Author(s): Utkarsh Gupta | Jasraj Fukane | Varshini Ramanan | Rohit Thakur
Journal: International Journal of Computer Applications
ISSN 0975-8887
Volume: icwet;
Issue: 4;
Date: 2012;
Original page
Keywords: Unimodal Biometric Authentication System (UBAS) | Multimodal Biometric Authentication System (MBAS) | Percentage Confidence (pC) or Accuracy Score | Genuine Acceptance Rate (GAR) | Imposter Acceptance Rate (IAR)
ABSTRACT
In many real-world applications, unimodal biometric systems often face significant limitations due to sensitivity to noise, intra class variability, data quality, pressure, dirt, dryness and other factors. Multimodal biometric authentication systems aim to fuse two or more physical or behavioral traits to provide optimal Genuine Acceptance Rate (GAR) Vs Imposter Acceptance Rate (IAR) curve i.e. Receiver's Operating Characteristic (ROC). This paper presents a real time multimodal biometric authentication system integrating finger and face traits based on weighted score level fusion. Each biometric trait produces a varied range of scores i.e. heterogeneous scores. Various scores normalization techniques have been developed for fusion of such scores. Whereas this paper presents a technique for producing compatible scores (homogeneous). We have observed interesting variations in ROC through experimental analysis by changing the number of Eigen Faces in Face Verification Module for considering real time vibrations of input face. The statistical analysis for optimized ROC using fusion of the two traits is also represented.
Journal: International Journal of Computer Applications
ISSN 0975-8887
Volume: icwet;
Issue: 4;
Date: 2012;
Original page
Keywords: Unimodal Biometric Authentication System (UBAS) | Multimodal Biometric Authentication System (MBAS) | Percentage Confidence (pC) or Accuracy Score | Genuine Acceptance Rate (GAR) | Imposter Acceptance Rate (IAR)
ABSTRACT
In many real-world applications, unimodal biometric systems often face significant limitations due to sensitivity to noise, intra class variability, data quality, pressure, dirt, dryness and other factors. Multimodal biometric authentication systems aim to fuse two or more physical or behavioral traits to provide optimal Genuine Acceptance Rate (GAR) Vs Imposter Acceptance Rate (IAR) curve i.e. Receiver's Operating Characteristic (ROC). This paper presents a real time multimodal biometric authentication system integrating finger and face traits based on weighted score level fusion. Each biometric trait produces a varied range of scores i.e. heterogeneous scores. Various scores normalization techniques have been developed for fusion of such scores. Whereas this paper presents a technique for producing compatible scores (homogeneous). We have observed interesting variations in ROC through experimental analysis by changing the number of Eigen Faces in Face Verification Module for considering real time vibrations of input face. The statistical analysis for optimized ROC using fusion of the two traits is also represented.