Special Sessions

The following three special sessions have been approved by the MMSP Technical Committee. Interested authors are encouraged to contact the special session organizers for paper submission.

• Special Session 1: Search and Beyond Search: Online Video Applications
• Special Session 2: Immersive Tele Collaboration
• Special Session 3: Global Motion Estimation and Mosaicing for Applications in Video Analysis and Coding

Note for Special Session Authors

The Special Session (SS) papers will be submitted to the same submission portal by choosing relevant Special Session as the Main Topic. The deadline of the SS submission is May 2, 2008. However, the potential authors MUST register their interests by lodging a PDF format file of paper abstract into the system before or on the regular paper submission deadline to GET a paper submission ID. This is because that the on-line submission will be closed after the regular paper submission deadline, which has now been extended to May 2, 2008. For final SS paper submission, authors MUST send the final copy with submission ID to the SS organizer.

Please note that any submitted SS paper with no ID will not be accepted by MMSP08!

Thanks for your contribution to MMSP 2008

Organizers
Xian-Sheng Hua (Microsoft Research, Asia, xshua@microsoft.com)
Tao Mei, (Microsoft Research, Asia, tmei@microsoft.com)
Jesse Jin (The University of Newcastle, Australia, Jesse.Jin@newcastle.edu.au)
Lingyu Duan (I2R, lingyu@i2r.a-star.edu.sg)

　Scope:

1. Video search and video search re-ranking
2. Social video and video sharing
3. Video recommendation
4. Video advertising
5. Online video annotation/categorization/tagging
6. Online video presentation, browsing and management
7. Online video authoring
8. Other video related technologies and applications on the Internet

The core technologies for different online video applications are closely related. “Multimodal Relevance” is one of the key technology that many applications involve …

Organizers
Cha Zhang (Microsoft Research, chazhang@microsoft.com)
Zicheng Liu (Microsoft Research, zliu@microsoft.com)

Immersive tele-collaboration has attracted increasing attentions recently, thanks to the widening geographic distributions in large corporations, rapidly dropping hardware costs, increasing network bandwidth and rising computation power. It opens up a wide variety of new research topics and applications, such as multi-view video capturing/compression/ transmission/rendering, spatial audio capturing and rendering, distributed audio/video processing, remote data collaboration, immersive computer and user interaction, etc.

Even though some of these topics are part of the traditional MMSP research areas, immersive tele-collaboration has some unique characteristics, which pose stringent requirements on the multimedia signal processing techniques being used. For instance, many researchers have worked on building camera arrays for free viewpoint viewing of a 3D scene. Unfortunately most of the developed techniques are too computational intensive to be able to render 3D scenes in real time. Real-time compression and transmission of multiple video streams is also a very challenging problem. When spatial audio are considered, real-time multi-channel echo cancellation remains a very difficult task.

Immersive tele-collaboration may also broaden the traditional research area of MMSP. For instance, in remote surgery applications or remote microscopy applications, control information from hepatic devices need to be transmitted to the remote site in a much higher frequency than traditional audio and video. These control information may alsoexhibit special requirement of error protection or error concealment, which is not traditionally considered in other multimedia applications.

The purpose of this special session is to bring together leading experts from various disciplines to exchange their work and ideas on multimedia signal processing techniques for immersive tele-collaboration. We notice that the main theme of MMSP 2009 will be “Multimedia for immersive telecommunication and human collaboration”. This special session will serve as a great prelude to gain momentum on this topic toward MMSP 2009.

Organizers
Thomas Sikora (Technische Universität Berlin, sikora@nue.tu-berlin.de)
Andreas Krutz (Technische Universität Berlin, krutz@nue.tu-berlin.de)
Martin Haller (Technische Universität Berlin, haller@nue.tu-berlin.de)

Global motion estimation in 2D-video sequences is a core technique in video analysis and　coding. Higher-order motion parameters of a video sequence can be used for object　segmentation, motion prediction, noise filtering, and to characterize the camera motion.　Extracted characterization for camera motions such as pan, tilt, zoom, rotation, track, boom,　and dolly can be used for video content analysis. Furthermore, short-term global motion　parameters are used to generate so-called background sprites or mosaics, which contains all　the background information over a certain video sequence. Theses background sprites are　employed for video coding and analysis. In recent years, there has been done much progress　in this field. Recent work has shown excellent results for coding and analysis of video　sequences.

The workshop on Multimedia Signal Processing (MMSP) deals with topics in video content　analysis and video coding. A special session about latest results in global motion estimation　and mosaic generation techniques should be of much interest for the MMSP community.　These techniques provide opportunities for a large number of related topics in image and　video processing, e.g. video compression, video summarization, video segmentation, object　segmentation, and structure from motion techniques for 2D/3D conversion. Even for bioinspired　multimedia signal processing, core image registration techniques can be used.