In data compression, vector quantization is a technique (usually lossy) in which the basic idea is to code a multidimensional vector space into a subspace of lower dimension. The lower-space vector requires less storage space and the data is thus compressed. The transformation into the sub-space is usually achieved through projection, or by using a code-book.
See also: speech coding
External links: http://www.data-compression.com/vq.html