A set of 17 grayscale face images (in JPG format) from 9 persons are contained in the attached .ZIP file. Each image is of dimension 195 x 231 (width x height) pixels and each pixel uses 8 bits for grayscale. Use the following 8 images as training images:
- subject01.normal,
- subject02.normal,
- subject03.normal,
- subject07.normal,
- subject10.normal,
- subject11.normal,
- subject14.normal,
- subject15.normal.
Use all 17 images (including the 8 training images) in the dataset as test images.
Non-face image: A non-face image (apple1) in JPG format is attached. This image is also of dimension 195 x 231. In addition to the 17 face images, test your algorithm on this non-face image.
Note: In the lecture slides, the training and test images are of size N X N (same horizontal and vertical dimensions.) The face images in the above dataset have different horizontal and vertical dimensions, but the Eigenface method works the same way and with no changes to the formulas.