User authentication using passowrd and hand gesture with leap motion sensor

Abstract

User Authentication is becoming a significant factor in the field of modern technology. It is a process that permits a device to confirm the recognition of somebody who interfaces with a system asset. In the world of AI, machine learning is currently one of the leading research fields which is looking into practical implementation. In this report, we propose a method where the user will enter the given password while leap motion sensor will compare the behavioural data of the user with an existing dataset. Leap motion controller is a sensor or gadget which can recognize 3D movement of hands, fingers and finger like articles with no contact. Moreover we will be discussing the benefits of using behavioural biometrics instead of physiological biometrics for security, and how behavioural biometrics can solve the faults of physiological biometrics. In addition, we will be discussing the benefits of using leap motion sensor along with password authentication to properly identify an user and how it can improve security. For our project, we chose to use Dynamic Time Warping and Naive Bayes Classifier algorithm. DTW algorithm will be useful by comparing two frames which differ in time or velocity when one user have multiple behavioral entries before identifying user as valid or invalid. Naive Bayes will classify a user as valid or invalid through allowing classifiers to learn user data through features. The proposed system has about 91% accuracy which rises to 93% in the best-case scenario. We believe that because Leap Motion is comparatively low cost at the exchange of an extra layer of security it provides, the proposed system can ensure a secure and efficient environment for user authentication.