Rosenfield's article was amazingly readable, honestly the first since I've
been at Georgia Tech that I didn't fall asleep from or daydream through. 
The article was based on image recognition of the face, tying in how the
futuristic Hal from 2001 might have "seen". Images are analyzed to create
boundaries of the face that separate different components. These edges can
then be analyzed for facial recognition. They can also be used to
determine emotion based on certain components being offset from an
emotionless face. Over time these images can be used to plot changing
emotions. I hope the rest of the papers are this interesting and easy to
follow.