FICS Course on Computational Imaging Algorithms for Light-Field Acquisition, Processing and Visualization

Organized by the Finnish Doctoral Programme in Computational Sciences, the PROLIGHT Marie Curie IAPP Action and 3D Media Group at the Department of Signal Processing, Tampere University of Technology

November 27-29, 2013 at Tampere University of Technology

The course addresses computational imaging aspects of sensing 3D visual scenes by multi-camera setups, their light-field and multi-view multi-depth representations and related processing, and their visualization through modern 3D displays. More specifically, the course addresses the problems of multi-camera rectification and disparity estimation, as well as scene representations ranging from pure image-based, such as Epipolar-Plane Image (EPI) based, to multi-view multi-depth based. Ray interpolation and view rendering techniques are discussed along with production rules for generating correct stereoscopic 3D and techniques for generating visual effects out of multi-camera data. Principles of operation of 3D displays are reviewed and Fourier domain analysis of such displays is presented.

Instructors: Frederik Zilly, Fraunhofer IIS, Germany and Dr. Robert Bregovic, Holografika, Hungary and Tampere University of Technology, Finland

Schedule:

Wednesday, 27 November
Room: TB 224

12:00 – 12:15 Opening  the course (Atanas Gotchev)

12:15 – 14:00 Lecture 1: Multi-camera 3D scene sensing (Frederik Zilly).
–  Feature-Point-Based Multi-Camera Rectification
–  Multi-Camera Disparity Estimation

14:00 – 14:15 Coffee break

14:15 – 16:00 Lecture 2: Light field representations (Robert Bregovic)
– Epipolar-plane image representation
– Light field analysis in Fourier domain

Thursday, 28 November
Room: TB 216

09:15 – 11:00 Lecture 3: Processing of multi-camera data (Frederik Zilly)
– View Rendering / Depth Image Based Rendering
– Visual Effects using Multi-Camera Data: Matrix-Effect, Vertigo-Effect, Synthetic Aperture

11:00 – 12:45 Lunch break

12:45 – 14:15 Lecture 4: 3D displays (Robert Bregovic)
– 3D displays: principles of operation
– Analysis of 3D displays in Fourier domain;

Friday, 29 November
Room: TB 216

09:15 – 11:00 Lecture 5: High-quality Stereoscopic 3D (Frederik Zilly)
– Production Rules for visually pleasant Stereoscopic 3D;

11:00 – 11:15 Instructions for writing the reports and closing the school (Atanas Gotchev)

Credits and grading

2-4 cr; pass/fail

2cr requirements: Attendance, literature survey and four-six pages report on a particular topic agreed with the instructors.

1-2 additional credits:  A more extensive study on a particular algorithm agreed with the instructor, including a software implementation (Matlab or C); presentation and a final report with test results.

Administration

Registration: Susanna Anttila  susanna.anttila@tut.fi
Send an email to the department secretary Ms Susanna Anttila with the following information:
– Name, affiliation, student number
– Are you a FICS student?
– Do you have any dietary requirements?
Registration deadline: 25.11.2013

Additional information: Atanas Gotchev atanas.gotchev@tut.fi (course organizer) and Ella Bingham ella.bingham@aalto.fi (FICS coordinator)

FICS students will have preference in registration if participation needs to be limited and non-FICS students will be admitted in the order of their registration request.

Bios of course instructors:

Frederik Zilly is Head of the group Computational Imaging and Algorithms in the Moving Picture Department of Fraunhofer IIS in Erlangen, Germany. He has received the Diploma degree in physics from Free University of Berlin, Germany, in 2006. Before joining Fraunhofer IIS, he was Scientific Project Manager in the Immersive Media & 3-D Video Group in the Image Processing Department at Fraunhofer HHI in Berlin where he was involved in several German and European research projects related to 3DTV. In this function he was mainly responsible for the development of the Stereoscopic Analyzer (STAN) and coordinated the activities of the Fraunhofer HHI concerning the European research project MUSCADE project. Frederik Zilly has been honored for his work on the assistance system STAN (Stereoscopic Analyzer) with the Award for Outstanding Merit in Young Scientists, the Rudolf Urtel Prize 2011. He served as reviewer for different international conferences and journals. His research focus lies on multi-camera image processing algorithms.

