Facial Expression Recognition Using Improved Local Binary Pattern and Min-Max Similarity with Nearest Neighbor Algorithm

Abstract

AbstractIn human-to-human interaction, important role is played by facial expression recognition (FER). Various modes of understanding and expressions out of verbal world are allowed by this. In many applications, it is a challenging task to analyze human facial expression automatically. For better human machine interaction, robust facial expression recognition (FER) system is proposed in this work. In this work, improved local binary patterns (ILBPs) and min-max similarity with nearest neighbor (MMSNN) algorithm are proposed for efficient training and recognition. Initially, the face detection is performed using Haar classifier method. From images, useful information is extracted by applying improved local binary pattern (ILBP) method. Finally, min-max similarity with nearest neighbor (MMSNN) algorithm is applied to train various expressions of face. Traditional methods are used to make a performance comparison of proposed MMSNN method. Mean recognition rate is achieved by proposed method, which shows its ability in comparison with existing methods. With respect to accuracy, f-measure, recall, and precision, better results are produced by the proposed MMSNN method. KeywordsFacial expression recognitionHaar classifier methodImproved local binary patternsMin-max similarity with nearest neighbor (MMSNN)