Related Experiment Video
Updated: May 8, 2026
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
Optimized periocular template selection for human recognition.
Sambit Bakshi1, Pankaj K Sa, Banshidhar Majhi
1Department of Computer Science and Engineering, National Institute of Technology Rourkela, Odisha 769008, India. sambitbaksi@gmail.com
Biomed Research International
|August 29, 2013
Summary
This study introduces four novel methods for optimal periocular template selection, balancing recognition accuracy with system speed. These techniques enhance human recognition systems by efficiently processing biometric data.
Area of Science:
- Biometrics
- Computer Vision
- Pattern Recognition
Background:
- Periocular region biometrics offers a robust alternative for human recognition.
- Template size impacts recognition accuracy and system performance.
- Optimizing template selection is crucial for efficient biometric systems.
Purpose of the Study:
- To develop and evaluate novel approaches for optimal periocular template selection.
- To address the trade-off between template size and recognition accuracy.
- To enhance the efficiency and effectiveness of periocular-based human recognition.
Main Methods:
- Proposed four distinct dynamic optimal template selection methods.
- Tested methods on publicly available UBIRISv2 and FERET databases.
- Evaluated performance based on recognition accuracy and computational efficiency.
Main Results:
- Achieved satisfactory recognition results using the proposed methods.
- Demonstrated the effectiveness of dynamic optimal template selection.
- Validated the approach on unconstrained, real-world datasets.
Conclusions:
- The proposed dynamic optimal template selection is effective for periocular recognition.
- This approach can improve the speed and accuracy of biometric systems.
- Applicable for individual recognition in organizations and national identification systems.
