基于改进VGG-16的手写数字识别方法研究  

Research on Handwritten Digit Recognition Method Based on Improved VGG-16

作　　者：李凌云 LI Lingyun(Hunan Polytechnic of Environment and Biology,Hengyang Hunan 421005,China)

机构地区：[1]湖南环境生物职业技术学院

出　　处：《信息与电脑》2021年第18期27-29,共3页Information & Computer

基　　金：基于深度学习的手写数字识别方法研究(项目编号:ZK2020-08)。

摘　　要：传统的朴素贝叶斯手写数字识别方法的识别效果差,准确率低。因此本文基于改进VGG-16设计了新的手写数字识别方法。处理手写数字图像,分割图像字符,在此基础上,基于改进VGG-16提取了手写数字特征,实现了手写数字识别。实验结果表明,设计的手写数字识别方法准确率高,能快速进行特征识别,有一定的应用价值。The traditional Naive Bayesian handwritten digit recognition method has poor recognition effect and low accuracy.Therefore,a new handwritten digit recognition method is designed based on the improved VGG-16.Handwritten digital images are processed and image characters are segmented.On this basis,handwritten digital features are extracted based on the improved VGG-16,and handwritten digital recognition is realized.The experimental results show that the designed handwritten digit recognition method has high accuracy,can quickly perform feature recognition,and has certain application value.

关 键 词：改进VGG-16 手写数字识别 特征提取 数字降维 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]