基于图像金字塔的离线签名鉴别技术  

Based on Image Pyramid of Offline Signature Verification

作　　者：张淑婧 ZHANG Shu-jing(The People’s Bank of China Wulumuqi Central Sub-Branch,Urumqi 830002,China)

机构地区：[1]中国人民银行乌鲁木齐中心支行,新疆乌鲁木齐830002

出　　处：《电脑知识与技术》2020年第16期193-194,197,共3页Computer Knowledge and Technology

摘　　要：手写签名鉴别已广泛应用于日常中的方方面面,现阶段对于离线手写签名鉴别技术的研究仍在发展中。首先对手写签名图像生成图像金字塔,再对图像金字塔中的每一层图像提取局部二值模式算子,最终形成基于图像金字塔局部二值模式的多维特征。采用决策树及梯度提升树两种分类器对公开数据库CEDAR中签名进行鉴别实验,两种分类器得到AER分别为13.21%和10.89%。实验结果表明,对签名图像建立空间金字塔能够有效提高签名鉴别成功率。Handwritten signature verification has been widely used in all aspects of daily life.At present,the research on offline handwritten signature verification technology was still developing.Firstly,the image pyramid was generated from the hand-written signature image.Then,Local Binary Pattern(LBP)was extracted from each layer of the image pyramid.A multi-dimensional fea⁃ture based on the LBP of the image pyramid was formed.The decision tree(DT)and gradient boosted decision tree(GBDT)were used to identify signatures in the public database CEDAR.The AER of the two classifiers was 13.21%and 10.89%respectively.The experimental results shown that the establishment of spatial pyramid for signature image can effectively improve the success rate of signature verification.

关 键 词：签名鉴别 图像金字塔 局部二值模式 决策树 梯度提升树 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]