Related Experiment Video
Updated: May 6, 2026
09:26
Quantification of Orofacial Phenotypes in Xenopus
Published on: November 6, 2014
9.2K
Membership-degree preserving discriminant analysis with applications to face recognition
Zhangjing Yang1, Chuancai Liu, Pu Huang
1School of Computer Science and Engineering, Nanjing University of Science and Technology, Nanjing 210094, China.
Computational and Mathematical Methods in Medicine
|November 14, 2013
Summary
We introduce Membership-Degree Preserving Discriminant Analysis (MPDA), a new feature extraction method for face recognition. MPDA effectively reduces data dimensionality while preserving crucial class information for improved accuracy.
Area of Science:
- Computer Science
- Artificial Intelligence
- Pattern Recognition
Background:
- High-dimensional data in pattern recognition often requires dimensionality reduction.
- Feature extraction is crucial for efficient and accurate data analysis.
- Existing methods may not fully capture the nuances of class separability.
Purpose of the Study:
- To propose a novel feature extraction algorithm, Membership-Degree Preserving Discriminant Analysis (MPDA).
- To enhance face recognition accuracy by incorporating fuzzy set theory into discriminant analysis.
- To improve dimensionality reduction techniques in pattern recognition.
Main Methods:
- Developed MPDA algorithm combining Fisher criterion and fuzzy set theory.
- Utilized Fuzzy k-Nearest Neighbor (FKNN) to calculate sample membership degrees.
- Modified between-class and within-class scatter definitions using membership degrees.
Main Results:
- MPDA demonstrated effectiveness in feature extraction for face recognition.
- Experimental results on ORL, Yale, and FERET databases validated the algorithm's performance.
- The proposed method showed superior performance compared to existing techniques.
Conclusions:
- MPDA offers an effective approach for dimensionality reduction in pattern recognition.
- Incorporating fuzzy membership degrees enhances the discriminative power of feature extraction.
- The algorithm shows significant potential for real-world face recognition applications.
