Facial Expression Recognition Based on G-2DPCA Feature and Sparse Representation Classification
Yan Wang, Ming Li, Congxuan Zhang, Hao Chen · 2018
In this paper, we proposed an effective method of facial expression recognition based on a G-2DPCA feature and Sparse Representation-based Classification (SRC). Gabor filters with five scales and eight directions are first employed for feature extraction. To address the high dimension of Gabor features, we select one out of forty Gabor filters with an optimal parameter pair of scales and directions to filter facial images. Two-dimensional principal component analysis (2DPCA) is utilized for image representation and data dimension reduction. It retains the 2D geometric structure of an image, and the image matrix does not need to transform into a vector, which reducing the computation time greatly. Finally, Gabor plus 2DPCA (G-2PCA) features are regarded as the atoms of dictionary in SRC. Experimental results demonstrate that the proposed method has better performance than the existing algorithms.