Gesture recognition implemented on a personal limited device

Abstract

For a biometrics system, one of the principal challenges is to protect the biometric reference template, as if a malicious individual is able to obtain this template, the genuine user would not be able to reuse the biometric for any application. A solution may be to use a new form of authentication based on gesture recognition. This type of authentication has the added advantage that in the case of compromise, the gesture can be changed yet still retain the advantages of the biometric input. In this paper, we investigate whether it is feasible to implement a Gesture Recognition system on a personal limited device such as a smart card. To do this, we set out an experiment using sample gestures based on practical results of gesture authentication trials and an optimised version of Dynamic Time Warping (DTW) algorithm to analyse the data captured. We implemented them on both a contact Smart Card (SC) and the more powerful Samsung Galaxy S4 mobile phone, using Host Card Emulation (HCE). The result of this experiment was that it would take around a minute for the SC and a second for HCE.