Data compression ratio

From Wikipedia, the free encyclopedia

Data compression ratio is a computer term used to quantify the reduction in data quantity produced by a data compression algorithm. A data compression ratio is much like a physical compression ratio used to measure the physical compression of substances.

Compression ratios are generally expressed as <compressed size>:<original size>, or as a percentage. If given as a percentage, the most common form is:

${\rm Compression\;Ratio} = \frac{\rm Compressed\;Size}{\rm Original\;Size}$

Thus a 10MB file that compresses to 2MB would have a compression ratio of 2:10, or 20%.

Sometimes the inverse is used, with the form:

${\rm Compression\;Ratio} = \frac{\rm Original\;Size - Compressed\;Size}{\rm Original\;Size}$

Thus a 10MB file that compresses to 2MB would have a compression ratio of (10-2):10, or 80%.

Loss-less compression of digitised data such as photographs and audio preserves all the information, but can rarely do much better than 50% compression. In contrast, lossy compression (for example JPEG, or MP3) can achieve much better compression ratios at the cost of a decrease in quality (visual or audio artifacts are introduced).

Data compression ratios are also closely related to the concept of coding rates, which express the same concept but relate the ratio to more tangible quantities. For example, the coding rate of an audio stream may be expressed in "bits per second", or still image compression may be expressed in "bits per pixel". In many instances, coding rates can be converted to compression ratios if the physical characterstics of content of the original file are known. Lower coding rates indicate a higher level of compression.