Micro-expression Recognition Using Dynamic Textures on Tensor Independent Color Space

Micro-expression is a brief involuntary facial expression which reveals genuine emotions and helps detect lies. It intrigues psychologists and computer scientists' (especially on computer vision and pattern recognition) interests due to its promising applications in various fields. Recent research reveals that color may provide useful information for expression recognition. In this paper, we propose a novel color space model, Tensor Independent Color Space (TICS), for enhancing the performance of micro-expression recognition. An micro-expression color video clip is treated as a fourth-order tensor, i.e. a four-dimension array. The first two dimensions are the spatial information, the third is the temporal information, and the fourth is the color information. We transform the fourth dimension from RGB into TICS, in which the color components are as independent as possible. The combination of dynamic texture in the independent color components can get higher accuracy than that in RGB. In addition, we define a set of Regions of Interest (ROIs) based on Facial Action Coding System (FACS) and calculated the dynamic texture histograms for each ROI. The experiments are conducted on two micro-expression databases, CASME and CASME 2, and the results show that the performance in TICS is better than that in RGB or gray.