Extracting patterns from human gaits with Android smartphones - 07/09/2010

Hongchang Wu

Abstract:

Biometric data are high in entropy and can be potentially used for secure authentication schemes. Research has shown that human gaits are unique and therefore can be used as a source for biometric data. Android smartphones are equipped with different kinds of sensors (e.g. accelerometers and magnetometers), and are ideal for getting biometric data from gaits. In this project, we collect walking data of different subjects walking in different environments from the accelerometer sensors of Droid. By employing wavelet transform and statistical pattern recognition, we achieve a high classification rate across different subjects and environments. Methods used for the gait analysis and alternative approaches will be elaborated. Limitations and challenges for future work will also be discussed.

-- JiaxiJin - 08 Jul 2010