FIELD: physics, video.
SUBSTANCE: invention relates to video processing technology, particularly to predictive video coding. The method for predictive video coding includes extracting a pixel unit from the current frame and dividing said unit into a main and an additional pixel set; encoding the main pixel set; outputting its code stream and obtaining restoration values of said set; interpolating based on the restored pixel values, containing a restoration value of the main pixel set, and using the result as the predictive value of the additional pixel set; encoding the additional pixel set with intra-frame prediction; outputting its code stream and obtaining restoration values of said set; merging the restoration values of the main pixel set with the restoration value of the additional pixel set and obtaining restoration values of the pixel unit. Implementation of the present invention is suitable for highly parallel computations and provides fast-compression coding for a high-definition video source.
EFFECT: high compression efficiency when encoding video, reduced amount of memory used, high coefficient of successful cache accessing.
15 cl, 12 dwg
