Centralized Binary Patterns Embedded with Image Euclidean Distance for Facial Expression Recognition

The existing local binary pattern (LBP) operators have three disadvantages: (1) They produce rather long histograms, which slow down the recognition speed especially on large-scale face database; (2) Under some certain circumstance, they miss the local structure as they don´t consider the effect of the center pixel; (3) The binary data produced by them are sensitive to noise. Aiming at these problems, we proposed centralized binary pattern (CBP) operator. CBP operator has several advantages: (1) It reduces significantly the histograms´ dimensionality by comparing pairs of neighbors in the operator; (2) It considers the center pixel point´s effect and gives it the largest weight, thus improving discrimination; (3) By modifying the sign function of existing LBP operator, it decreases the white noise´s influence on face images. Moreover, for the purpose of improving the robustness to small perturbation (deformation) of expressional images, we introduced image Euclidean distance (IMED) and embedded it in CBP. Experiments on two well-known facial expression databases demonstrate that the proposed method outperforms other modern approaches and show that IMED can enhance the performance of CBP in facial expression recognition.