3. Assume that 200 face images are available for training, and the face images are represented by Gabor features, extracted using Gabor filters of 5 center frequencies and 8 orientations. The Gabor feature of the i-th training image is denoted as fi, where i = 1, ..., 200. Assume that principal component analysis is applied to the training images, and the computer used can compute the eigenvectors and eigenvalues of a matrix with dimensions up to 1,000x1,000. (a) What is the dimension of the feature vector fi? (b) Describe how to compute the principal components of the feature vectors of the training samples. You must state the dimensions of all the vectors and matrices involved. (c) How many eigenfaces are available for face recognition? Explain your answer. 3. Assume that 200 face images are available for training, and the face images are represented by Gabor features, extracted using Gabor filters of 5 center frequencies and 8 orientations. The Gabor feature of the i-th training image is denoted as fi, where i = 1, ..., 200. Assume that principal component analysis is applied to the training images, and the computer used can compute the eigenvectors and eigenvalues of a matrix with dimensions up to 1,000x1,000. (a) What is the dimension of the feature vector fi? (b) Describe how to compute the principal components of the feature vectors of the training samples. You must state the dimensions of all the vectors and matrices involved. (c) How many eigenfaces are available for face recognition? Explain your