Publication: Employing feature extraction for middle part of finger creases on FPGA
Loading...
Date
2023-07
Authors
Lum, Kah Hong
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Biometric systems are used in many applications nowadays due to their high conveniency and security. However, biometric systems nowadays have weaknesses
such as unrecognized biometric traits, high risk database and long recognition time. This project had proposed a FPGA based finger creases recognition system as a solution. The parallelism property of FPGA can improve the biometric system recognition speed. The finger creases with less worn-out chance can be recognize more easily. The feature extraction and classification algorithm use in this project required small amount of biometric that have been protected to store in database. To develop a finger creases recognition system, some problem needed to be solved in this project. Since the finger creases identifier is a contactless biometric system, the variation and blurring of finger creases image is a challenge to the accuracy of finger creases recognition system. Besides, the limitation of the DE1-SoC development board will affect the execution time of finger creases recognition system. The process of this project was categorized into three phases: image pre-processing phase, feature extraction phase and image classification phase. The feature extraction method use in this project is Local Binary Pattern (LBP) and Centre Symmetric Local Binary Pattern (CS-LBP) while the classification method use in this project is Chi Square test and Score Level Fusion. The feature extraction and classification method are developed into several different combination and the result is determined. The result show that the LBP multiple finger creases recognition system has the highest accuracy which 96% and able to compare the input finger creases with 5 finger creases subject imultaneosly. This making the finger creases recognition system suitable for practical applications such as access control and security systems. The findings have significant implications for improving security and identification systems. Overall, this study provides a valuable contribution to the field of biometric recognition and demonstrates the feasibility of using finger creases as an alternative biometric identifier.