Facial Expression Recognition Using LBP and LPQ Feature Fusion with SVM Classification

The MATLAB program for facial expression recognition presented in this work employs a fusion of LBP (Local Binary Pattern) and LPQ (Local Phase Quantization) algorithms, with SVM (Support Vector Machine) serving as the classifier for expression recognition. The implementation extracts texture features using LBP to capture micro-patterns in facial images, while LPQ analyzes phase information to handle blur-invariant characteristics. The code combines these complementary features through feature-level fusion, creating a robust representation that enhances recognition accuracy. The program utilizes MATLAB's Computer Vision Toolbox for image preprocessing and feature extraction, followed by SVM classification through Statistics and Machine Learning Toolbox functions. Key implementation aspects include histogram computation for LBP/LPQ features, feature normalization, and SVM model training with cross-validation. This integrated approach demonstrates significant effectiveness in facial recognition applications, providing improved accuracy for facial expression classification results.