Related Experiment Videos
Implementation of a high-speed face recognition system that uses an optical parallel correlator
Eriko Watanabe1, Kashiko Kodate
1Department of Mathematical and Physical Science, Japan Women's University, 2-8-1 Mejirodai, Bunkyo-ku, Tokyo 112-8681, Japan. eriko@fourier.jwu.ac.jp
Applied Optics
|March 9, 2005
Summary
A novel optical parallel correlator achieves rapid, fully automatic face recognition. This system identifies individuals from a database of 4000 faces in under 1.5 seconds with low error rates.
Area of Science:
- Optics and Photonics
- Biometrics and Pattern Recognition
- Computer Vision
Background:
- Traditional face recognition systems face challenges with speed and accuracy, especially with varied facial appearances.
- The need for high-speed, automated identification systems is growing in security and surveillance applications.
Purpose of the Study:
- To develop and evaluate a fully automatic, high-speed face recognition system.
- To demonstrate the efficacy of an optical parallel correlator for real-time biometric identification.
Main Methods:
- Implementation of a custom-designed 1000 frame/s optical parallel correlator.
- Development of a binary real-only matched filter optimized for face recognition.
- System optimization based on minimizing false-rejection rate (FRR) and false-acceptance rate (FAR) using biometrics guidelines.
Main Results:
- Achieved operational speed of less than 1.5 seconds for 1:N identification (4000 images), including preprocessing.
- Obtained low error rates: 2.6% FRR and 1.3% FAR in 1:N identification experiments.
- Successfully identified challenging facial images, including those with thin glasses or heavy makeup.
Conclusions:
- The optical parallel correlator offers a viable solution for fast and accurate automatic face recognition.
- The developed system demonstrates robustness in identifying individuals under difficult conditions.
- This technology has significant potential for applications requiring rapid biometric authentication.