In-air signature verification system using Leap Motion

Abstract

Signature verification is a widely explored field due to its high acceptance and its compromise between security and comfort. Recently, different techniques have appeared to improve the capture, processing, and classification of signatures. In this work, authors present a novel and robust in-air signature verification system, which applies the use of Leap Motion controller to characterize in-air strokes, due to its stability and good performance for this task, as it will be demonstrated. Therefore, a database has been built for developing the experiments, which is composed of 100 users, with 10 genuine and 10 forgery samples per user. The implemented system is tested against two tests of impostor samples, zero effort attacks and active impostors. The second type of attacks are developed by different users, who showed very good abilities with the sensor. The classification is done by a Least Square Support Vector Machine. The equal error rate was 0.25% and 1.20%, respectively. The proposed system achieves very good results in comparison with the state-of-the-art one, which suggests that in-air signature processing gives an opportunity to increase systems’ security.