Exploiting the Symmetrical Characteristic of Faces to Classify Face Images

In this paper, we propose to exploit the symmetrical characteristic of the face to represent and classify face images. The proposed method first partitions each face image into two halves, i.e. the left part and the right part of the face image. The method then uses a linear combination of the left parts of all the training samples to represent the left part of the testing sample. Also, this method employs a linear combination of the right parts of all the training samples to represent the right part of the testing sample. Finally, the method combines the two representation result to classify the testing sample. The conducted face recognition experiments show that the proposed method can produce a lower error rate than the comparison method.