主权项 |
1. Prediction method of a pixel block of an image that comprises
a search step for candidate patches during which a set of candidate patches is formed from at least one patch belonging to an image other than the image to which the block to be predicted belongs, each patch being formed by a block and a neighbourhood that is causal of this block and
a block prediction step comprising a dictionary definition sub-step during which it is determined, for each dictionary,
a first patch among the patches of said set of candidate patches, said first patch being close, in terms of content, to the patch formed from the block to be predicted and from its neighbourhood, andother patches of said set of candidate patches, each one of them being close, in terms of content, to this first patch,the dictionary thus containing said first patch and said other patches; a neighbourhood prediction sub-step during which, for each dictionary, a prediction of the causal neighbourhood of the block to be predicted is determined using a weighted linear combination of neighbourhoods of patches of the dictionary; weighting parameters which optimise the prediction are chosen, and a block prediction sub-step during which the pixel block of the image is predicted by a weighted linear combination of the pixels of the blocks of the patches of a dictionary, the weighting parameters of said linear combination being those optimums that have been determined during the neighbourhood prediction step. |