FIELD: information technology.
SUBSTANCE: image decoding method is carried out by extracting information from a bit stream which indicates an internal prediction mode; and performing internal prediction over the current unit, wherein the internal prediction includes: determining one of (i) a left-side neighbouring pixel of a first previous unit neighbouring the left side of the current unit and decoded before the current unit, and (ii) an upper neighbouring pixel of a second previous unit neighbouring the upper side of the current unit and decoded before the current unit, wherein the left-side neighbouring pixel is defined based on j*dy>>n, and the upper neighbouring pixel is defined based on i*dx>>m, where dx, dy, m and n are integers, and performing internal prediction using the determined one of the left-side neighbouring pixel and the upper neighbouring pixel, wherein the image is hierarchically divided from a plurality of maximum coding units to encoded depth coding units, according to depths, wherein the unit for encoding the current depth is one of a rectangular block of data, separated from the unit for encoding higher depths, and divided into units for encoding lower depths, independent of neighbouring coding units.
EFFECT: high efficiency of decoding video.
34 dwg, 5 tbl
