Related Experiment Video
Updated: Aug 5, 2026
11:34
High-resolution, High-speed, Three-dimensional Video Imaging with Digital Fringe Projection Techniques
Published on: December 3, 2013
High-speed face recognition based on discrete cosine transform and RBF neural networks
Meng Joo Er1, Weilong Chen, Shiqian Wu
1Computer Control Laboratory, School of Electrical and Electronic Engineering, Nanyang Technological University, Singapore 639758, Singapore. emjer@ntu.edu.sg
IEEE Transactions on Neural Networks
|June 9, 2005
Summary
This study introduces an efficient face recognition method using Discrete Cosine Transform (DCT), Fisher's Linear Discriminant (FLD), and Radial Basis Function (RBF) neural networks. The approach enhances speed and accuracy while maintaining robustness to illumination variations.
Area of Science:
- Computer Vision
- Machine Learning
- Biometrics
Background:
- High-speed and accurate face recognition remains a challenge.
- Illumination variations significantly degrade recognition performance.
- Existing methods often struggle with computational efficiency and robustness.
Purpose of the Study:
- To develop an efficient and robust high-speed face recognition system.
- To improve recognition accuracy under varying illumination conditions.
- To facilitate faster training of neural networks for face recognition.
Main Methods:
- Dimensionality reduction using Discrete Cosine Transform (DCT) by discarding low-frequency coefficients.
- Clustering of DCT coefficients to enhance Fisher's Linear Discriminant (FLD) efficiency.
- Utilizing FLD to extract invariant facial features and RBF neural networks for classification.
Main Results:
- The proposed method achieves high training and recognition speeds.
- Excellent recognition rates were demonstrated in simulations.
- The system exhibits significant robustness to large area illumination variations.
Conclusions:
- The integrated DCT, FLD, and RBF network approach offers an efficient solution for high-speed face recognition.
- The method effectively handles illumination variations, improving overall system reliability.
- This technique facilitates rapid training and high performance in biometric applications.
