Face Recognition Based on Gabor with 2DPCA and PCA

In this paper, a new method combined Two-Dimensional Principal Component Analysis (2DPCA) with Principal Component Analysis (PCA) is proposed to extract Gabor features for face recognition. Gabor wavelet has been widely used in the face recognition task because it's good imitation of human visual. However, the huge redundancy of Gabor features limits its application. When processing an image use a group Gabor nuclear with five scale and eight directions, the date obtained is enormous. The traditional Two-Dimensional Principal Component Analysis (2DPCA) can limit relativity between Columns, but the number of features is still large, which affects the speed of classification. To resolve this problem, the author uses a method based on Gabor wavelet matrix applied to 2DPCA features matrix and Principal Component Analysis (PCA) for the feature extraction in this paper. The experiment results showed that the performance is superior to single 2DPCA or Gabor with 2DPCA.