Parallel motion estimation in video coding
Methods for improved parallel motion estimation are provided that decouple the merging candidate list derivation and motion estimation for merge mode and skip mode and the advanced motion vector predictor (AMVP) candidate list construction from regular motion estimation to increase the coding quality in parallel motion estimation while meeting throughput requirements. This decoupling may be accomplished by modifying the availability rules for spatial motion data (SMD) positions for construction of the candidate lists. As part of the decoupling, largest coding units (LCUs) of a picture may be divided into non-overlapping parallel motion estimation regions (PMER) of equal size. Within a PMER, motion estimation for merge mode, skip mode, and normal inter-prediction mode may be performed in parallel for all the prediction units (PUs) in the PMER.
1. A method for derivation of spatial motion data (SMD) candidates for a prediction unit (PU) in video coding or decoding, the SMD candidates for inclusion in an inter-prediction candidate list for the PU, the method comprising:
determining an SMD position corresponding to an SMD candidate for the PU, wherein the PU is comprised in a parallel motion estimation region (PMER);
selecting at least some motion data of the SMD position as the SMD candidate when the SMD position is not in the PMER and the motion data is available; and
indicating that the SMD candidate is not available when the SMD position is in the PMER.
2. The method of claim 1 , further comprising:
determining an alternative SMD position for the SMD candidate when the SMD position is in the PMER, wherein the alternative SMD position is in a neighboring PMER of the PMER; and
selecting at least some motion data of the alternative SMD position as the SMD candidate when the motion data of the alternative SMD position is available,
wherein indicating that the SMD candidate is not available comprises indicating that the SMD candidate is not available when the SMD position is in the PMER and the motion data of the alternative SMD position is not available.
3. The method of claim 1 , further comprising:
receiving an indication of a size of the PMER.
4. The method of claim 3 , wherein the size of the PMER is one selected from a group consisting of 4×4, 8×8, 16×16, 32×32, and 64×64.
5. The method of claim 1 , wherein the SMD position is one selected from a group consisting of a left down neighboring position, an above neighboring position, an above right neighboring position, a left bottom neighboring position, and an above left neighboring position.
6. The method of claim 2 ,
wherein the SMD position is one selected from a group consisting of a left down neighboring position, an above neighboring position, an above right neighboring position, a left bottom neighboring position, and an above left neighboring position, and
wherein the alternative SMD position for the left down neighboring position is a corresponding position just outside a left boundary of the PMER, the alternative SMD position for the above neighboring position is a corresponding position just outside a top boundary of the PMER, the alternative SMD position for the above right neighboring position is a corresponding position just outside the top boundary of the PMER, the alternative SMD position for the left bottom neighboring position is a corresponding position just outside the left boundary of the PMER, and the alternative SMD position for the above left neighboring position is one selected from a group consisting of a corresponding position just outside the left boundary of the PMER and a corresponding position just outside the top boundary of the PMER.
7. The method of claim 6 , wherein the corresponding position for the left down neighboring positions is at luminance coordinates (xM−1, yP+nPSH−1), the corresponding position for the above neighboring position is at luminance coordinates (xP+nPSW−1, yM−1), the corresponding position for the above right neighboring position is at luminance coordinates (xP+nPSW, yM−1), the corresponding position for the left bottom neighboring position is at luminance coordinates (xM−1, yP+nPSH), the position just outside the left boundary corresponding to the above left neighboring position is at luminance coordinates (xM−1, yP−1), and the position just outside the top boundary corresponding to the above left neighboring position is at luminance coordinates (xP−1, yM−1), wherein (xP, yP) are luminance coordinates of an upper left corner of the PU, (xM, yM) are luminance coordinates of an upper left corner of the PMER, nPSH is a height of the PU, and nPSW is a width of the PU.
8. The method of claim 1 , wherein the SMD candidate one selected from a group consisting of a spatial merging candidate and a spatial motion vector predictor (MVP) candidate.