Media Technology MSc

MSc Thesis


Title

Using Gesture-Paths for Identification on Mobile Devices

Student(s)

Roland Brouwer

Year

2012

Abstract

Research on gesture-recognition engines has allowed these engines to be implemented in a wide range of apparatus in recent years. It allows users to control touch-interfaces with more freedom and has exposed a broader audience to use gestures in day to day applications. Research in this field has sought to make gesture-recognition engines to be as flexible as possible, thus allowing users more freedom in the performance of gestures.
In this study it has been proposed that participants can be identified by analyzing the execution of gestures on platforms which utilize touch-screens. In order to do so a number of gesture-samples have been collected and analyzed by letting participants play a game on a mobile phone equipped with a capacitive touch-surface. The game has been developed to enforce users to perform the gestures as fast as they can without being able to reflect on what has been drawn. To compare the collected gesture-paths for dynamic characteristics the dynamic time warping algorithm has been employed.
With this method a 88-99% recognition rate for finding variation in gestures has been achieved. However for future work it is recommended that gestures are compared with the same method while using a standard environment instead of a game-environment.

Full Reference

Roland Brouwer, "Using Gesture-Paths for Identification on Mobile Devices", Master's Thesis for the Media Technology programme, Leiden University (The Netherlands), 2012