Micro-expression recognition using optical flow and local binary patterns on three orthogonal planes
Tian Chang, Fei Long, Jianming Huang · 2019
Micro-expression recognition attracts more and more attentions from computer vision due to its wide range of potential applications. This paper proposes an approach for micro-expression recognition, Local Optical Flow Binary Patterns on Three Orthogonal Planes (LOFBP-TOP), by combining optical flow and LBP-TOP. We first compute optical flow on cropped micro-expression videos, then we encode the magnitude and orientation component of optical flow with LBP-TOP to get LOFBP-TOP features. Furthermore, ReliefF is used to do feature selection on LOFBP-TOP features. Finally, SVM is used in classification. Experimental results on two datasets, CASME2 and SMIC, show that our method has better recognition performance than some popular methods in micro-expression recognition, such as LBP-TOP and HOOF.