Page 136 - Video Coding for Mobile Communications Efficiency, Complexity, and Resilience

Section 4.6. Block-Matching Methods 113

:
w current , w left , w above , w above-left w above-left w above
2N × 2N windows centered around current
N × N block and its left, above, and above-
left neighboring blocks, respectively.
d , d l , d , d a-l :
c a
motion vectors of the current block and
d a-l a d
its left, above, and above-left
neighboring blocks, respectively.
2N N
current block.
l d c d
this part of the current block
is predicted using the 4
vectors shown.
w left
neighboring block. w current
Figure 4.8: Overlapped motion compensation for the top-left quadrant of the current block

N × N block from the reference frame to the current N × N block in the
current frame. In OMC, however, the estimated block motion vector is used
to copy a larger block (say, 2N × 2N ) from the reference frame to a position
centered around the current N × N block. As illustrated in Figure 4.8, since
they are larger than the compensated blocks, the copied blocks overlap, hence
the name overlapped motion compensation. Each copied block is weighted
by a smooth window, with higher weights at the center and lower weights
toward the borders. This means that the estimated motion vector is given
more in*uence in the center of the block, and this in*uence decays toward
the borders, where neighboring motion vectors start taking over. This ensures
a smooth transition between blocks and therefore reduces blocking artefacts.
Overlapped motion estimation and compensation can also be implemented in
the frequency domain, as proposed by Young and Kingsbury [95].
Another view of the OMC process is that each pel in the current N × N
block is compensated using more than one motion vector. For example, in
Figure 4.8, each pel is compensated using four motion vectors. The set of
motion vectors is decided according to the spatial position of the pel within the
block. A pel in the top-left quadrant of the current block will be compensated
using the motion vector of the block itself, plus the motion vectors of the
blocks to the left of, above, and above left of the current block. Each vector
provides a prediction for the pel, and those four predictions are weighted
according to the spatial position of the pel within the block. For example, as
the spatial position of the pel gets closer to the left border of the block, a
higher weight is given to the prediction provided by the motion vector of the
block to the left.