Facial Expression Recognition Based on Fusion Features of Center-Symmetric Local Signal Magnitude Pattern

Local feature descriptors play a fundamental and important role in facial expression recognition. This paper presents a new descriptor, Center-Symmetric Local Signal Magnitude Pattern (CS-LSMP), which is used for extracting texture features from facial images. CS-LSMP operator takes signal and magnitude information of local regions into account compared to conventional LBP-based operators. Additionally, due to the limitation of single feature extraction method and in order to make full advantages of different features, this paper employs CS-LSMP operator to extract features from Orientational Magnitude Feature Maps (OMFMs), Positive-and-Negative Magnitude Feature Maps (PNMFMs), Gabor Feature Maps (GFMs) and facial patches (eyebrows-eyes, mouths) for obtaining fused features. Unlike HOG, which only retains horizontal and vertical magnitudes, our work generates Orientational Magnitude Feature Maps (OMFMs) by expanding multi-orientations. This paper build two distinct feature maps by dividing local magnitudes into two groups, i.e., positive and negative magnitude feature maps. The generated Gabor Feature Maps (GFMs) are also grouped to reduce the computational complexity. Experiments on the JAFEE and CK+ facial expression datasets showed that the proposed framework achieved significant improvement and outperformed some state-of-the-art methods.