METHOD AND APPARATUS FOR ENCODING OR DECODING A VIDEO SIGNAL USING A SUMMARY REFERENCE PICTURE

摘要

In interframe video encoding temporal prediction is carried out that is based on motion compensation using a motion vector prediction from a reference frame. But in low bit rate coding, temporal prediction of blocks having texture content achieves only a low Rate-Distortion performance. According to the invention,for blocks having texture content a prediction from a reduced-size texture summary reference picture SRP (i.e. a position referencing) is carried out instead, wherein the components of the prediction vector are derived from the components of vectors for reconstructed blocks adjacent to the current block, and wherein the components of the current block vector refer to the position of the content of the current block within the SRP. The encoded data (SRPI) of the SRP and of the components of the block vectors are transmitted and evaluated at decoder side.