Lecture: 3 hours
Discussion: 1 hour
Prerequisite: Courses 275A, 277
Grading: Letter; projects (70%), homework (30%)
Catalog Description:
A number of technologies allow us to capture information about real objects
in the form of three-dimensional point clouds. Processing this data to
create high-quality models for computer graphics presents many interesting
challenges. This course reviews recent research in this area and provides
background on useful techniques.
Goals:
I. Prepare students for research in model capture and analysis
II. Practice research techniques, including reading,
theoretical understanding, and experimentation
III. Learn basic algorithmic, mathematical, and
programming techniques relevant to model capture and analysis and other
problems in computer graphics and computer vision.
Expanded Course Descriptions:
I. Capture technologies: stereo, laser range scanner, structured light
II. Alignment: 2D image alignment, Horn's algorithm, ICP, global alignment
III. Surface reconstruction: Zero-set models, Voronoi methods, point-set surfaces
IV. Analysis: Feature detection and matching
Textbook:
There is no appropriate textbook for this topic. We will read research
papers, selected background material, and technical materials related
to specific hardware and software.
Project:
Students will work on projects individually or in pairs. Projects will
be chosen by the students in consultation with the instructor. Students
will have the opportunity to use simple data capture hardware to create
original models.
Instructor: N. Amenta
Prepared by: N. Amenta (October 2003)
Overlap Statement:
There is no significant overlap with other course
offerings.