|
MPEG4 Video Floods Internet
MPEG4 coding technology can compress video streams from 64 kbits/s to 2 Mbits/s. This technology will enable video streams to become commonplace on the Internet and video will finally become as ubiquitous a medium as text or graphics are now.
The world of video, once confined to the television (TV) set, is on the verge of an explosive expansion (Fig 1) with an advance that will allow video to be pumped through a range of networks connected to the Internet (Fig 2). What will make it possible is the next generation of high-efficiency encoding (data compression), known as Moving Picture Coding Experts Group Phase 4 (MPEG4). In September 1999, the standardization of MPEG4 began after six years of hard work. The advance was the acceptance of the basic version (Version 1) as an international standard to satisfy the demands of service providers and equipment manufacturers. It offers high-efficiency encoding of video streams, reducing them to data speeds between the range of 64 kbits/s and 2 Mbits/s. With MPEG4, it will be possible to pass video streams over the Internet with room to spare, whether through physical mediums or wireless systems.
With the new standard, video will not simply play the traditional role of providing entertainment, as it has on television. Instead, it will join text, graphics, voice and audio as a new medium for providing information. It will be possible for everyone to use video as easily as they use text today.
In today’s Internet access environment, the most common uses of video are to accompany voice conversations, or to provide free clips with purchased music software. Given the current peak access speed of 64 kbits/s, which imposes restrictions on resolution and motion (frame rate), applications are limited.
A Fast Future
The situation will change in the near future, though, as Internet access speeds continue to climb and image resolution improves (Fig 3). The use of cable and asymmetric digital subscriber line (ADSL) modems in the home will lead to Internet access at speeds ranging from several hundred kbits/s to several Mbits/s.
The same trend can be seen in wireless access. The IMT-2000 mobile communication system, slated for deployment by NTT Mobile Communications Network, Inc (NTT DoCoMo) of Japan in the spring of 2001, aims at a final speed of 2 Mbits/s. As one major Internet equipment manufacturer commented, “MPEG4 will deliver more than sufficient image quality once speeds rise to several hundred kbits/s to several Mbits/s.”
MPEG4 will help construct a new world of digital imagery, marking a departure from the existing world of TV.
Video performance is usually measured against TV, which means that the digital imagery of the MPEG standard must deliver at least the resolution of TV broadcasting. MPEG2 was developed to accomplish this, and will play an important part in expanding the potential of TV, such as by supporting an emerging market for digital video disk (DVD) players and digital TV receivers. MPEG4 is in a separate class, and will open a new dimension for video in non-TV applications, primarily via the Internet.
Radio broadcaster Tokyo FM Broadcasting Co, Ltd (FM TOKYO) predicts, “In the future, the number of MPEG4 terminals will exceed MPEG2 terminals.” While the MPEG2 terminal is moving toward its probable deployment target of one per home, MPEG4 may well spread until everyone has one of these standard-based systems connected to their portable equipment, such as a mobile phone (Fig 4).
Microsoft Adopting MPEG4
There are two major reasons why MPEG4 will become widespread. The first is that it will be offered as a standard function on personal computers (PC). The second is that it will become the standard technology for video distribution and the coming mobile communication services.
Microsoft Corp has already announced that it will support MPEG4 as a standard function, and will ship Windows 2000 with Windows Media Player 6.4 software to play (decode) MPEG4 video streams. The technology has already been implemented in the firm’s Internet Explorer 5.0 World Wide Web (WWW) browser and Windows Media Technologies 4.0 digital content distribution system.
Since MPEG4 standardization began in 1993, Microsoft has been an active participant in the process having, for example, provided copies of evaluation software. And the firm has proposed its own technologies for incorporation in the MPEG4 standard on a number of occasions.
Other Firms Adopt MPEG4
The decision by Microsoft to support the standard has galvanized peripheral equipment developers and content providers to also embrace MPEG4 technology.