Dr. Robert Bregovic is Visiting Senior Researcher at Holografika, Hungary. He has graduated as Doctor of Technology (with honours) at Tampere University of Technology, Finland in 2003 with a work on optimal design of perfect-reconstruction and nearly perfect-reconstruction multirate filter banks. He has been Senior Researcher at the Department of Signal Processing, Tampere University of Technology (2003-2015), Visiting Research Fellow at Temasek Laboratories, Nanyang Technological University, Singapore (2005-2006), and Research Fellow at TUT, before going for a research secondment to Holografika, Hungary within the PROLIGHT-IAPP project. His research interests are in the Fourier domain signal analysis, design of filters and filter banks and light field sensing, processing and display. He has co-presented a tutorial on signal processing methods for 3D displays at the recent IEEE Int. Conference on Multimedia and Expo (July 2013, San Jose, CA).

First PROLIGHT Workshop on Modern Signal Processing Methods for Ultra-realistic 3D Displays

Organized jointly by the PROLIGHT-IAPP Marie Curie Action and the 3D-ConTourNet COST Action

 October 29-30, 2013, Nantes France

Workshop chairs: Atanas Gotchev and Jaakko Astola, Tampere University of Technology, Finland

Workshop synopsis

The ultimate aim of visual media technologies is to achieve higher realism of the reproduced visual scenes in order to involve the user in the story being told or to help him/her to accomplish some task in a more productive way. Ultra-realistic 3D displays are expected to recreate an exact optical replica of the visual scene, therefore they constantly require capture and processing of vast amount of multi-modal data. Such data is available only for computer-generated scenes, while the creation and manipulation of real-life scenes is still problematic. In order to fill the gap between creation and manipulation of rich visual content and its highly-realistic visualization, there is a need to understand the complexity of the data and how it is constrained by the available capture equipment and the target display.

The PROLIGHT workshop addresses the problems of capture, analysis, modelling, compression and visualization of light fields of real-world scenes. It gathers together researchers aiming at advancing the theory of light-field modelling and its practical application on next-generation 3D displays. The workshop is a venue for presenting novel methods and techniques based on modern signal processing theory made operational for modern 3D displays.

Workshop agenda

Tuesday, 29 October 2013

11:30 – 11:35 Opening, Atanas Gotchev

11:35- 12:00 Latest development in 3D content representation and processing, Robert Bregovic, Tampere University of Technology, Finland and Holografika, Hungary

12:00 – 12:25 Lightfield media production, Michael Schöberl, Fraunhofer IIS, Germany

12:25 – 12:50 Content creation challenges for 3D light-field displays; Peter Kovacs, Holografika, Hungary and TUT, Finland

12:50 – 13:10 Plenoptic function reconstruction from non-uniform distributed samples, Suren Vagharshakyan, Tampere University of Technology, Finland

13:10 – 13:30 Design of high-speed full-parallax 3D visualization test setup for subjective experiments, Atanas Boev, Tampere University of Technology, Finland

Wednesday, 30 October 2013

10:30 – 10:55 QoE for 3D media, Patrick Le Callet, University of Nantes, France

10:55 – 11:30 Imaging and display of 3D micro and macro scenes, M. Martinez-Corral, H. Navarro, A. Llavador, A Dorado, and G. Saavedra, University of Valentia, Spain

11:30 – 11:45 A Novel Scene Representation For Digital Media, Christopher Haccius, Intel Visual Computing Institute, Germany

11:45 – 12:10 Compression of light field images, Yun Li, Roger Olsson, Mårten Sjöström, and Ulf Jennehag, Mid Sweden University, Sweden

12:10 – 12:35 Modeling of plenoptic capture systems, Mitra Damghanian, Roger Olsson, and Mårten Sjöström, Mid Sweden University, Sweden

Atanas Boev, Robert Bregovic, and Atanas Gotchev presented a tutorial at ICME 2013. The title of the tutorial was: Signal Processing Methods for Stereoscopic and Multi-View 3D Displays. Abstract of the tutorial can be found here.

For more details about the tutorial, please contact one of the presenters.