for his invention of motion compensation in the framework of coding concepts for moving pictures and for his fundamental contributions to hybrid coding enhanced by estimation.
Coding Motion Pictures – The Creation of the MPEG
Consumer electronics are benefiting from a major step forward in the evolution of communication technology. Audio visual systems – such as television, Internet applications or digital storage media – increasingly encode their source signals using digital coding techniques defined by the Motion Picture Experts Group (MPEG). Created under the auspices of ISO/IEC International Standardisation Organisation/International Electrotechnical Commission), the MPEG standards define highly efficient techniques for encoding audio and video signals while maintaining a high rate of data compression.
To date, the following MPEG standards have been adopted world wide:
MPEG-l: Storage applications, e. g. CD,1.5 MB/s, non-interlaced (progressive scan)
MPEG-2: Television and HDTV applications, interlaced and non-interlaced
MPEG-4: Internet and Virtual Reality applications
A further standard, MPEG-7 for multimedia applications, is under discussion. These standards form an important basis for digital audio visual systems such as digital television. They enable users to communicate across system boundaries, especially important for future multimedia applications. The economies of scale resulting from the standards make it possible to produce complex receiving decoders, e. g. for digital television sets, in large numbers at low prices.
Without MPEG there would be no digital television. Without Leonardo Chiariglione there would be no MPEG. Inspired by the standardisation work that the Joint Pictures Expert Group (JPEG) performed under the guidance of Hiroshi Yasuda (a former fellow student of Chiariglione’s at Tokyo University), Chiariglione, a well-travelled polyglot humanist, founded the MPEG in 1988. Like the JPEG on which it was modelled, MPEG was established under the auspices of ISO/IEC. Showing remarkable tenacity and persuasiveness, Chiariglione brought together experts in the coding of motion pictures from the world’s leading communication technology labs, harnessing their talents to create the MPEG standards in the 1990s. In meeting after meeting, internationally renowned experts engaged in heated discussions with one another and with Chiariglione to find the best
possibilities. One participant remembers how Chiariglione would “bang heads” until the new standard was agreed upon. Nothing and no one was going to stop him. The motto he adopted was, “One audio visual system for the whole world – the eyes and ears of people are the same everywhere.” By involving working groups from all over the world, a steady source of technical sophistication was assured; the international participation ensured broad acceptance of the standards and avoided incompatible proprietary solutions.
In satellite TV it is already customary to transmit programs encoded using the MPEG-2 standard. A small t7at square antenna, about 40 cm long, enables viewers in Munich or on the Cote d’Azur to receive unencrypted broadcasts of some 35 digital German-language TV programs and an equal number of digital radio programs at a single satellite position. In addition, a number of encrypted pay-TV programs is also available. This method of transmission allows four to eight digital programs to share the bandwidth formerly occupied by one analog channel. In addition to the many other advantages of digital technology, this high bandwidth efficiency explains the economic success and market breakthrough enjoyed by digital transmission technology for audio visual signals. One example may serve to illustrate this impressive result of decades of hard work put in by many researchers and developers:
Kotelnikov showed that an analog signal of cutoff frequency W can be represented by 2W samples. In order to digitalize the analog signals, a certain number of quantizing steps is always needed. In many cases 256 steps, i. e. 2 or 8 bits, suffice. It follows that the data rate must be 16W in bits per second. As anybody who is active in the field of communication technology knows, this data rate, as first shown by Nyquist, requires a transmission channel with a cutoff frequency of at least 8W for binary lowpass transmission. Of course, such a mechanically exact digitisation of the analog signal produces a data rate that requires eight times the transmission bandwidth of the analog signal – normally a completely unsatisfactory situation.
Fortunately, however, audio and video signals contain redundant, irrelevant material. As early as 1948, Shannon described in his theory of information that it is possible to encode such sources in a way that substantially cuts down the amount of data. For decades afterwards, a large number of articles were published on the topic of source coding. They contained excellent ideas that led to many practical improvements. However, the possibilities of refinement and optimisation of various methods, such as differential picture coding or transformation procedures, were exhausted long before data rates could be reduced to an acceptable level. Finally, in the late 1980s, a proposal made by Fabio Rocca in 1969 could be fruitfully employed owing to advances in microelectronics. In his contribution to the symposium Return Bandwidth Cornpression at MIT in April 1969, Rocca had entitled the description of his proposal “Television Bandwidth Compression Utilizing Frame-to-frame Correlation and Movement Compensation.”
Rocca’s idea was to measure movement in partial areas of the frame during motion picture sequences, in effect estimating movement, and to transmit only the difference in signals between the current frame and a motion compensated preceding frame. This is accomplished by calculating the motion compensated frame, using estimated movement vectors, from the frame previously transmitted. This process finally made the decisive increase in coding efficiency possible. The practical exploitation of this hybrid coding did not become feasible until a few years ago, however, when large-scale integration made highly complex digital circuitry and signal processors available. Today, when motion pictures are encoded anywhere in the world there is a very high chance that the concept of motion compensated hybrid coding is employed, and of course this is true whenever coding is based on the television standard MPEG-2.
In view of their accomplishments, the Curatorium of the Eduard Rhein Foundation has chosen Leonardo Chiariglione and Fabio Rocca to receive the Foundation’s Technology Award for the outstanding contributions that they, along with many other experts, have made to the creation of the MPEG standards.
Prof. Dr. Broder Wendland. Universität Dortmund