A Novel Tensor Perceptual Color Framework based Facial Expression Recognition

Abstract

The Robustness of Facial Expression Recognition (FER) is based on information contained in color facial images. The Tensor Perceptual Color Framework (TPCF) enables multilinear image analysis in different color spaces. This demonstrates that the color components provide additional information for robust FER. By using this framework color components RGB, YCbCr, CIELab or CIELuv space of color images are unfolded to 2-D tensors based on multilinear algebra and tensor concepts. The features of this unfolded image are extracted by using log-Gabor filter. The optimum features are selected based on mutual information quotient method in feature selection process. These features are classified using a multiclass linear discriminant analysis classifier. Experimental results demonstrate that color information has significant potential to improve emotion recognition performance due to the complementary characteristics of image textures.