Sharp Corp, for example, released its Internet Viewcam VZ-EZ1 MPEG4 videocamera at the same time Microsoft shipped Windows Media Player 6.4. A spokesperson for Sharp explained, “When we considered the video-play environment of most PC users, we felt we had little choice but to go with Windows Media Player 6.4, which is a standard function of the OS (operating system). So we ended up using MPEG4 video compression.”
The story is the same for FM TOKYO. When it began its “TFM TV” Internet video program in July 1999, the firm offered content in two versions: Windows Media Player 6.2 format and the RealPlayer G2 format developed by Real Networks, Inc of the US. As of May 1999, RealPlayer G2 had established itself as an industry standard, garnering over 3.2 million user registrations. “Recently, though, the number of people accessing with Media Player 6.2 has surpassed those with RealPlayer G2,” revealed a source at FM TOKYO.
Mobile Communications Use
MPEG4 has also been drawn upon for use in IMT-2000 and satellite digital radio broadcasting, both of which are on the verge of commercial service. NTT DoCoMo has stated that the “visualphones” (multimedia terminals) used in IMT-2000 service will use MPEG4. Mobile Broadcasting Corp of Japan, founded by firms including Toshiba Corp, Toyota Motor Corp and Fujitsu Ltd, will begin satellite digital radio broadcasting in the S band (2.6GHz) in 2001.
Both services will use the MPEG4 video compression scheme. In parallel with these developments, standardization is well under way on a domestic Japanese version now that MPEG4 Version 1 has been established as an international standard. In addition to IMT-2000 and mobile broadcasting services, MPEG4 is also likely to be used as the compression scheme for simple video transmissions primarily accompanying voice and audio services, such as Broadcast Satellite digital radio broadcasting and terrestrial radio broadcasting.
The development of an MPEG4 video codec integrated circuit (IC) for these terminals is now a high-priority project (Fig 5), with Toshiba and other Japanese manufacturers apparently in the lead.
According to carrier Internet Initiative Japan Inc (IIJ), “There is no particular reason to want to use MPEG4 if all that’s involved is the PC-based Internet of today. This is because RealPlayer G2 offers more than enough performance. Once mobile distribution services get started, though, MPEG4-based video communication and distribution may well take off. If they do, then MPEG4 content will deluge the Internet. That’s why MPEG4 cannot be ignored.”
Diverse Encoding Tools
A variety of applications making effective use of the high-efficiency coding that MPEG4 offers are beginning to emerge (Fig 6). MPEG4, however, offers more than just high compression. While not yet utilized in a product, it provides a range of data compression tools for use with a diverse range of data, including computer graphics (CG) and animation. MPEG2 only provided support for natural video captured with a video camera. MPEG4 can now handle natural video up to high-definition TV (HDTV) signals. MPEG4 is also very robust with respect to errors in encoded data – a handy characteristic for mobile communications, where transmission path conditions vary on a second-to-second basis.
If an error that cannot be fixed by error correction code (ECC) is detected, the damage is minimized through a robustness function. The error is identified, and its propagation is halted as much as possible to confine it as a local error. This technology is also called error concealment.
Of all the MPEG4 tools available, object encoding is one of the most interesting. This technology classifies objects by type (people, backgrounds, etc.) and then codes each object separately. This means that the image can be transmitted, stored and edited in object units. The technology is entirely different from the coding in frame/field units used in MPEG1 and MPEG2.
Of the specifications contained in Version 1, many carriers and manufacturers will adopt the simple profile, which defines compressing tools for rates of up to 384 kbits/s (Fig 7). This is the profile with the least load when it comes to the mobile communications that MPEG4 was originally intended for.
While standardization will continue, many people in the industry feel that Version 1 is sufficient for mobile video communications. The technical specification for Version 2, an expanded version that includes tools that offer higher coding efficiencies, was completed in July 1999, and is expected to be adopted as an international standard in July 2000. Discussion continues on Version 3, which will include tools that would not fit in Version 2.
