Prediction of a movement vector of a current image partition having a different geometric shape or size from that of at least one adjacent reference image partition and encoding and decoding using one such prediction

Abstract

The invention relates to a method for predicting a movement vector (MVp) of a partition (P) of a current image (I) from a plurality of n reference movement vectors associated respectively with n reference partitions that have been previously encoded and decoded. For a spatial prediction of one such vector, when the geometric shape of the current partition is different from that of k adjacent reference partitions (pr, pr, . . . prk), with k≤n, the movement vector of the current image partition is determined from a function of at least one reference movement vector belonging to a set of k reference movement vectors associated respectively with k adjacent reference partitions.