Over 650 pre-registrations for MPEG-7 technology
Subjective tests prove error resilience MPEG-4 Video
Rome, Italy, December 1998 – A total of 657 pre-registrations have been received in response to the call for proposals for the new MPEG-7 multimedia description standard. These pre-registrations can represent relatively ‘small’ items like a color descriptor, but also ‘large’ submissions, such as a proposal for a Description Definition Language. Many pre-registrations when taken together form a larger, comprehensive proposal. A total of 74 parties have pre-registered proposals with MPEG, industrial and academic groups around the world among them. This signals that the intended user base takes MPEG-7 very seriously. A complete list of parties that have pre-registered is annexed to this press release.
The 46th MPEG meeting took place from 7 - 11 December, in Roma, Italy, at the invitation of UNINFO. MPEG-4 evaluation tests demonstrated the error resilience and the performance of scalable coding in the MPEG-4 Visual part of the standard. Superb performance was demonstrated for the MPEG-4 ‘Simple Scalable Profile’. This is a low complexity video profile that allows the embedded coding of video sequences with two layers of different temporal resolution. A receiver may decide to either decode and display a video sequences with low frame rate at low bitrate, or at high frame rate by also accepting an incremental enhancement bitstream. Subjective testing results illustrate that the efficiency of doing this is equal to, or even better than a single layer approach. The results will shortly be made available on the MPEG home page (see below).
The MPEG-7 standard will be quite different from MPEG-1, -2 and -4. Its goal is not to define a representation for faithful reproduction, but to define a description that can be used for (automatic) search and identification of multimedia content. Pre-registration was mandatory for all intending to submit something in response to the Call for Proposals, issued last October. The proposals are due by February 1st, and will be evaluated in a special meeting in Lancaster, UK, 15-19 February 1999. MPEG was pleased to see both video and audio technology well represented in the pre-registrations.
To verify the performance of error resilient video in MPEG-4, subjective tests were carried out simulating MPEG-4 video carried in realistic multiplex and ditto radio channels, at bitrates between 32 kbit/s and 384 kbit/s. The test used a simulation of the residual errors after channel coding at bit error rates up to 10-3. The test methodology was based on a continuous quality evaluation over a period of three minutes. In such a test, subjects constantly score the degradation they experience. The results show that the average video quality achieved on the mobile channel is high, that the impact of errors is effectively kept local by the tools in MPEG-4 video, and that the video quality recovers quickly at the end of periods of error. These excellent results were achieved with very low overheads, less than those typically associated with the GOP structure used in MPEG-1 and MPEG-2 video.
Work is still ongoing on MPEG-4 Version 2, notably on supplying spatial scalability next to the temporal scalability already present. Using this technology, a receiver can except and display a low resolution version of a video at lower bitrate and display an enhanced resolution regions of interest by accessing incremental bitstreams.
The verification of the Body Animation Transform (BAT) has shown the high quality achievable for linking the motion of standard models of body joints to the animation of proprietary 3D body shapes. This capability can support such special effects as the bunching of muscles or shifts in clothing.
For 3D Model Coding, a method of partitioning large compressed 3D mesh models (e.g. CAD, landscapes) into chunks has shown support for continuous rendering of the incremental build-up of 3D models in a terminal while tolerating dropout of corrupted model pieces. Such decoded models can be rendered in the terminal while the ultimate model gradually arrives, so the user has emerging hints about final model details with transmission granularity traded for reduced compression efficiency. This capability will help accommodate different channel characteristics including mobile channels and the Internet, including the opportunity for users to have quick previews of models. Current representation schemes of 3-D models do not take into account either streaming or network errors very well.
Also in the MPEG-4 area, much progress was made on defining ‘MPEG-4 Conformance’. This is a part of the standard that defines how implementers can test whether their system complies to the standard. As for MPEG-1 and MPEG-2, the Conformance part follows the other parts of the standard in time. The difference in this case is a year: conformance will be ready by the end of ’99 and it will be officially promoted to International Standard by February 2000. Conformance defines, for the tools in the Systems, Audio, Visual and DMIF parts of the standard, mechanisms by which the performance of any commercial embodiment can be judged against the performance of the reference embodiment. To assist in this task, a large number of specific test bitstreams were defined and —in the case of Audio— many of them were also made immediately available.
Future MPEG meetings will be held in Korea (March '99), Canada (July '99), Australia (October '99), Hawaii (December '99) and the Netherlands (March 2000). A special MPEG-7 Evaluation meeting takes place in Lancaster, UK, February ’99.
For further information about MPEG, please contact:
Dr. Leonardo Chiariglione, (Convenor of MPEG)
Via G. Reiss Romoli, 274
10148 Torino, ITALY
Tel.: +39 11 228 6120; Fax: +39 11 228 6299
This press release and much other MPEG-related information can be found on the MPEG homepage:
The MPEG homepage has links to other MPEG pages, which are maintained by some of the subgroups. It also contains links to public documents that are freely available for download to non-MPEG members.
Journalists that wish to receive MPEG Press Releases automatically can contact the author:
KPN Research, Netherlands
tel. +31 70 332 5310
fax +31 70 332 5567
Annex - Alphabetical list of parties (companies, consortia, universities) that have pre-registered a proposal in response to the MPEG-7 Call for Proposals