July 3-5, 1996
LAAS-CNRS, Toulouse, France
Center for Biological and Computational Learning
Massachusetts Institute of Technology, Cambridge
A New Image Representation for Visual Learning in Computer Graphics
and Computer Vision
The last few years have seen new successfull approaches to object
recognition and to computer graphics based directly on images without
the use of intermediate 3D models. I will show that most of these
techniques depend on a representation of images that induces a linear
vector space structure and in principle requires dense correspondence.
This image representation allows the use of learning techniques for
the analysis and for the synthesis of images, that is for both
computer vision and computer graphics.
The key assumption hidden in most view-based approaches to object
recognition and object detection is that the relevant images are
vectors. This is not true unless they are set in correspondence. I
will review how this representation can be used to learn
In particular, I will describe techniques that represent a somewhat
unorthodox approach to computer graphics and computer animation.
- to estimate parameters such as expression and pose from images.
- to interpolate in multiple dimensions new images and images sequences
- to extrapolate from single images and generate