Compression

Jose Maria Caro (Editor: Rachel Chertkoff)

The more the capacity of communication and transmission systems improves, the bigger the documents (video, image and data) get. Big data objects need large amounts of data storage and, as the volume of data storage grows, the time needed to access that data increases.

Every single time a data file (a document, a video or still image, an audio file, etc.) is digitized it produces a huge amount of information. Due to its size, this resultant information is difficult and costly to store and transmit, so it must be compressed into smaller documents without losing vital information.

Compression is a technology that applies a number of transformations that reduce the size of the initial digital data (text, documents, video, sound or graphics) as much as possible by using an algorithm, but always assuring its later reproduction without any degradation of information.

In order to reduce the volume of data, this algorithm exploits spatial and temporal redundancies and eliminates the data that can not be displayed by the imaging devices used by the consumer. In other words, compression eliminates all the data repetition among bits inside a file. For instance, if a file contains 200 bits with the number 36, which corresponds to the color of a portion an image, compression will reduce it to only two bits. One byte will represent the color (the number 36) and the other one will indicate how many times that color is repeated inside the image (200).

Most compression systems contain the following elements (Lakhani 1996):

• Digitalization, Sampling and Segmentation: transformation of an analog signal/file into digital data and division of that data into blocks.
• Redundancy Reduction: dropping of duplicated information.
• Entropy Reduction: representation of digital data using fewer bits.
• Entropy Coding: designation of code words (bit strings) shorter in length to minimize the number of bits required to code a file.

There are two main types of compression technologies:

• Lossless Compression: there is no change or loss in the original data during the compression and decompression processes. It is mostly used for text compression (spreadsheets, word processors, database files, program executables) and repetitive data files (binary images and grayscale images) where pixel repetition is huge. Lossless technologies can reduce a file from 1/10 to 1/50 of its original size.

• Lossy Compression: this is used for files with very little redundancy in pixels. The loss of some information is the key factor of this type of compression. Lossy compressors use the ability of the human eye and ear to fill in the missing information. The most important factor here is how much information can be lost before the human eye/ear realizes the gap. This technology is mainly used in multimedia applications (video, video teleconferencing, etc.)

Some common Lossy standards are: Joint Photographic Experts Group (JPEG), Motion Pictures Experts Group (MPEG), Intel DVI, CCITT H.261 (Px64) Video Coding Algorithm and Fractals.

Finally, it is also important to distinguish between two types of compressors:

• Fixed Compressors: these compress one or more files together and store them in their reduced form. These files cannot be used until a special decompression program decompresses them.
• Real Time Compressors: they compress data while it is written/introduced in the computer/unit, and decompress it automatically when reading it.

Compression technologies have been used for many years to reduce both the amount of data to store and the transmission time of that data. It was around 1940 when C.E. Shannon developed, for the first time ever, ‘a probabilistic vision of information as well as its representation, transmission and compression’ (Sierra-Llamazares Alguacil 1997).

In the 1970s, compression of images and audio was already used through analog techniques to compress the video and audio bandwidth. These technologies, as well as the digital ones still used now, helped enormously to reduce the cost for establishing television and telephone circuits.

The arrival of computer/digital technologies produced a revolution in the communication industries, but compression is still mandatory in order to take down storage and transmission costs. The eighties were the revolutionary decade for the use of compression in the telephone industry as the nineties are for the use of compression in the television industry.

Now more than ever, time and space are money, digital files are extremely big, and it is the job of compression to save that time and space.

Standardization has been essential for the development of this technology and for all the manufacturers to work in the same track. CCITT developed in the early 80’s and MPEG established by the International Standards Organization in 1988 are some example of standards.

The MPEG standard is based in an encoding system. 3DO, C-Cube, DiviCom, IBM, SGS Thomson and AT&T are some of the companies which develop encoders for this system. Their cost can range from $10,000 up through to six figures. In 1993 AT&T presented a real time MPEG 2 compression system at a price of $90,000$ (information updated until 1994).

A few companies (AT&T, General Instrument, MIT, Philips, Sarnoff Labs, Thomson, and Zenith) together with the MPEG 2 committee and the FCC formed an alliance in the early nineties to define the advance digital television system which includes the High Definition Television (HDTV) system.

Compression technologies have many different applications, for example: facsimile systems, printer systems, document storage and retrieval systems, video teleconferencing and telephony systems, electronic multimedia messaging systems, military systems, television and audio systems, and multimedia systems.

Each of these applications uses one kind of compression technology. Sometimes that technology shares its standard with another.

Now, let’s take a deeper look into some of the most commonly used standards and their different applications:

• CCITT Groups 3 1-D, 3 2D and 4 2D Compression: they are based on different versions of run-length encoding and their primary uses are facsimile systems and software-based document imaging systems.
• JPEG: designed for continuous-tone images (not restricted to dual tone-black and white), it offers different types of resolution. Even though it discards any information with little visual effect, this system does not lose the detail of the original image.
• CCITT H.261 Video Coding Algorithm: mainly used for videophone and video conferencing systems, this is a real time compressor. It uses both lossy and lossless compression.
• MPEG: is world-wide recognized video, film, graphic and text compressor. It is based in four techniques:
• Pre-processing: this takes away non-essential visual information.
• Motion Compensation: the system assumes that the actual frame maybe similar to the preceding and the following one.
• Temporal prediction: the system predicts the following picture and eliminates redundancy within a single frame.
• Quantization coding: it codes the predicted pictures with a reference to a past one.

