Hand Gesture Authentication Using Optimal Feature Selection and Dynamic Time Warping based K-nearest Neighbor
Jungpil Shin, Md. Al Mehedi Hasan, Md. Maniruzzaman · 2022
Network security is becoming a critical issue due to the spread of computer networks to ordinary people. User authentication is one of the most important part for network security. Although passwords are used in many personal computers, it has many well-known problems and it is antiquated. On the other hand, fingerprint or face based authentication has the risk of information leakage by reproducing fingerprint or face from the photograph. In this paper, we have designed a hand gesture-based user authentication using leap motion to overcome of information leakage problem. We have collected data from 25 participants and used the data to analyse the accuracy of our authentication system. The experimental results are also measured by FAR (False Acceptance Rate) and FRR (False Rejection Rate). At first, Feature selection has been used to select important feature out of 246 features. Finally, Dynamic Time Warping (DTW) based K-Nearest Neighbor (KNN) Classifier has been used to authenticate a person. Our system has produced 96.73% accuracy in user authentication. It is also showed 1.25% FAR and 3.62% FRR. Our results indicate that proposed hand gesture with Leap Motion can be viable authentication approach.