Electronic Design

MPEG-4 Testing Begins As Efforts To Enhance Coding Move Forward

A number of companies and organizations are beginning to conduct tests on the MPEG-4 transport medium. So far, their work bodes well for this platform.

This standard will play an important role in a third-generation wireless system that will be investigated in France this month. During this test, France Telecom of Paris will provide 100 people with handheld PCs, representing future wireless devices. It will then offer these users high-speed voice, data, and animated images. Initially, services will be confined to localized information, on-demand radio, TV, and video telephony. Additionally, the company says that it is the first operator to test such wireless services in France.

Enhancement of the MPEG-4 standard continues as well. In July, the Moving Picture Experts Group (MPEG) met in Beijing, China. Group members responded to the MPEG-4 Video "Call for Evidence," issued during the March meeting, with several contributions on new tools for improved video coding. Their results indicate that there may be ways to further improve upon MPEG-4's coding efficiency. The attendees decided to proceed with a study to determine if new technology that warrants standardization exists.

MPEG-4 is supposedly the most complex standard yet in the multimedia sector. But this raises a moot point. Some wonder if the world really needs another standard for audio and visual information (see the table). Others ask if MPEG-4 deserves to supplant the three proprietary video formats that are now in use. Examining MPEG-4's qualities, as well as the legacy of earlier MPEG formats, may answer these questions.

"MPEG-4 is the ultimate convergence technology," says Larry Horn, vice president of licensing and business development, MPEG LA, Chevy Chase, Md. His company currently licenses the MPEG-2 and the IEEE 1394 standards. "Each of the MPEGs has really been about compression, but for very different purposes. All these platforms have embraced various compression technologies that operate both spatially, within frames, and temporally, from frame to frame."

Compression is a vital part of the process. With uncompressed data, one byte would be used to store each of the primary colors in each pixel of a televised video. Then, an hour-long video would require a whopping 20 billion bytes. HDTV would need 100 billion to 200 billion bytes.

As Horn points out, each standard has earned its place by meeting some essential needs, but not others. "MPEG-1, with a typical bit rate of 1.5 Mbits/s, was aimed at CD/video resolution, but it didn't deliver up anywhere near the quality required for high entertainment value, such as broadcast television, cable TV, or direct satellite, or even movies," he explains. Arriving in 1991, MPEG-1 is still widely used in the Far East in digital video players.

"MPEG-2 was quite an advancement, bringing the quality up to what was required by broadcast television, cable TV, and direct satellite, closely approximating movie quality," Horn continues. Arriving in 1994, MPEG-2 was introduced for compression and transmission of digital television signals. Today, MPEG-2 video compression is part of international standard ISO/IEC IS13818-2, and it is the de facto standard for entertainment video.

The MPEG-2 standard places no restrictions on video encoder implementation. It thereby has let encoder designers introduce new technology, improving compression efficiency and picture quality. Since MPEG-2 was introduced, compression efficiency has improved 30% to 40%.

Virtually no one doubts that MPEG-2 is likely to remain a dominant standard. It has succeeded in establishing an inventory of standard decoders in existing consumer products and in chip libraries. This is in spite of the fact that there is no provision for interactivity within the content. But that all changes with MPEG-4.

Object Oriented
MPEG-4 is the first technology platform designed for network-based multimedia delivery. Streaming media architectures such as MPEG-4 comprise three domains. The authoring-encoding-publishing (AEP) is the content-creation area. Next, the server component takes the compressed, encoded data and puts in it on a server, packetizes it, and sends it out onto the network. Finally, the player component resides on the user side, in the PC.

Until now, users' hands were tied. They could do little more than play, fast forward, and rewind. But with MPEG-4, authors can supply users with interactive content. Users can define any object within the media and then synthetically add in other digital objects. Or, they can search for objects in the media or add in data.

The figure portrays how an audio/visual scene in MPEG-4 is described. "Primitive media objects" can be thought of as leaves in a descriptive tree, while "compound media objects" encompass entire branches. As an example, a visual object corresponding to a person in a scene and a corresponding voice can be tied together to form a new compound media object containing both visual and aural components of that talking person. Such grouping lets authors and users construct complex scenes. Once delivered, MPEG-4 lets them manipulate meaningful sets of objects. For a full discussion of this process, point your browser to www.cselt.it/mpeg/standards/mpeg-4/mpeg-4.htm.

What is particularly revolutionary about MPEG-4, however, is that the user's vantage point can be translated to anywhere in the scene.

"We are going to introduce an object-based CODEC and player in the early part of next year," says Bill Ponton, vice president of sales at e-Vue Inc. of Iselin, N.J. This company specializes in MPEG-4 authoring, encoding, server, and player technology. In support of experiments such as the France Telecom program, Ponton adds, "MPEG-4 is unusually well suited for the wireless space because of its error resilience and ability to selectively degrade objects within a scene when the channel becomes degraded."

The video bit range is unusually wide (see the table, again). When encoded with MPEG-4, a bit stream in a wireless system can throttle back when it encounters noise.

"Let's say the wireless channel folds back the bandwidth," Ponton hypothesizes. In anticipation of such events, he says, the author can decide how to treat the various objects in a scene. Perhaps there are some important objects that need to appear very crisp. The author can allocate bits to those particular objects, allowing less-important objects in the scene to degrade. Now imagine a scene where the background isn't that important. The author can degrade the background relative to the foreground or with regard to some other object.

It looks as if MPEG-4 promises something for everyone. First, it will let AEP produce content that has far greater reusability and flexibility than is possible today in media like digital television, animated graphics, web pages, and their extensions. It also will become possible to better manage and protect content owner rights. Meanwhile, network service providers will benefit because MPEG-4 supplies transport mechanisms with generic quality of service (QoS) descriptors. Finally, end users will enjoy higher levels of interaction with content, within any limits set by the author, while avoiding the likelihood of noninteroperability that can come about with proprietary formats and players.

Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.