Computer Vision

This is the course page for the computer vision course, for the spring semester, 1999-2000, being taught by Dr. Subhashis Banerjee at the Department of Computer Science and Engineering, IIT, New Delhi.

General Information
Papers
Documentation
Software
Vision Sites

Project List

General Information

• Monday, 15:00-16:30
• Thursday, 16:00-17:30

Assignment 1 on stereo is now available online.

Papers

1. Finding Point Corrspondences in Motion Sequences Preserving Affine Structure
   G. Sudhir, Subhashis Banerjee, Andrew Zisserman.
   Computer Vision and Image Understanding, Vol. 68, Nov. 1997
2. Object Tracking Using Affine Structure for Point Correspondences
   Gurmeet Singh Manku, Pankaj Jain, Amit Agarwal, Lalit Kumar, Subhashis Bannerjee.
   IEEE CVPR '97, June 19-21, San Juan.

Documentation

1. General
   
   • Projective Geometry Tutorial, by Mohr and Triggs
   • The Geometry of Multiple Views, by Andrew Zisserman.
     • Part 1: Epipolar Geometry and the Fundamental Matrix
     • Part 2: Uncalibrated 3D Reconstruction
     • Part 3: Planes
     • Part 4: Points, Lines and Curves
   • Mathematics for Computer Vision, by Carlo Tomasi
2. Affine and Projective Reconstruction
   
   • Affine Structure from Motion.
     Koenderink and van Doorn, Journal of the Optical Society of America, A8:377-385, 1991.
     (No electronic copy available. Paper copy may be obtained from the Central Library).
   • Projective Structure from Two Uncalibrated Images: Structure from Motion and Recognition
     Amnon Shashua, (A.I. Memo 1363)
   • Dr. Banerjee's notes on affine projection

Software

• Horatio: the vision library that solves all your problems, that is, if you like C, don't like C++, and want to make things run FAST.
  
  • Download Horatio 0.313 sources
  • Compiled libraries (Horatio, and dependencies) for RedHat 5.2 and RedHat 6.0, 6.1. All you have to do is `tar xzf` in /usr/local.
  • Online Manual
  • Sample program with Makefile. Please note that while compiling, the libF77.a library must be linked after the object file conatining main.
• MATLAB: the high-performance language for technical computing. We have the compilers, plus toolboxes for optimization, signal processing, simulink, state flow, wavelets, and control. The image processing toolbox is not available.
• MATLAB Helpdesk
• Learning MATLAB: a beginner's guide.
• Using MATLAB: MATLAB user's guide.
• Sample Programs Directory

Vision Sites

• The Computer Vision Homepage, CMU
• Compendium of Computer Vision A tremendous amount of information, including image processing, camera calibration, geometry, mathematics. If you find something especially useful, let me know, and I'll mirror it locally.
• Robotvis, the robotics and vision group at INRIA, Sophia Antipolis. There are some impressive demos, check out the one on Epipolar Geometry

Maintained by Vaibhav Vaish.
Last updated on 17^th Feb, 2000.