After successful completion of the course, students are able to identify key video processing techniques such as filtering, optical flow, and stereo matching, to describe how they work, and to assess their relevance / applicability to a given problem. In addition, students can evaluate selected video processing algorithms in consideration of the state-of-the-art and compare them with reference results from relevant benchmarks.
This course deals with selected problems of video processing, incuding the basics of video recording and imaging geometry, the estimation of movement (optical flow), filtering techniques, methods of video stabilization and the field of application of 3D video/film production and stereoscopic display.
In an introductory part of the course, we review principles of video recording, image generation and project geometry, whereby important basic terms of image processing are briefly recapitulated in order to support students with only little previous knowledge in the field. In the main part of the lecture, essential video processing techniques such as optical flow methods, filtering techniques (e.g. bilateral filters, guided image filters) and associated possible applications (e.g. cost volume filtering), as well as methods of video stabilization are presented. Furthermore, benchmarks, which are made available to the research community for evaluating algorithms, are explained and demonstrated.
The lecture slides will be made available online (on TUWEL) after each lecture unit. The lecture is accompanied by exercises in which selected problems are implemented in small groups.
The lecture slides will be provided on TUWEL and are supplemented by suitable video material, demos, and weblinks, in order to illustrate the lecture content and current research questions.
!UPDATE (07.11): The lecture will begin a week later than originally planned, on 16 November (INSTEAD of 9 November)
IMPORTANT: The German title of this course was changed from VO Videoverarbeitung to VO Video Analyse. The English title remains unchanged: VO Video Analysis.
The prelecture meeting takes place in FAV Hörsaal 1 - INF on October 5th, 2022, at 11 am. Time and place differ from the usual lectures, which are held at 10 am FAV Hörsaal 2. 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 9th, Nov 16th, Nov 23rd, Nov 30th, Dec 07th, Dec 14th, Dec 21st (The dates in january are only substitution dates, if necessary).
All lecture dates will be held on campus.
If necessary, lectures can be held online at short notice.
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 exam preparation.
The exam serves to assess how well the processes and algorithms presented in the lecture have been understood. You will be asked to give your answers as written explanations, formulas, sketches, pseudo-code etc. The final grade for the practical course consists of the grades you receive for training exercises that you have to hand in and a final conversation.
The exam at the end of the semester will be a written closed-book exam and will take place online via Zoom. To participate in the exam you therefore need to have access to the internet, a computer with a working microphone and a webcam.
The lecture slides and any additional materials are provided for download via TUWEL.
The course requires basic knowledge in image processing, for example as taught in related courses of the Bachelor's Program Media Informatics and Visual Computing (e.g. course VU 186.822).