基于PNN网络的大写英文字母电子签字识别研究  

Research on the Recognition of Electronic Signatures with Uppercase English Letters Based on PNN Network

作　　者：王晓娟 杨永昕 WANG Xiaojuan;YANG Yongxin(Faculty of International Business and Management,Chongqing Institute of Foreign Studies,Chongqing 401420,China;Chongqing Survey Institute Co.,Ltd.,Chongqing 401120,China)

机构地区：[1]重庆外语外事学院国际商贸与管理学院,重庆401420 [2]重庆市勘测院有限公司,重庆401120

出　　处：《河南科技》2025年第4期26-29,共4页Henan Science and Technology

基　　金：2022年重庆市教育委员会科学技术研究项目(KJQN202202304)。

摘　　要：【目的】随着计算机技术发展,越来越多的电子签字被应用于现实生活中。为解决电子签字识别问题,利用鲁棒性较好的PNN神经网络来识别输入的电子签名,并转换成可编辑的英文字母。【方法】在识别过程中,为提高识别的正确率,需要对收集到的电子签字进行处理。先提取签字中的黑色有字母部分,再将提取到的黑色有字母部分统一大小,并将统一大小后的黑色有字母部分放在大小一样的白色画布上。经过处理后的电子签字可作为PNN网络输入变量,进行网络识别。【结果】以一组倾斜程度较大的电子签字进行识别验证,结果表明PNN网络整体识别效果较好,对倾斜程度较大的电子签字识别效果较差。【结论】该方法能高效识别倾斜程度较小的电子签字,未来应对倾斜程度较大的电子签字展开进一步的研究。[Purposes]With the development of computers,more and more electronic signatures appear in real life.In order to solve the problem of recognition of these electronic signatures,this paper uses a ro⁃bust PNN neural network to recognize the input electronic signatures into editable English letters.[Meth⁃ods]In the process of recognition,in order to improve the accuracy of recognition,it is necessary to pro⁃cess the collected electronic signatures.Firstly,the black lettered part in the signature is extracted,and then the extracted black lettered part is unified in size,and the unified black lettered part is placed on a white canvas of the same size.The processed electronic signature can be used as an input variable of the PNN network for network recognition.[Findings]In order to test whether the processing method in this paper is effective,a group of electronic signatures with a large degree of skew are added to the paper,and it is verified that the PNN network has a poor recognition effect on electronic signatures with a large de⁃gree of skew.[Conclusions]This method can efficiently identify electronic signatures with less tilt,and further research should be carried out on electronic names with greater tilt in the future.

关 键 词：PNN网络 模式识别 字母处理 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]