IMAGE PREDICTION USING THE REPARTITIONING OF A REFERENCE CAUSAL AREA, AND ENCODING AND DECODING USING SUCH A PREDICTION

摘要

The invention relates to a method for decoding a data stream (F) representative of an image or an image sequence, said stream (F) including data representative of at least one portion of said images (IE), wherein said method comprises the step of predicting a partition movement vector (P1; P'1 ) of a current image portion relative to a reference causal image (Cs; Ct) that has been partitioned according to a predetermined encoding mode for said area. The prediction step comprises, relative to the current partition (P1; P'1 ), the steps of: again partitioning (D3) at least one portion (MBra, MBrb, MBrc; MBnE-i) of the reference causal area into a plurality of reference partitions; determining (D4) the movement vectors (MVM -MVr3; MVr'1) respectively associated with reference partitions (Pr4a, Pr16b, Pr13c; Pr'1 ) resulting after the new partitioning; determining (D5) a predictive movement vector from said determined movement vectors (MVr1-MVr3, MVr'1 ); and predicting (D6) the movement vector of said current partition from said determined predictive movement vector.