Biometric Pattern Verification On Smartphone

Abstract

In the 21th century, almost everyone owns at least one smartphone. A lot of people tend to rely on their smartphones for information storage, performing daily tasks, communication, so privacy of users and information security are the important topics that worth exploring.

The first barrier to keep users’ phones secure is always the screen locks. There are different types of screen locks, for example the PIN, 9-dot pattern, or phone locking using fingerprint. However, these patterns or features are insecure and unsafe as they can be easily copied by others.

Therefore, this project aims to develop an android app that can create an unlock pattern for verifying the identity of users through their biometric characteristics. This can be done by collecting coordinates, pressure, contact area, velocity of the pattern, and other biometric data of users when they custom their specific drawing pattern on the touch-screen smartphones. Then Neural Network is used for building a specific verification for users. Finally, the trained algorithm is used to verify the patterns on the smartphone. The experimental results on dataset of 5 participants show that the algorithm has satisfactory performance on pattern verification with high accuracy.