Facial Expression Recognition with Marginal Fisher Analysis on Local Binary Patterns

A novel approach to facial expression recognition with Marginal Fisher Analysis (MFA) on Local Binary Pattern (LBP) is proposed. Firstly, each image is transformed by an LBP operator and then divided into 3 × 5 non-overlapping blocks. The features of facial expression images are formed by concatenating the LBP histogram of each block. Secondly, MFA algorithm based on Graph Embedding (GE) is applied for dimensionality reduction. Finally, Support Vector Machine (SVM) is used to classify the seven expressions (anger, disgust, fear, happiness, neutral, sadness and surprise) on Japanese Female Facial Expression (JAFFE) database. The maximum facial expression recognition rate of the proposed algorithm reaches to 65.71% for person-independent expression recognition, which is better than LBP + LDA algorithms. The experiment results prove that the facial expression recognition with MFA on LBP is an effective and feasible algorithm.