Multi-Factor Authentication System Using Keystroke Dynamics and Biometric Face Recognition with Feature Optimization
Jishnu Prasad, Mary Amala Bai · 2024
This project encompasses capturing and optimizing keystroke and biometric data for user authentications and behavioral analysis. Based on the use of computer vision and machine learning techniques, it captures keystroke patterns and facial features. As for keystroke data, it is collected using the pynput library, which logs key press and release times to calculate the duration of each keystroke that could be analyzed. An OpenCV library's Haar Cascade classifier is used to capture the biometric data, that is facial features, by detecting and extracting facial images through the user's webcam. These facial images are then further encoded in base64 format and are stored as part of the biometric profile of users. The GWO algorithm is further utilized for better feature identification and optimal data to enhance performance. The GWO algorithm progressively enhances the facial feature data to achieve the optimal set of features. All captured data will be in the user's specific folder for privacy and data organization. The system works as an advanced mechanism of multi-factor authentication by combining behavioral biometrics such as keystroke dynamics with physiological biometrics like facial features for better accuracy and security. This project could be applied in fields such as cybersecurity, personal identification, and analysis of user behavior.