Multilingual Highlighting CAPTCHA

Shirali-Shahreza, Sajad; Shirali-Shahreza, M.

doi:10.1109/itng.2011.85

Cited by 5 publications

(2 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Previous work that has investigated handwriting input as online verification has been based on some form of CAPTCHAs (Completely Automated Public Turing Tests to Tell Computers and Humans Apart). In Highlighting CAPTCHA [4] the user must trace an obfuscated word with a stylus, however relying on precise user handwriting is cumbersome to perform on a mobile device due to the inaccuracy of user input [18], [22]. Other approaches require the user to trace a symbol with their mouse or finger, for example a gesture 1 or a math equation [5], arguing that these symbols are language-independent.…”

Section: Related Workmentioning

confidence: 99%

“…On the web, for example, it is possible to analyze mouse movements at no cost and at large scale [3]. Furthermore, some websites are starting to ask their users to solve some form of handwritten captcha [4]- [6], based on the assumption that handwriting input is very natural for humans. Indeed, today's mobile devices have a touchscreen, so users can simply handwrite with their fingers or a stylus both quickly and effortlessly.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Human or Machine? It Is Not What You Write, But How You Write It

Leiva¹,

Diaz²,

Ferrer³

et al. 2020

Preprint

View full text Add to dashboard Cite

Online fraud often involves identity theft. Since most security measures are weak or can be spoofed, we investigate a more nuanced and less explored avenue: behavioral biometrics via handwriting movements. This kind of data can be used to verify whether a user is operating a device or a computer application, so it is important to distinguish between human and machine-generated movements reliably. For this purpose, we study handwritten symbols (isolated characters, digits, gestures, and signatures) produced by humans and machines, and compare and contrast several deep learning models. We find that if symbols are presented as static images, they can fool state-ofthe-art classifiers (near 75% accuracy in the best case) but can be distinguished with remarkable accuracy if they are presented as temporal sequences (95% accuracy in the average case). We conclude that an accurate detection of fake movements has more to do with how users write, rather than what they write. Our work has implications for computerized systems that need to authenticate or verify legitimate human users, and provides an additional layer of security to keep attackers at bay.

show abstract