Related Experiment Video
Updated: Jul 31, 2026
13:44
Detection of Architectural Distortion in Prior Mammograms via Analysis of Oriented Patterns
Published on: August 30, 2013
Tensor Rank Preserving Discriminant Analysis for Facial Recognition
Summary
A new tensor-based facial recognition algorithm, tensor rank preserving discriminant analysis (TRPDA), preserves spatial data lost in traditional methods. TRPDA enhances accuracy by utilizing the natural structure of facial images for improved recognition.
Area of Science:
- Computer Vision
- Pattern Recognition
- Machine Learning
Background:
- Traditional facial recognition algorithms reshape images into vectors, losing spatial information.
- Existing tensor-based methods can be computationally expensive.
Purpose of the Study:
- Propose a novel tensor-based feature extraction algorithm for facial recognition.
- Improve facial recognition accuracy and efficiency by preserving spatial constraints.
Main Methods:
- Introduced tensor rank preserving discriminant analysis (TRPDA).
- TRPDA involves two stages: low-dimensional tensor subspace extraction and discriminative locality alignment.
- The algorithm optimizes tensor spectral analysis directly, reducing computation.
Main Results:
- TRPDA effectively utilizes the natural structure of facial image tensors.
- The method preserves intra-class rank order information.
- Experimental results on three facial databases demonstrate TRPDA's effectiveness.
Conclusions:
- TRPDA offers an efficient and effective approach to facial recognition.
- The algorithm overcomes limitations of traditional vector-based and some tensor-based methods.
- Preserving spatial and rank order information is crucial for robust facial recognition.