Standard for IMT-2000
The simple profile will be the standard for the IMT-2000 next-generation mobile communication system. 3GPP, the industry body responsible for drawing up technology specifications for IMT-2000, formally adopt MPEG4 into the 3G-324M portable multimedia equipment specification in December 1999.
“We are recommending MPEG4 for use in IMT-2000 terminals equipped with video communications functions,” said Hiroyuki Yamaguchi, senior research engineer, Multimedia Laboratories, NTT DoCoMo.
MPEG4 has been adopted for more than just its high-efficiency coding, though. Another attraction is the robustness of the data with respect to errors. Another high-efficiency video coding scheme that is extremely robust is the International Telecommunication Union Telecommunication Standardization Sector (ITU-T) Rec.
H.263 encoding scheme for teleconferencing and telephony. While H.263 is designed to conceal errors on a one-to-one (1:1) connection, though, MPEG4 can minimize the propagation of errors by working only with received data (Fig 8). In other words, MPEG4 not only resists errors in 1:1 video communications, but also in one-to-multiple (1:N), such as in video distribution.
30% More Efficient
Outside of mobile communication applications, Toshiba, Hitachi Ltd and Microsoft all use the MPEG4 simple profile in their digital content distribution systems that handle video imagery. The single most important reason for its adoption is that MPEG4 offers, on average, 30% better encoding efficiency than MPEG1.
Toshiba’s digital content distribution system was planned from the ground-up with MPEG4 in mind. “MPEG4 was the only choice for a compression scheme, because we are planning to run video through low-speed wide-area networks (WAN),” explained Hidekazu Izumi, senior manager, Product Planning Group, Mobile Computing Division, Digital Media Equipment & Services Company, Toshiba.
Hitachi had been planning an MPEG1 editing system, but decided to switch to MPEG4 instead. “MPEG4 offers significantly better compression. We don’t need MPEG1 anymore,” said Akira Ikuma, senior engineer, Distribution and Service Systems, Information System Division, Hitachi.
Minimum Speed 64 kbits/s
Systems designed for the MPEG4 simple profile are likely to use 64 kbits/s for duplex video communication, and 384 kbits/s for simplex video distribution. The primary reason for this stems from the IMT-2000 service offered by NTT DoCoMo. In the initial stages, the uplink (terminal to base station) will probably be 64 kbits/s, and the downlink (the reverse) will likely be 384 kbits/s.
The speed of 64 kbits/s is the lowest possible rate for video communication, representing 10 to 15 frames/s for a resolution of 176 pixels x 144 pixels (quarter common intermediate format, or QCIF). The 64 kbits/s rate was actually the initial top-end target for MPEG4, but continual rises in network speed have led most manufacturers to position it at the bottom instead.
The top speed in the MPEG4 simple profile, 384 kbits/s, is used to provide the highest image quality in a number of applications. For an image format of 352 pixels x 288 pixels (common intermediate format, or CIF), this translates to 20 frames/s.
“Most demonstrations of video communication for IMT-2000 are made at 384 kbits/s,” explained Yamaguchi. “At that speed the images are as clear as they are on standard TV sets. They may be slightly worse when it comes to motion (frame rate), but in practical terms the image quality equals that offered by television receivers.” Sharp’s MPEG4 video camera, the Internet Viewcam VN-EZ1, is designed to handle high-resolution imagery with a 320 pixel x 240 pixel format (quarter video graphics array, or QVGA).
These simple-profile systems that use MPEG4 also come with functions for encoding audio and voice, in addition to video data. At present, however, there is no application that uses MPEG4 audio itself.
Object Encoding Technology
Object encoding is one of the few technologies in MPEG4 that the broadcast industry is interested in, even though it continues to pursue the commercial application of MPEG2 (digital TV broadcasting). By handling video as a collection of components (objects), the new technology promises improved program editing efficiency, and simplified storage and transfer within the broadcasting station. This feature came into existence as a by-product.
