University of California, Berkeley

Electrical Engineering and Computer Sciences Department

Announcements | Readings | Lecture Notes | Resources | Homework | Course Handouts
 

EE290T, Fall 2011
Advanced Topics in Signal Processing:
3D image processing and Computer Vision

Tuesday, 9:00 am - 11:00 am
212 Cory Hall

Prerequisite:  Signals and Systems at the level of EE120, basic linear algebra.

Course Announcement

Academic Dishonesty Policy

Grading:
35 % class participation and paper presentation
65 % project: Proposals for project due on October 4th

Lecturer:
Professor Avideh Zakhor
507 Cory Hall
Phone: (510) 643-6777
avz@eecs

Office Hours:
Tuesday, 11:00 - 12:00 noon, 507 Cory Hall

Course Administrative Assistant:
Lea Barker
(510) 642-2384
leab@eecs

Texts:

(a) ″Computer Vision: Algorithms and Applications″ by Richard Szeliski; Springer; available online at: http://szeliski.org/Book/

(b) ″Multiple View Geometry in Computer Vision″ by Richard Hartley and Andrew Zisserman, second edition, Cambridge Press (Look up electronically at OskiCat)

(c) Recommended: ″Image Based Modeling″ by Long Quan, Springer

(d) Tutorial Notes by Marc Polleyfeys on ″Visual 3D modeling from images″    Webpage    PDF

Announcements:

  • 11/15/11
    Homework 3 and the files you need to complete it are now posted under Homework.

  • 10/11/11
    Suggestions for class projects are now available under Course Handouts.
    All readings have been combined under Readings

  • 10/4/11
    The slides for the Sept. 13th lectures are now available under Lecture Notes.
    Photofly has been added to the Resources

  • 9/6/11
    There will be a lecture on Sept. 13th, after all. Dr. Allen Yang will talk about 3D reconstruction from a single image, and Rick Garcia (my PhD student) will talk about dynamic 3D scene reconstruction using structured light systems.

  • 8/30/11
    No lecture on Sept. 13th. Make up lecture to be announced shortly.

  • Welcome to EE290T!

Readings

  • Chapter 2.1 of Rick Szeleski's book

  • ″The 3D Model Acquisition Pipeline,″ by Fausto Bernardini and Holly Rushmeier, Computer Graphics Forum V. 21, No. 2, pp149-172
    PDF

  • ″3D Challenges and a Non-In-Depth Overview of Recent Progress,″ by Luc Van Gool, Bastian Leibe, Pascal Muller, Maarten Vergauwen and Thibaut Weise, 3DIM 2007.
    PDF

  • ″A Flexible New Technique for Camera Calibration,″ by Zhengyou Zhang, IEEE Transactions on Pattern Analysis and Machine Intelligence, V. 22, No. 11, pp 1330-1334, 2000.
    PDF

  • ″Single View Metrology,″ by A. Criminisi, I. Reid, and A. Zisserman, International Journal of Computer Vision 40(2), 123-148, 2000.
    PDF

Back to top

Lectures:

  1. Aug. 30, 2011
    Introduction to 3D computer vision 
    3D indoor modeling using lasers and cameras

  2. Sept. 6, 2011
    Introduction to 2D projective geometry and hierarchy of transformations
    Notes on projective geometry
    Handwritten notes on projective geometry
    Projections and Camera Calibration

  3. Sept. 13, 2011
    Distributed Sensing and Perception via Sparse Representation Allen Y. Yang
    Introduction to Structured Light (SL) and SL Based Phase Unwrapping Rick Garcia

  4. Sept. 20, 2011
    Recovery of metric and affine properties, 3D projective geometry

  5. Oct. 4, 2011
    Polleyfeys camera model
    Torralba Camera Model

  6. Oct. 7, 2011 (make up class for 9/27)
    Parameter Estimation
    Parameter Estimation, Part 2

  7. Oct. 11, 2011
    Camera Calibration

  8. Oct. 18, 2011
    Single-View Geometry

  9. Oct. 25, 2011
    3D Perception
    Vanishing Points and Line

  10. Nov. 1, 2011
    Epipolar Geometry

  11. Nov. 8, 2011
    3D Reconstruction

  12. Nov. 15, 2011
    Computing F

  13. Nov. 22, 2011
    Feature Detection

  14. Nov. 29, 2011
    Triangulation

Back to top

Resources:

Audio for the EE290T lectures, Fall 2009
Photofly
Point cloud library: open source library with code for processing point clouds
Open CV
Photo-synth from Microsoft
Bundler software from Noah Snavely
2011 CVPR proceedings (password protected)

Homework:

  • HWK 1: Take your camera to any scene , indoors or outdoors, at Berkeley, and take a few images by just rotating it, i.e. no translation. Then try to stitch the images together using some of the techniques we discussed in class.

  • HWK 2: Take your camera to an outdoor scene in Berkeley and take a few images of a building or an object. Then use photofly software from Autodesk to reconstruct the 3D geometry of the scene. Try different scenarios in order to find out where the algorithm excels and where it fails. Suggest a few ways of fixing its shortcomings.
    Due:  Nov. 15, 2011

  • HWK 3:     Problem set    Zip files
    Due:  Nov. 29, 2011

Handouts:

Back to top
 Last updated 11/28/2011 by LB