Multimedia compression

Multimedia compression is a general term referring to the compression of any type of multimedia, most notably graphics, audio, and video.

Because multimedia typically derives from data sampled by a device such as a camera or a microphone, and because such data contains large amounts of random noise, traditional lossless compression algorithms tend to do a poor job compressing multimedia. Multimedia compression algorithms, traditionally known as codecs, work in a lossy fashion:

1. Transform the data according to a model designed to reduce sample-to-sample correlation, concentrating the important signal in a few data values.
2. Quantize the data, most of which has become noise. Some codecs use a scalar quantizer followed by run-length encoding; others use vector quantization.
3. Use entropy coding such as Huffman coding to reduce the number of bits that the most common values use.