IMAGE COMPRESSION METHOD Russian patent published in 2010 - IPC G06T9/00 

FIELD: information technology.

SUBSTANCE: following operations are performed: image data are obtained at the input of an optimum encoder; the image is evaluated through the encoder and the class of the image is determined based on a set of compression filters; the image is divided into multiple non-overlapping rank blocks, adaptive granulation of the rank blocks is applied; conversions are pre-calculated for all domains using the encoder based on the set of filters for the predetermined class and a set of domains is created; vectors of characteristics for all domains are calculated using the encoder; for each rank block, a domain block is sought among previously calculated blocks, as well as transformation in form of rotation/reflection which meet error requirements; if the rank block is not covered after comparing all domains, it is divided into smaller rank blocks and the previous step is performed, if the minimum size of the rank block is reached and the rank block is not covered with the allowable error, the transformation and domain block with minimum error level are stored.

EFFECT: high degree of image compression.

4 dwg