不变矩特征在数字验证码识别中的应用

摘要

网络给人们提供丰富资源和极大便利,同时,互联网的安全性问题也随之而来.验证码正是加强网络系统安全性的常用手段.不变矩方法在目标识别领域的应用较为广泛,但在验证码识别方面的应用却不多.论文以数字验证码为例,先对验证图像进行去噪、定位、归一化等预处理,然后计算待识别数字的不变矩,再进行特征匹配,得到识别结果.以Matlab软件为开发平台来进行设计实现及仿真,并构建相应的GUI界面.实验结果表明利用不变矩在识别数字验证码方面具有可行性.在教学过程中应用该成果,可以使学生强化基本理论,培养学生的创新能力和理论联系实际的能力.%Network provides tremendous resources and great conveniences to people,at the same time,however,the Internet security is becoming more and more serious. The verification code appears to be the important product of web system security. The in?variant moment method is widely used in the field of object recognition,however,it is not widely used in the field of verification code identification. In this paper,digit verification code is operated as an example. Firstly,the code image to be recognized requires denoising,localization,normalization and other operations. Then the invariant moments of the digit image to be identified are calcu?lated as features. Then features of digit images are used to match in order to obtain recognition result. The Matlab software is used as a running and test platform for program implementation and simulation,and corresponding GUI interface is also constructed. The feasibility of recognition of verification code using moment invariant is illustrated by the experimental results. The application of this fruit strengthens fundamental theory,improves the students'actual operation level and cultivates the students'ability of innovation and combining theory with practice .