Algorithm Design for Grip-Pattern Verification in Smart Gun

The Secure Grip project1 focuses on the development of a hand-grip pattern recognition system, as part of the smart gun. Its target customer is the police. To explore the authentication performance of this system, we collected data from a group of police officers, and made authentication simulations based on a likelihood-ratio classifier. This smart gun system has been proved to be useful in the authentication of the police officers. However, its authentication performance needs some further improvement, especially when data for training and testing were collected with some time in between. We present and analyze the simulation results of the authentication experiment. Based on the analyses, we propose some methods to improve the system’s authentication performance. Keywords— grip pattern recognition, smart gun, likelihood-ratio, support vector machine