The goal of this course is to make the students acquainted with selected research topics in the field of video analysis and related applications. The addressed research topics include efficient video filtering, interactive object segmentation, motion analysis (optical flow), 2D-to-3D conversion, and 3D video capturing, video representation and view synthesis for 3D video/TV applications.
In this course, we deal with selected topics in the area of video processing, including motion estimation (Optical Flow) and other label-based problems (e.g., video object segmentation), high-speed video recording, and 3D video/TV (2D-to-3D conversion, 3D formats and displays, 3D video production).
In an introductory part of the course, we briefly review principles of image processing, in order to support students with only little background in the field. Based on introduced methods for efficient video filtering, the main part of the course discusses an approach to solve label-based problems (Cost-Volume Filtering), such as interactive video object segmentation. The course further focuses on both theoretical foundations (e.g., optical flow equation) and practical applications/current developments in the field of video analysis (especially 3D video/TV). The course contents are conveyed using a variety of image/video materials and demos. Furthermore, some selected research topics are discussed.
In other words, the course addresses the following questions:
- How is an image filter extended to a video filter?
- How to extract an object from a video and paste it onto another background video. How to deal with semi-transparent pixels at borders?
- How can one detect and estimate motion in a video?
- How are 3D movies generated?
- Is it possible to convert a 2D movie into a 3D movie?
- How does a 3D display work?
- How are additional views for 3D generated?
The lecture slides will be provided on the web (via TUWEL) after each course unit. The lecture is accompanied by a lab exercise, in which students will work in small groups on the definition and implementation of selected problems.
The prelecture meeting takes place on October 8th, 2018, at 10 am in Zemanek seminar room (Favoritenstraße 9-11). Please note that in order to successfully complete the excercise part of the lecture, it is mandatory to attend the prelecture meeting. The dates of further lectures are going to be as follows: Nov 5th, 12th, 19th, 26th, Dec 3rd, 10th, and 17th (Jan 7th and 14th and 20th, scheduled as a substitution date)
- This course represents 1.5 ECTS (= 38 hours). An estimate of 14 hours of these will be used for the actual lecture and the remaining 24 hours for in-class pre-exam preparation.
The course requires basic knowledge in image processing, for example, as taught in related courses of the bachelor program Media Informatics (e.g. course VL 183.126). Some important principles of image processing are briefly reviewed in the introductory part of the lecture, as an additional support for students with only little previous knowledge.
