Multiparameter Space Decision Voting and Fusion Features for Facial Expression Recognition

Area of Science:

• Artificial Intelligence
• Computer Vision

Background:

• Facial Expression Recognition (FER) is a key area in artificial intelligence.
• Developing accurate FER methods remains a significant research challenge.

Purpose of the Study:

• To propose a novel multiparameter fusion feature space and decision voting-based classification for enhanced facial expression recognition.
• To improve the classification of similar facial expressions.

Main Methods:

• Determining fusion feature space parameters via cross-validation accuracy of Multiscale Block Local Binary Pattern Uniform Histogram (MB-LBPUH) descriptor filtering.
• Constructing fusion feature spaces using multiclass linear discriminant analysis (LDA).
• Employing fused MB-LBPUH and Histogram of Oriented Gradient (HOG) features for expression representation.
• Implementing a nearest neighbor-based decision voting strategy for classification.

Main Results:

• The proposed model demonstrated superior performance compared to existing algorithms on the JAFFE, CK+, and TFEID datasets.
• The fusion of MB-LBPUH and HOG features effectively represents facial expressions.
• The decision voting strategy successfully addressed challenges in classifying similar expression classes.

Conclusions:

• The developed facial expression recognition method offers a significant advancement in the field.
• The combination of feature fusion and decision voting provides a robust and accurate FER system.