Computer Vision and Photogrammetry
Course description:
In this course will look at different techniques to acquire 3D models from images. During the course labs you will implement a 3D reconstruction pipeline, from 2D images to a 3D geometric model.
- Camera geometry and calibration
- Feature detection and Matching
- Stereo and Structure from Motion
- Registration, Meshing and Texturing
Resources
- http://www.cs.unc.edu/~marc/tutorial/
- http://www.cs.unc.edu/~marc/tutorial04/
- http://www.robots.ox.ac.uk/~vgg/hzbook/
Tentative Syllabus 2016-2017
| Class Date | Contents | Details | Lecture notes | Lecture notes PDF | Lab |
|---|---|---|---|---|---|
| 4 October, Tue | Introduction and image formation | Photogrammetry Overview and Projection Matrix. Details about the course. | Lecture 1 | Lecture 1 | Lab0: Install Python (Anaconda) and get familiar with image processing libraries |
| 11 October, Tue | Image formation, camera calibration, pose estimation | Camera matrix, calibration with linear methods | Lecture 2 | Lecture 2 | |
| 18 October, Tue | Homographies and Epipolar geometry | Homography calculation, epipolar constraint, Essential and Fundamental Matrix | Lecture 3 | Lecture 3 | Lab 1 |
| 25 October, Tue | Two View Geometry | Epipolar geometry, Computing F and E. Triangulation | Lecture 4 | Lecture 4 | Lab 2 |
| 8 Novemeber, Tue | Stereo | Problems of stereo matching. Image rectification. | Lecture 5 | Lecture 5 | Lab 3 |
| 22 Novemeber, Tue | Keypoint detectors | Harris, LoG, DoG, MESR | Lecture 6 | Lecture 6 | Lab 4 |
| 29 November, Tue | Feature Descriptors and Matching | SIFT, NNDR, precision, recall, accuracy. | Lecture 7 | Lecture 7 | Lab 5 |
| 6 December, Tue | Self-Calibration | Projective Geometry, Duality, self-calibration | Lecture 8 | Lecture 8 | Lab 5+ |
| 13 December, Tue | Reconstruction from Sequences | Projective Reconstruction, Sequencial recontruction | Lecture 9 | Lecture 9 | Marking Lab 5 |
| 20 December, Tue | Bundle adjustment | Bundle adjustment, sparsity of the bundle adjustment problem. | Lecture 10 | Lecture 10 | Marking Delayed Labs |
| 10 January, Tue | Meshing Pointclouds | Signed Distance Functions, Normal Computation, Radial Basis Functions, Poisson Reconstruction | Lecture 9 | Lecture 11 | Lab 6 |
| 17 January, Tue | Texturing | Multi-view Texturing, Parameterization, Least-squares conformal Maps | Lecture 12 | Lecture 12 | |
| 24 January, Tue | SLAM | Lecture 13 | Lecture 13 | Marking Lab 6 | |
| 31 January, Tue | Review & Closure | Recapitualtion, Feedback, Closure | Lecture 14 | Lecture 14 |
Requirements
Image Processing is advisable. Linear algebra.
Examination
There will be no exam. Your mark will depend on your implementation and documentation of the 3D reconstruction system.
Similar Courses at other Universities
in progress…