Facial Expression Recognition Algorithm Based on Local Feature Extraction

This paper employs the Center Symmetric Local Binary Pattern (CSLBP) algorithm for extracting fine features in facial expression recognition. Additionally, the Rotation Invariant Local Phase Quantization (RILPQ) algorithm is utilized to capture accurate features for recognizing blurry facial expressions. The proposed approach achieves comprehensive and detailed extraction of different facial expression features. By constructing a multi-nonlinear SVM classifier structure model, the system accomplishes the classification of facial expression features and facilitates facial recognition. The experimental statistical data indicates that our algorithm is most suitable for a 9×9 pixel block grid. The accuracy of facial expression feature extraction is around 92%, and the accuracy of facial expression classification and recognition is also approximately 92%. For images with an 8×8 feature dimension, the algorithm's processing time is approximately 0.23 seconds.