Information Security Analysis of Chinese Character Captcha Based on Siamese Neural Network

Abstract

CAPTCHA (Completely Automated Public Turing Test to Tell Computers and Humans Apart) plays a pivotal role in the network security of information systems, and has been widely used in various fields. This paper uses Tensor Flow as the neural network framework and uses Siamese network to reduce noise information and extract the characteristic information of the captcha. The experimental results show that the recognition rate of the Siamese neural network for the four-character image captcha with noise interference is 97.40%. Therefore, the siamese neural network has a strong recognition ability for complex captchas. At the same time, this high recognition ability leads to serious security risks in the information system. Therefore, the current Chinese character captcha design technology still needs to be strengthened and improved.