A Framework of Human Emotion Recognition Using Extreme Learning Machine

Human emotion recognition has been challenging issue in field of human-computer interaction. In order to form an interaction that is more natural between human and computer, the computer should be able to discern and respond to human emotion. In this paper, an approach for recognizing human emotion is proposed. The proposed approach operates HAAR-classifier to detect mouth, eyes, and eyebrow on face, and, to extract features from them, it uses Gabor wavelet. Before classifying the features, PCA is performed to reduce its dimension. The proposed framework employs SLFNs with ELM as its learning algorithm to classify the features. In this experimental, the proposed framework is tested in two cases, personalize and generalize face case, with ten subjects expressing six basic emotions and neural state. The robustness of ELM is evaluated with comparing it to K-NN and SVM. Preliminary experiment shows that the proposed approach has promising performance in personalize face case.