An information theoretic mouse trajectory measure

Abstract

In this paper, we propose the Relative Trajectory Information (RTI) measure, an information theoretic measure to evaluate mouse pointer trajectories. The measure is used to score the level of smoothness of mouse pointer trajectories. We show that, by leveraging Gaussian processes and information theory, RTI accounts for relative differences in timestamps of the mouse pointer trajectories. RTI also does not require explicit descriptions of targets, in either their location or size. Our experimental analysis shows how RTI can capture the motion signature of a user with severe motion disabilities and distinguish it from the motion signature of smooth trajectories obtained in a control experiment. © 2011 Springer-Verlag.