“Object encoding was originally developed as a framework to make it possible to use the optimal encoding scheme for each different type of data – natural video, CG, animation, or whatever,” explained Professor Hiroshi Yasuda, University of Tokyo, Japan. This would have meant that overall coding efficiency would be better than that provided by MPEG2 or other conventional technologies.
Japan Broadcasting Corp (Nippon Hoso Kyokai, or NHK), the British Broadcasting Corp (BBC), Sony Corp, and Heinrich-Hertz-Institut (HHI) of Germany are cooperatively drawing up object-encoding based MPEG4 studio standards (Fig 9). Yutaka Tanaka, director, Advanced Audio and Video Coding Research Division, NHK Science & Technical Research Labs commented, “The goal is to standardize encoding tools which are easy to use in program production.”
The first stage MPEG4 studio standard is the Studio Simple Profile (SSP), which was released as a working draft in July 1999. It is expected to become a formal international standard by Version 3 or higher.
The profile stresses assurance of compatibility with MPEG2, which is finding increasing use in program production systems. “Studio Simple Profile was developed from existing encoding tools, specifically to ensure that no major changes would be needed in existing MPEG2 program production systems,” explained Yoichi Yagasaki, senior research scientist, HomeNet Labs, Home Network Company, Sony. The firm plans to use the MPEG4 SSP in its MPEG World broadcast equipment systems running MPEG2.
Object encoding tools will also be incorporated into both the simple profile and core profile. The simple profile will only be able to handle rectangular objects, with any-shape object encoding supported only by the core profile.
Matsushita Electric Industrial Co Ltd has released encoding/decoding software that is compliant with the core profile on the Internet, for experimental use. Toshiya Takahashi, manager, Communications Group, Multimedia Development Center at Matsushita commented, “We really want to offer object encoding of any shape, and plan to continue public testing for some time to collect more content information.”
Object extraction, however, is manual work. Automatic object extraction technology is not covered within MPEG4 standardization, and is still at the research level. In Japan, NHK Research Labs and several other organizations are involved.
MPEG4 Codec ICs
When it becomes possible for individuals to use mobile terminals to send and receive video, each terminal will have to mount its own MPEG4 codec integrated circuit (IC). Major semiconductor firms are already swinging into action in anticipation of the opening of a major new market.
The coding circuits in the first-generation MPEG4 codecs announced thus far vary from vendor to vendor. The MPEG4 standard has considerable leeway and can be used in a wide variety of applications, so manufacturers stress different features and functions depending on which terminals and applications they intend to pursue. This is quite different from the situation in which MPEG2 codec development took off, because the MPEG2 chip was aimed squarely at specific applications, namely digital broadcasting receivers and DVD players.
NEC Corp, for example, only developed MPEG4 integrated circuits for the IMT-2000 mobile communication system. The firm constructed the coding circuitry using a digital signal processor (DSP) originally designed for existing mobile phones, with a dissipation of only 94mW. NEC has used the chip in a prototype portable videophone terminal for IMT-2000. Toshiba has developed a low-power custom IC aimed for use in terminals for IMT-2000 and mobile digital broadcasting.
Hitachi, on the other hand, is developing a general-purpose codec with no particular application in mind. The firm believes that widespread adoption of video communication will come about in 2002 or 2003, and plans to accumulate expertise in general-purpose codecs before developing specialized codec ICs for mobile communication terminals.
The MPEG4 codec IC developed by Matsushita Electric Industrial can handle CIF-size (352 pixels x 288 pixels) video streams at 15 frames/s. In an effort to establish the chip as a general-purpose item, it also supports the ITU-T recommendation for low-bitrate compression (H.261), and Wavelet conversion. Only the prototype chip has been disclosed and it is not known if the product version will offer the same functions and features.
by Masahiro Katou,
Chikashi Horikiri
- Websites:
- BBC
FM TOKYO
Fujitsu
HHI
Hitachi
IIJ
Microsoft
NEC
NHK
NTT DoCoMo
Real Networks
Sharp
Sony
Toshiba
Toyota
University of Tokyo
