FIELD: physics, computer engineering.
SUBSTANCE: invention relates to computer engineering. The method for interframe prediction for multiview video sequence coding, wherein a coded frame of the coded view video sequence is represented as a set of blocks; at least one restored frame from the coded view video sequence, designated as a reference frame, is determined; synthesised frames corresponding to the coded frame and each of the reference frames are generated using restored depth maps and restored video sequence frames from at least another view, the method further including: determining, in the synthesised coded frame, the synthesised current frame which corresponds to the coded block; obtaining at least one synthesised motion vector for the synthesised current block using the synthesised reference frames; outputting at least one adjusted motion vector for the coded block using a pattern mapping technique using a pattern of the coded block, using reference frames and using the obtained synthesised motion vector; calculating a prediction block for the coded block using adjusted motion vectors.
EFFECT: reduced amount of overhead information in the field of compressing multiview video sequences with depth maps.
9 cl, 9 dwg
