CS 482
Computer Vision

Time/Location: Monday 4:30-7:10,   Robinson B203
Instructor: Dr. Jana Kosecka
Office: 4444, Research II
email: kosecka@cs.gmu.edu
Course website http://cs.gmu.edu/~kosecka/cs482/

This course will cover essentials of computer vision. We will learn basic principles of image formation, image processing algorithms and different algorithms for 3D reconstruction and recognition from single or multiple images (video). Apllications to 3D modelling, video analysis, video surveillance, object recognition and vision based control will be discussed.  
This course is of interest to anyone seeking to process images or camera information, or to acquire a general background in issues related to real-world perception, image processing, object and scene recognition and multi-view geometry

Sample Syllabus:
    1. Representation of  3-D scenes : rigid body motion, euclidean, affine and projective transformations.
    2. Image formation: geometric and photometric aspects of image formation process, binary, grey level and color images
    3. Image features and Correspondence:  geometric and photometric features, feature detection and matching, optical flow
    4. Stereo - Two view geometry : camera pose and 3D structure recovery from two views, camera calibration, 3-D reconstruction
    6. Image Matching and Tracking : matching of multiple views, tracking and video analysis
    8. Grouping and Segmentation : detection and recovery of multiple motions
    9. Detection and Recognition of 0bjects in Images: object representations and classification methods
   10. Selected topics: vision based control, image based rendering pipeline, vision for human computer intraction, recognition

Grading: Homeworks (about every 2 weeks) 40% Midterm: 30% Final project: 30%
Prerequisites: linear algebra, calculus
Lecture Materials:  
Lecture slides, lecture notes provided by instructor

Recommended Textbooks:
Computer Vision: Stockman and Shapiro, Prentice Hall.
Invitation to 3D Vision:  From Images to Geometric Models: Y. Ma, S. Soatto, J. Kosecka and S. Sastry (for part I of the course)
Introductory Techniques for 3D computer Vision. E. Trucco and A. Verri, Prentice-Hall, 1998
Computer Vision: A Modern Approach: D. Forsythe and J. Ponce, Prentice-Hall, 2003

Required Software:
Matlab, OpenCV. Homeworks will require using Matlab and OpenCV. You can buy a student version in Johnson center or use it remotely from ITE labs. OpenCV is an C/C++ open source computer vision library.

Outcome: Students will obtain basic understanding of images formation process, processing of digital images and video and will gain familiarity with different algorithms for 3D reconstruction and recognition of objects in images. In this context of these computer vision problems, the students will use elemetary geometry, linear algebra, probabilistic inferenec and basic machine learning and pattern recognition techniques. Students will obtain capabilities for implementing learned algorithms in C/C++ and Matlab.