CENG 608

3D Photography

This course covers algorithms and applications to extract 3D information (especially shape) from images. It starts with the camera model and calibration, 2D and 3D projective geometries and extracting feature points. Then it covers passive 3D reconstruction techniques such as single view reconstruction, structure from motion, shape from silhouettes. Active sensing techniques (time of flight cameras, structured light, laser scanners etc.) that directly obtain 3D data are also briefly covered.

Course Objectives

To teach the imaging geometry of cameras. To describe the algorithms used in 3D information and shape extraction and to improve ability to implement these algorithms. To form acquaintance with the applications of this research area. To improve the ability of written and oral presentation.

Recommended or Required Reading

Computer Vision: Algorithms and Applications, Richard Szeliski, Springer, 2010.

Learning Outcomes:

1. Describe the fundamental concepts in imaging geometry

2. Identify state-of-the-art techniques used in 3D information and shape extraction from images

3. Construct programs that implement algorithms of 3D information extraction and 3D reconstruction

4. Prepare written and oral presentation for the performed work

 

Topics
Camera models and calibration
Feature point extraction and matching
Feature point tracking and optical flow
2D projective geometry
RANSAC and image alignment using feature points (image warping)
3D projective geometry and epipolar geometry
Two-view and multi-view structure-from-motion
3D simultaneous localization and mapping
Depth extraction with stereo matching
Image segmentation
Shape-from-X
3D reconstruction
Active sensing techniques

 

Grading

Midterm 25%

Homework 15%

Presentation 30%

Final 30%