Estimation of motion at the level of the decoder by matching of models

Abstract

Methods and apparatus are provided for decoding and coding a data signal representing at least one current image split up into partitions. The method of decoding includes, for at least one current partition to be reconstructed in the image, steps of: prediction of the current partition from at least one already decoded causal partition, and reconstruction of the current partition, from the predicted current partition, delivering a reconstructed current partition. Following the aid reconstruction step, the decoding method implements steps of: determination of at least one already decoded causal partition using a motion estimation applied to the reconstructed current partition, and modification of the reconstructed current partition by calculation of a function applied both to the determined causal partition and to the reconstructed current partition, delivering a modified reconstructed current partition.