FIELD: information technology.
SUBSTANCE: invention relates to data compression techniques, particularly a method and apparatus for multistep quantisation. The result is achieved due to that, in the method for multistep quantisation, a reference codebook is obtained in accordance with the codebook of the previous step, the codebook of the current step is obtained in accordance with the reference codebook and a scaling factor, and the input vector is quantised using the codebook of the current step.
EFFECT: saving memory and more efficient use of resources.
11 cl, 7 dwg
Authors
Dates
2012-06-20—Published
2008-10-31—Filed