Acceleration Axis Selection in Biometric Technique Based on Gesture Recognition

This article proposes a biometric technique based on gesture recognition performed directly in a mobile device embedding an accelerometer. As time consumption is an essential requirement, this article aims to discover the most distinctive acceleration axis information in order to find the best strategy considering EER and consumed time. Best EER result of 2.5% has been obtained when the information of accelerations on each of the three axis is analyzed. When reducing the information inspected to only two or one acceleration axis signals, EER values are 2.98% and 4.34% respectively. Preprocessing various acceleration signals by calculating their magnitude outcomes with higher EER values. All this work has been developed from a database of 34 individuals who have performed their identifying gestures, and three falsifiers who have attempted to forge each original in-air signatures from studying video records.