CPS 196.1
Introduction to Computer Vision
Spring 2004

Course Outline

This course explores some of the most successful ideas in computer vision at an introductory level. Basic linear algebra and probability are prerequisites, but the class will be essentially self-contained. For a choice of the following topics (depending on time and interest), we will cover the basic concepts, look at one or two algorithms that work well, listed below, and discuss limitations and possible extensions.

• Image processing
  • Convolution and smoothing
  • Choice of median filtering or the Manduchi-Tomasi bilateral filter
• Edge detection
  • Derivatives of image intensity
  • Canny's edge detector
• Tracking points in video
  • Moravec's interest operators
  • The Lucas-Kanade tracker
  • The Shi-Tomasi feature detector
• Motion segmentation
  • Weiss-Adelson layers
• Stereo vision
  • Sum of squared differences
  • The Woodfill-Zabih stereo matcher
  • Dynamic-programming stereo
• 3D geometry from video
  • Tomasi-Kanade factorization
• Recognition
  • Turk-Pentland eigenfaces

Back to the CPS196.1 Home Page