It is important to point out that MPEG compression also takes advantage of the properties

of the human vision (lower sensitivity to high frequencies of luminance and chrominance).

The MPEG standard consist of different standards:

• MPEG 1: introduced in 1991, this produces videocassette recorder (VCR) quality images. One of its main uses is personal computer video.
• MPEG 2: covers video audio and systems and offers broadcast quality and great levels of compression.

Many social scientists divide communication in three classes, according to Andleigh (1996):

1. Unexpected communication which interrupts the user from performing his or her normal duty.
2. A class of communication where the user must wait for the process to finish before he or she can start new task.
3. A class of communication at the proper speed until the task is completed.

When using computers, humans have a ‘patience factor ranging from two to four seconds’ (Andleigh 1996:53). Any response from the computer that goes beyond that range disturbs the working rhythm of the consumer. That is why compression could be valuable by reducing transmission time.

Storage space is the other force that helps in the developing of new compression systems. As digital technology advances, consumers work with bigger files. The problem is that hard disks remain small, so in order to store those files, they must be compressed into smaller ones.

On the other hand, compression itself is the driving force of many other emerging technologies such as Digital Television, Digital Telephony, etc.

Also, without compression technologies the computer hardware technologies would have evolved in a very different way. For instance, and due to the large amount of data that a digital file has, without compression, 1.4 MB floppy disks would not exist anymore.

Compression standards have achieved their goal; to create a set of rules for the development of the technology worldwide. In other words, standardization makes easier for the equipment from different manufacturers to work together and avoids the need for customized software and hardware. This way the same compressors and decompressors can be used with different systems. For example, the MPEG standard is compatible both with the PAL and NTSC television systems.

To better understand this point it is important to say that standards committees, like the MPEG committee, only set the technical standards and it is the manufacturers who create the different patents for their own products.

Unfortunately, it seems that even though the future will bring better and cheaper compressors, it will also bring more standards, and this can create confusion both in the market and in the consumers.

As technology gets slowly better and the communication industries join together, the quality of compression algorithm should get better, its prices should get lower and its future should get bigger. The only problem is that technology doesn’t grow at the same speed as the market necessitates, but experts say we should not worry: technical problems can always be solved.

The MPEG is already presenting the MPEG 4 and MPEG 7 algorithms, which add features to speed up the work and make it easier.

SIZE AND CODING BITS RATES OF VIDEO APPLICATIONS (Lakhani 1996)

COMPRESSION RELATED SITES (Pascual 1998)

• File Compression:

• Programs related with compression:

• Compressor’s codes:

• MPEG Compression:

• MPEG Standard:

• Software related to MPEG-2 Audio:

• Audio MPEG files:

• Graphic Formats:

• Newsgroups:

Andleigh, P.K. and Thakrar, K (1996) Multimedia Systems Designs, Prentice Hall PTR.

Baldwin, T., Mc Voy, D.S. and Steinfield, C. (1996) Convergence: Integrating Media, Information & Communication, SAGE Publications.

Bhatt, B., Birks, D. and Hermreck, D. (1997) ‘Digital television: making it work’, Spectrum 3:22-23.

Borthick, S.L. (1998) ‘New horizons in video compression’, DVC, Desktop Video Communications, January/February 1998.

Crowcroft, J. (1997) ‘Video compression’: online. Available HTTP: http://ee.mokwon.ac.kr/~music/tutorials/mmbook/node131.html (20 March 1998).

Lakhani, G. (1996) ‘Demand for video compression’: online. Available HTTP: http://cs4sum.cs.ttu.edu/lakhani-doc/compression_papers/video_tutorial/node1.html (20 March 1998).

Long, M. (1996) ‘Understanding MPEG 2 digital video compression’: online. Available HTTP: http://www.sat-net.com/mlesat/Article7.html (20 March 1998).

MPEG FAQ, ‘The MPEG standard’: online. Available HTTP: http://www.gti.ssr.upm.es/~vadis/faq_MPEG/mpeggeneral.htm (15 February 1998).

Pascual, J.A. (1998) ‘La Compresion de Datos (El Pilar Oculto de la Informatica): online. Available HTTP: http://www.canaldinamic.es/PCMANIA/PC056/PO/pc056poportac000.html (20 March 1998).

Shen, K., Delp, E. and Proff, J. (1998) ‘Parallel video compression’: online. Available HTTP: http://yara.ecn.purdue.edu/~ke/compression/compression.html (20 March 1998).

Sierra-Llamazares Alguacil, S. (1997) ‘Compresion de Imágenes (Formatos): Introduccion’: online. Available HTTP: http://www.uco.es/~i42sials/comp/node1.html (20 March 1998).

Wong, W., Chen, F., Xie, G. and Jiang, E. (1995) ‘Video compression’: online. Available HTTP: http://heart.engr.csulb.edu/~slobo/authoringProject2/reference/video/video1.html (20 March 1998).

Back to Home Page