Computer Vision I: Variational Methods

What it is about

You learn how many tasks of image processing (e.g. denoising, desmearing, segmentation, optical flow, stereo depth estimation, 3D reconstruction) are formulated and solved as variational problems. In the end you know the Euler–Lagrange approach and PDEs, efficient solution methods as well as convex formulations and relaxations and you can implement central concepts in Matlab.

What you will be able to do

• Understand the fundamentals of variational methods
• Apply Euler–Lagrange formulations and the associated PDEs
• Formulate computer vision tasks as variational problems
• Apply efficient numerical solution methods
• Assess convex formulations and relaxations
• Implement concepts in Matlab

What the module consists of

• VorlesungPresentation of the main concepts
• Tutorium / ÜbungDeepening through exercises, discussions and programming tasks

Teaching method

• VorlesungIntroduction of the theoretical foundations
• TutoriumsübungenDeepening, interactive problem solving and programming practice