected from 75persons（20samples/person）in Uyghur handwritten signature database.96% of highest identification rates were obtained.Experimental results show that the density feature is a kind of effective feature for Uyghur handwritten signature,and it can comprehensively describe and capture the writing characteristics of Uyghur handwritten signature.

关 键 词：维吾尔文 签名识别 密度特征 特征距离 向量间距离 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]