An Effectual Approach for Facial Expression Recognition Using Adaptive Neuro Fuzzy Classifier

Abstract: Human facial Expression recognition by computer with high recognition rate is still a challenging and interesting task. A lot of emotional information is conveyed by facial expression alone rather than voice tone and spoken words. Facial expression recognition system plays an important role in many areas such as human-computer intelligent interaction system. In this paper, an Automatic Facial Expressions Recognition System is proposed that would recognize five principal expressions, which are Happy, Sad, Neutral, Angry and Disgusted. The proposed system uses an efficient approach for the recognition of those expressions on the basis of some extracted features. For the detection of the frontal face proposed method uses well known Viola Jones face detection technique. Once face detection is performed, feature of interested region that is eyes and mouth are extracted. In feature extraction, local binary pattern (LBP) is proposed as a feature. After the extraction of the LBP feature for the classification or recognition, the proposed method includes adaptive Neuro Fuzzy Classifier (ANFIS) to efficiently cluster the obtained LBP features. The whole system will implemented on the dataset of 150 images of frontal facial expressions of Happy, Sad, Neutral, Angry and Disgusted from Karolinska Directed Emotional Faces (KDEF) Database by using MATLAB 2013(b) and expected to improve expression recognition efficiency.

Keywords: Facial Expression Recognition, Facial Expressions, Feature Extraction, LBP, ANFIS