1- AT&T database is in the net, and it has very large number of face images. I need a database taken from AT&T database, it should consists of 10 persons and each person has 10 images, total 100 images. For example the database consists of folder like person 1 and this folder has 10 face images of (person 1), 6 for training and 4 for testing.
2- The system should be built with one of these following techniques:
-Principal Component Analysis (PCA)
-Linear Discriminant Analysis (LDA)
-Elastic Bunch Graph Matching (EBGM)
3- The codes should be running after the implementation as follow:
-After the system is built in MAT LAB ,If I select a face image from the 100 images for example image 9 of (person 4) which is testing image and enter it to the system in MAT LAB, it should be able to identify the entered images and telling me that the entered face is for (person 4) by displaying the entered image and person name which is (person 4). And the opposite If I enter an image that is out of the 100 images such as I enter an image from the remaining images of AT&T database like image 1 of (person 15) the system should tell me that the entered image is not identified and it is out of my database.
4- The codes should include the codes for displaying the extracted features that are extracted from the training images, any features that the system selected to make the calcification between the people faces images and faces images of the same person from the 60 training images.
5- The required items: A folder consists of the MAT LAB file with the implemented codes, the used database, and a document that has the displaying of the extracted features from some training images, these feature can be presented by graphs or any types of data demonstrations.
6- The image should be with the same quality as it is available in the net.
7- It is not necessary that programming is novel.
7- The work should be submitted within 2 days.
[login to view URL]
this link is an example of how the system should work