Android平台下的验证码识别研究  

Research on Recognition of the Verification Code Based on Android

作　　者：包乾 李文超[1] 张庆东 Bao Qian;Li Wenchao;Zhang Qingdong(School of Computer and Communication Engineering,Liaoning Shihua University,Fushun Liaoning 113001,China)

机构地区：[1]辽宁石油化工大学计算机与通信工程学院,辽宁抚顺113001

出　　处：《科技通报》2017年第9期73-75,219,共4页Bulletin of Science and Technology

摘　　要：为了在开发移动应用APP时,实现与已有系统的无缝连接,在研究验证码图像特征的基础上,首先采用大津法对待验证图片进行预处理,通过对图像进行灰度化、二值化、分割、归一化处理,去除离散噪声;然后利用Hopfield神经网络对字符和背景颜色随机变化的验证码进行识别。通过测试表明,应用这种方法开发的移动应用服务APP,能够有效识别验证码图像,解决系统开发中的透明集成问题,为用户提供完美的操作界面。In order to realize a seamless connection with existing systems in the development of themobile APP,based on the study of verification code image features,we used Fomier-Mellin topreprocess the verification image.By grayscale,binary,segmentation,normalization processing,weremoved the discrete noise.Then we used hopfield neural network to identify verification code whosecharacters and background colors randomly vary.Tests have shown that the mobile APP developed bythis method can effectively identify verification code image,and it can solve the problem of transparentintegration in system development,and provide users with the perfect interface.

关 键 词：验证码 大津法 HOPFIELD神经网络 ANDROID 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]