## CSC492 Face Recognition

• Ravinders site
• Ravinders report
• Look up some simple references on Face Recognition (Biometrics, eigen-faces approach to face recognition).
• on the web
• Find a paper/course/website devoted to face recognition for example
• in the research literature
• Find the name of a few journals that discuss Face Recognition regularly.
• Get a simple paper discussing a simple aspect of Face Recognition or a summary of the state of the art. look at the U of T library site including online journals.
• Pick a good reference on Eigen Faces and start studying it. Your goal is to be able to explain the intuition behing this approach. We will aim for underlying details later on. Let me know which you have picked, give me a list of relevent background material you will need to understand.
• A few references:
• Todo
• Understand covarance and least squares and connection to eigenfaces
• Understand intuition behind eigenfaces
• Understand how to reconstruct a face using the eigenfaces. First we must understand what an eigenface actually is.
• Learn OpenGL or Java2D api (we need to be able to grab pixels from an image gif, jpg)
• Open a window with a white background
• As above but now draw a blue pixel
• Load an image (jpg or gif)
• Find r,g,b for pixel at (100,100)
• Find good C/C++ linear algebra package (free/opensource)
• Week 06 For this week, can you have a Java image loader and a block of code that loads images, computes the average image and the difference images (Image_i-Average). This will be the first steps towards the Eigenface algorithm. Can you also write up some brief notes describing the algorithm (see section 2.1 of the Turk and Petland paper). This will become part of your report.
• Week 07
1. For this week, generalize the code you created so that it loads all images in a single directory. It should compute the average image as well as all image differences.
2. Investigate LAPack. You should know how to create a matrix, find its eigenvectors and eigen values.
3. Have a summary of the algorithm to create eigenfaces (read Calculating Eigenfaces in "Eigenfaces for Recognition).
4. At some point, we should try to determine how a simple face recognition algorithm (ie compare pixels) will compare with the eigenface approach. Write a few notes on a possible simple approach you would like to implement.
• Week 09
1. Summarize (in a report) the eigenface approach
2. Suggest a simple approach to face recognition and measure the eigenface approach against it (ie average greyness in a b/w inage).
3. Extend this approach to determine if a face is present in an image (independent of resolution, independent of orientation). Think about how we might do this.
• Week 13