Virtual Keyboard Logging Counter-Measures Using Human Vision Properties

This paper describes keylogging counter-measures for virtual keyboards, widely used to authenticate in various applications and contexts, such as online banking services or touch screen mobile devices. Due to this massive use and to the malware landscape, the security of a virtual keyboard at authentication time is fondamental. Our work is based upon several human vision properties like motion perception, visualas similation and visual interpolation. We, first, generate a frame filled with noise and then we manipulate this noise in order to make a user see shapes, e.g. letters or digits. The recognitionof these shapes by a malware would require high analysis capabilities to automatize. This way, we achieved to make a human-readable virtual keyboard resilient to a large scope of screenshot-based keylogging methods.