Description:

The topics covered by this course include digital image formation, image features and feature detection, structure from X, image segmentation, object recognition, motion analysis, and stereopsis. Students will also work in groups to solve some real world problems. Some potential project topics are automatic object recognition, human face detection, video surveillance, and content-based image retrieval.

Syllabus:

1. Introduction
2. Image and video acquisition
3. Camera models
4. Image features - edge, corners, lines, Hough Transform, deformable contours
5. Camera calibration
6. Camera motion estimation and 3D scene reconstruction
7. Stereopsis
8. 2D motion analysis - optical flow estimation, differential techniques
9. Shape from X - reflection model, shape from shading, shape from texture, shape from defocusing and focusing
10. Tracking - Kalman filtering, correlation-based tracking, change-based tracking, 2D layer tracking, tracking of articulated objects
11. Image segmentation
12. Object recognition - Feature, invariants, subspace method, face detection and recognition

Reference:

Computer Vision -- A modern approach, by David Forsyth and Jean Ponce, Prentice Hall, 2002.

Time and Place:

TuTh   2:00 p.m. - 3:45 p.m.,   Baskin Engineering 372

Office hour:

Wednesday   3:00 p.m. - 4:00 p.m.,   Engineering 2, Room 333

Instructor:

Hai Tao (http://www.soe.ucsc.edu/~tao/)
Email:   tao@soe.ucsc.edu
Office:  Engineering 2,  Room 333

Evaluation: Coursework will be weighted as follows:

 Homework   25%
 Mid-term      30%
 Project          45%

Homework: Homework will be collected every next week on Thursday at the end of the class.

Homework Assignments: Solution

Lecture notes:

     Introduction,

     Image acquisition and camera model,

     An additional homework problem,

     Image noise and filtering,

     Image features: edges and corners,

     Hough transform: lines and curves,

     Model fitting and robust regression,

     Camera calibration,

     Epipolar geometry and the 8-point algorithm,

     Tomasi-Kanade factorization,

     Zhang's correspondence algorithm,

     Rectification and depth computation,

     Image motion analysis,

     Optical flow,

     Additional homework problems,

     Hidden Markov model (HMM),

     Object tracking and Kalman filtering,

     Object recognition - interpretation tree,

     Object recognition - appearance subspace,

     Image segmentation,

     Stereo algorithms (By Dan Kong),     

     Stereo algorithms (By Xiaoye Lu),

     Shape from shading (1),

     Shape from shading (2),

     Shape from texture,

Some Project Topics:

• 3D reconstruction and camera motion analysis

• Resolution enhancement

• Multiple view tracking

• Object tracking with occlusion

• Face detection

• Facial feature detection

• Facial motion analysis and event detection

• Image-based realistic facial animation

• Content-based image retrieval

• Image-based relighting

• A marker-driven pointing device

• Optical flow estimation

• Shape from brightness functions

Computer vision links:

• CMU Computer Vision Home Page
• CVonline: Computer Vision Online Course Notes
• USC Annotated Computer Vision Bibliography
• PEIPA - the Pilot European Image Proecssing Archive

Class Photo:

webmaster@cse.ucsc.edu

Back to the SOE Class Home Pages.
Back to the SOE Home Page.