Method of bi-prediction video coding based on motion vectors from uni-prediction and merge candidate
A method of bi-predictive Inter prediction mode using a uni-predictive motion vector and a second motion vector associated with a Merge candidate is disclosed. In the encoder side, the coded motion information for a uni-prediction motion vector for the current block is generated using a motion vector predictor selected from a motion vector prediction (MVP) candidate list. Furthermore, a Merge candidate with a second motion vector pointing to the other prediction direction is identified according to a Merge index. The uni-prediction motion vector and the second motion vector from the Merge candidate are used for bi-prediction encoding of the current block. In the decoder side, the corresponding decoding process is disclosed.
1. A method of video decoding for blocks coded in Inter prediction mode, the method comprising:
receiving a coded bitstream associated with a current block;
determining, from the coded bitstream, a current motion prediction direction for the current block, wherein the current motion prediction direction is either bi-prediction or uni-prediction;
if the current motion prediction direction is uni-prediction with a first reference list:
determining, from the coded bitstream, a Merge candidate according to a Merge index pointing to the Merge candidate in a Merge candidate set;
deriving, from the coded bitstream, a first motion vector according to coded motion information with respect to a motion vector predictor in a motion-vector-prediction (MVP) candidate set, wherein the first motion vector refers to a first reference picture in the first reference list; and
applying bi-prediction motion compensated decoding to the current block based on the first motion vector and a second motion vector of the Merge candidate pointing to a second reference list different from the first reference list.
2. The method of claim 1 , wherein the coded motion information with respect to the motion vector predictor includes a reference index pointing to the first reference picture, a MVP index pointing to the motion vector predictor in the MVP candidate set, and a motion vector difference (MVD) between the first motion vector and the motion vector predictor.
3. The method of claim 1 , wherein if the uni-prediction corresponds to list L0, the second motion vector of the Merge candidate pointing to list L1.
4. The method of claim 1 , wherein if the uni-prediction corresponds to list L1, the second motion vector of the Merge candidate pointing to list L0.
5. The method of claim 1 , wherein the Merge candidate set is derived according to a Merge candidate set generation process for Merge mode coding.
6. The method of claim 1 , wherein the Merge candidate set is derived from an initial Merge candidate set by re-locating a bi-prediction Merge candidate to two uni-prediction candidates or re-locating a uni-prediction Merge candidate to an empty entry of the initial Merge candidate set, wherein the initial Merge candidate set is generated according to a Merge candidate set generation process for Merge mode coding.
7. The method of claim 1 , wherein the Merge candidate set is generated by inserting into the Merge candidate set with different Merge candidates from initial Merge candidates for initial Merge candidate set, wherein the initial Merge candidates are inserted into the initial Merge candidate set according to a Merge candidate set generation process for Merge mode coding.
8. The method of claim 1 , wherein the Merge candidate set for the second reference list is derived from another Merge candidate set for the first reference list.
9. The method of claim 1 , wherein whether to apply the bi-prediction motion compensated decoding to the current block is according to a flag explicitly signaled or implicitly inferred.
10. The method of claim 1 , wherein whether to apply the bi-prediction motion compensated decoding to the current block is according to a size of the current block, wherein the current block corresponds to a coding unit (CU) or a prediction unit (PU).
11. A method of video coding for blocks coded in Inter prediction mode, the method comprising:
receiving input data associated with a current block;
determining a current motion prediction direction for the current block, wherein the current motion prediction direction is either bi-prediction or uni-prediction;
if the current motion prediction direction is uni-prediction using a current motion vector referring to a first reference picture in a first reference list:
generating coded motion information for the current motion vector using a motion vector predictor selected from a motion-vector-prediction (MVP) candidate set;
determining a Merge candidate from a Merge candidate set; and
applying bi-prediction motion compensated encoding to the current block based on the current motion vector and a second motion vector corresponding to the Merge candidate pointing to a second reference list different from the first reference list.
12. The method of claim 11 , wherein the coded motion information with respect to the motion vector predictor includes a reference index pointing to the first reference picture, a MVP index pointing to the motion vector predictor in the MVP candidate set, and a motion vector difference (MVD) between the current motion vector and the motion vector predictor.
13. The method of claim 11 , wherein if the uni-prediction corresponds to list L0, the second motion vector of the Merge candidate pointing to list L1.
14. The method of claim 11 , wherein if the uni-prediction corresponds to list L1, the second motion vector of the Merge candidate pointing to list L0.
15. The method of claim 11 , wherein the Merge candidate set is derived according to a Merge candidate set generation process for Merge mode coding.
16. The method of claim 11 , wherein the Merge candidate set is derived from an initial Merge candidate set by re-locating a bi-prediction Merge candidate to two uni-prediction candidates or re-locating a uni-prediction Merge candidate to an empty entry of the initial Merge candidate set, wherein the initial Merge candidate set is generated according to a Merge candidate set generation process for Merge mode coding.
17. The method of claim 11 , wherein the Merge candidate set is generated by inserting into the Merge candidate set with different Merge candidates from initial Merge candidates for initial Merge candidate set, wherein the initial Merge candidates are inserted into the initial Merge candidate set according to a Merge candidate set generation process for Merge mode coding.
18. The method of claim 11 , wherein the Merge candidate set for the second reference list is derived from another Merge candidate set for the first reference list.
19. The method of claim 11 , wherein whether to apply the bi-prediction motion compensated encoding to the current block is according to a flag explicitly signaled or implicitly inferred.
20. The method of claim 11 , wherein whether to apply the bi-prediction motion compensated encoding to the current block is according to a size of the current block, wherein the current block corresponds to a coding unit (CU) or a prediction unit (PU).