We are studying shape representations, and their application to object recognition. In particular, we are exploring the use of "shape contexts" to recognize objects based on their overall shape.

For more information on using our shape matching techniques to break EZ-Gimpy, a visual CAPTCHA, please follow this link.