Face recognition by using 4D Discrete Cosine Transform

In this study, a new method is presented for face recognition by using DCT4-faces that are calculated by the 4D discrete cosine transform. In this method, firstly, 3D struct is obtained for each person by collecting training face images on top of each other. Then 4D DCT is applied to 4D struct which is obtained from 3D structs of each person. Lasty, 2D inverse DCT is applied to each 2D coefficient layer separately. After this operation, DCT4-faces are found. By taking the inner product of DCT4-faces and original face images, feature vectors are calculated to be used in face recognition. The main difference of the presented method from other DCT-based face recognition methods is that DCT4-faces does not directly use DCT coefficients as feature vector.