| 摘要 |
The invention concerns a method for encoding a multi-view video stream comprising at least two video sequences, comprising a step of motion information prediction for a square or rectangular block of a current picture of a current view from a list of candidate motion information predictors, a motion information predictor including at least a motion vector, and a step of entropic encoding of an item of information representative of the position of a selected motion information predictor in the list. The method of the invention comprises, for the encoding of a current block, determining in the list of candidate motion information predictors of the current block a motion information predictor comprising a disparity motion vector pointing to a block of a picture of a video sequence corresponding to a view different from the current view, and in case of positive determination, including in the list of candidate motion information predictors at least one refined motion information predictor comprising a refined disparity motion vector predictor obtained from the determined disparity motion vector so as to obtain a refined list of candidate motion information predictors. |
