JOURNAL ARTICLE

Discriminative uncorrelated neighborhood preserving projection for facial expression recognition

Abstract

In this paper, we propose a novel supervised algorithm named discriminative uncorrelated neighborhood preserving projections (DUNPP), which is a variant of Neighborhood preserving projections (NPP). Combining with class relations between data samples in each local area, the DUNPP method can find a discriminative subspace where the within-class structure is preserved, while the margin between points from different classes is maximized. Also, a simple uncorrelated constraint is added to the objective function of DUNPP to remove redundancies contain in original data and ensure the independence of features, so that the recognition performance can be further enhanced. Experimental results on a widely used facial expression database verified the effectiveness and robustness of our proposed method.

Keywords:
Discriminative model Pattern recognition (psychology) Robustness (evolution) Subspace topology Artificial intelligence Uncorrelated Computer science Constraint (computer-aided design) Margin (machine learning) Projection (relational algebra) Mathematics Algorithm Machine learning Statistics

Metrics

2
Cited By
0.28
FWCI (Field Weighted Citation Impact)
19
Refs
0.56
Citation Normalized Percentile
Is in top 1%
Is in top 10%

Citation History

Topics

Face and Expression Recognition
Physical Sciences →  Computer Science →  Computer Vision and Pattern Recognition
Blind Source Separation Techniques
Physical Sciences →  Computer Science →  Signal Processing
Neural Networks and Applications
Physical Sciences →  Computer Science →  Artificial Intelligence
© 2026 ScienceGate Book Chapters — All rights reserved.