Abstract
The traditional text captcha adds adversarial perturbation based on a single character, and it is necessary to joint and compound the captcha picture, so that the characters and the background are easy to segment. In order to strengthen the security of text captcha, an AECAPTCHA text captcha defense algorithm with overall adversarial perturbations is proposed. The algorithm uses multi-label classification training text captcha as a pre-training model. Therefore, the AECAPTCHA algorithm is used to generate multi-character overall adversarial perturbation, eliminate the steps of splicing, compositing, and adding colors, then superimpose to generate safe text captcha with adversarial examples properties. The experimental results show that text captcha with the overall adversarial perturbation reduced the accuracy to 0.06%, effectively improves the security of the text captcha, and does not affect the user’s rapid identification and use.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.