3D Computer Vision

Goals:

The course is designed to develop the skills needed to generate synthetic images of 3D objects and to recover 3D structure from one ore more views (projections) of 3D objects. The objective is to develop software to recognize 3D features in one or more 2D views (images) of a scene based either on actual photographs or synthetic images (computer graphics generated) and to use this software to adapt 3D models to correspond to actual scenes.

Text:

None, but see references.

Topics:

Data structures for 3D objects
Camera models and projective geometry
3D Computer Graphics using OpenGL
VRML (Virtual Reality Modeling Language)
Camera calibration
Image feature identification
Shape recovery in multiple view geometries
Object tracking, Optical flow

Computer Usage:

Students will use MATLAB extensively.

Prerequisites:

Ideally students should be familiar with object-oriented programming (ECE 538), MATLAB, and image processing (ECE 563). However, accommodations will be made for students with experience in only one of these areas.

Target Population:

This course is intended to support research in robotic vision and could be of interest to students in Computer Engineering, Mechanical Engineering, Electro-Optics, and Computer Science.


Maintained by John Loomis, last updated 18 Dec 2013