A Biometric-Finger Vein Authentication System for Security Purpose using Deep Learning Technique

Abstract

Biometrics uses human physiological characteristics, is one way of protecting personal information. The usage of finger vein as a form of biometric become most popular method recently. Finger vein authentication provides a high level of security and accuracy, making it a reliable biometric authentication method. Finger vein authentication system compares the vascular structure of a person's finger with previously acquired data. This technique involves identifying patterns in vein images of human fingers below the skin's surface. The proposed system aims to enhance the security of user authentication by utilizing the unique features of finger vein patterns. The finger vein image is acquired from the database. The preprocessing done in order to remove the noise by means of Gaussian median filter in spatial domain and frequency domain. The segmentation of the image carried out through line tracking method which provided the better contrast image. The system utilizes Convolutional Neural Networks for feature extraction and the features are matched with the finger vein database. Then Deep Learning Approach used for classification of finger vein patterns between genuine and imposter users. For real-time communication the scanner scans the finger vein and send to Arduino board for storage followed by MATLAB for processing and classification of the images. The result is sent through GSM module as alarm or message. Then the information also stored in IoT for future references. with the user a GSM Module is integrated. The proposed system gives an accuracy of 96%. Thus, system is beneficial in several security applications such as access control, identity verification, banking system, and financial transactions.