Mean AR-Local Binary Pattern For Facial Expressions Classification Based on KNN Classifier

According to its significance, facial expression recognition (FER) has reached an incredibly prominent position in the field of image processing, beginning in the 1990s and continuing to the present day. Face recognition, feature extraction, and face categorization are the three common steps in FER. In contrast to the various FER systems described for recognizing just part of face expressions, the system presented in this study can identify all eight fundamental facial emotions (neutral, pleased, angry, contempt, surprise, sad, fear, and disgust). We employ the Japanese Female Facial Expression (JAFFE) and the Extended Cohn-Kanade (CK+) datasets to verify our results. The presented methods LBP Patterns in this work namely, Asymmetric Region LBP(AR-LBP), Mean Asymmetric Region LBP (Mean AR-LBP), We used GLCM feature extractor technique for feature selection. For the given input image, The GLCM calculates different features factors named Contrast, Correlation, and Homogeneity, Entropy, Mean, variance, and dissimilarity which were results to a feature vector. The findings of the given approach, which employed K-Nearest Neighbor (KNN) to categories facial emotions, are then contrasted with those of the standard LBP Pattern. The recognition rate utilizing the provided technique is 96% for the CK + data set and 97% for the JAFFE data set, as shown by the testing results.