FIELD: physics, computation hardware.
SUBSTANCE: invention relates to computer engineering. Proposed method comprises the stages whereat: Current block adjacent initial pixels are filtered, the block being coded to form filtered adjacent pixels. Determination is executed to select reference pixels for in-frame forecast with respect to current block among one of filtered adjacent pixels and initial adjacent pixels. Reference pixels are selected from filtered adjacent pixels if it is defined to select reference pixels from filtered adjacent pixels. Reference pixels are selected from initial pixels if it is defined to select reference pixels from initial adjacent pixels. In-frame forecasting is executed relative to current block via application of selected reference pixels. Note here that current block makes the unit of coding obtained by splitting of current frame in compliance with the maximum coding unit. The latter represents coding unit of maximum size and coding depth representing an hierarchical data of the maximum coding unit splitting.
EFFECT: higher efficiency of video compression.
14 cl, 26 dwg, 3 tbl
