Stefan Karlsson/PersonalPage/Education/MultiScaleCourse
This is the web page for the course Multiscale and Multidimensional Analysis. The course description will be available on the official university site soon. The schedule for the course will be made ad hoc, using doodle or by agreement over email.
Reach us at Stefan.Karlsson(AHTT)hh.se or feralo(AHTT)hh.se for any questions.
News
No news yet
Course organization
There will be roughly one new exercise made available to you for every session, and there will be a total of 4 exercises including a project. For each exercise there are several tasks for you to perform. These are to be completed and discussed during the sessions. Instructions on how to report on them will be given in the descriptions for the exercises. Communication WILL NOT be done through blackboard.
Exercises and sessions (files will be made available before each session)
1. Linear scale space (Stefan): 1 WEEK
- Waveletets (Gabor, Mexican Hat)
- Pyramids (Gaussian, Laplacian, etc.)
- Median filtering
2. Directionality analysis (Fernando): 1 WEEK
- Structure tensor, HOGs, Gabor
- Edges, corners
3. Non-linear scale-space (Stefan): 1.5 WEEK
- Variational formulations
- Non-linear filtering
- De-noising
→ PRACTICAL ASSIGNMENT
4. Feature analysis (Fernando): 1.5 WEEK
- Segmentation, clustering
- Feature extraction, pattern matching and classification
→ PRACTICAL ASSIGNMENT
5. Applications for computer vision/object detection (Fernando, Stefan): 1 WEEK
- SIFT
- Viola-Jones
- Perona-Malik
→ FINAL PROJECT ASSIGNMENT
