Advanced Topics in Computer Science: 
Geometric Modeling for Computer Graphics

Thomas Funkhouser

Department of Computer Science
Princeton University

General Information | Syllabus | Coursework  | Students  | Links

General Information


The objective of this course is to investigate methods for automatic analysis of 3D data.  This objective is motivated by recent developments which have combined to accelerate the proliferation of 3D models:
  1. Laser range-finders are making acquisition of detailed 3D models practical.

  2. Many scanned 3D models are already publically available, and repositories of interesting 3D objects are just coming on-line.
  3. The World Wide Web is enabling access to 3D models constructed by people all over the world.

  4. Currently, several web sites allow free downloads for thousands of publically available 3D models.
  5. Graphics hardware and CPUs are becoming faster and cheaper at an astounding rate, causing an increasing demand for 3D models.

  6. It is now possible to buy a PC-based graphics accelerator capable of displaying a million textured polygons per second for just a few hundred dollars.
These developments are changing the way we think about 3D data.  Today, the primary challenge of 3D modeling is how to synthesize computer-based descriptions for interesting objects.  In the future, when 3D models are ubiquitous, the interesting research problems will become how to search for 3D models and how to analyze them.  Research in retrieval, recognition, classification of 3D models will follow the same trends as can already be observed for text,
images, audio, video, and other media.

The focus of the course is to investigate various representations of shape and how different representations can be used for analysis and comparison of 3D objects .  For each potential represenation, we will consider: "how can we form a model from acquired 3D data (range images, unorganized sets of polygons, voxels, etc.)?" and "how can we use the model for reconstruction, simplification, recognition, retrieval, and classification of 3D objects?"  Potential applications for the proposed methods include computer-aided design, medicine, electronic commerce, entertainment, and education.