基于AlexNet的手写希腊字母识别研究  

A Study on the Recognition of Handwritten Greek Letters based on AlexNet

作　　者：向本华 王徽 XIANG Benhua;WANG Hui(Liaoning Technical University,Huludao Liaoning 125105,China)

机构地区：[1]辽宁工程技术大学,辽宁葫芦岛125105

出　　处：《通信技术》2021年第5期1103-1108,共6页Communications Technology

摘　　要：针对手写希腊字母识别的问题,使用了一种基于卷积神经网络(AlexNet)的图像识别方法。利用采集的手写体希腊字母大小写样本,对其预处理后,将其划分为训练集和测试集,对AlexNet模型进行训练,测试得到的大小写字母识别准确率分别达到98.27%和96.07%。结果表明,将AlexNet应用于手写希腊字母识别有着良好的识别效果,验证了该方法的合理性和有效性,具有一定的现实意义。In view of the problem of handwritten Greek letter recognition,this paper uses a convolution neural network(AlexNet)based image recognition method.After preprocessing the handwritten Greek letter case sample,it is divided into training set and test set.After training the AlexNet model,the accuracy of case letter recognition increases to 98.27%and 96.07%respectively.As a result,the application of AlexNet to handwritten Greek letter recognition has a good recognition effect,which verifies the rationality and effectiveness of the method,and has certain practical significance.

关 键 词：手写希腊字母识别 卷积神经网络 AlexNet 图像识别 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]