Augmented Local Binary Patterns for Facial Expression Recognition

Abstract

Facial expression recognition has attracted the attention of many researchers due to its potential application areas like Human Computer Interaction, Sign Language Recognition, Virtual Reality Systems, etc. For effective facial expression recognition, extraction of proper and sufficient features from the facial image is the most important step. This paper presents a brief review of the most popular appearance based feature extraction technique called Local Binary Pattern with an intention to evaluate its applicability for facial expression recognition and proposes additional features to be used for augmenting the local binary patterns to improve the performance of an automatic facial expression recognition system. The proposed method which uses histograms with suitable number of bins for representing pixel intensity values augments the local binary patterns quite well and achieves considerable improvement in the recognition efficiency as against that obtained by local binary patterns. The method is applied to images from two well-known databases namely JAFFE and Cohn Kanade in a two phase process wherein the first phase establishes the need of augmenting local binary patterns with pixel intensity values while the second phase uses the augmented feature vector for recognition.