Restriction on sub-block size derivation for affine inter prediction
A video coding device selects a current block for sub-block based affine inter-prediction, and derives a sub-block size for a sub-block of the current block. The sub-block size includes a sub-block width and a sub-block height. The device determines a unidirectional width threshold (TwU), a bidirectional width threshold (TwB), a unidirectional height threshold (ThU), and a bidirectional height threshold (ThB), where a total value of TwB and ThB exceeds a total value of TwU and ThU. The device determines that the affine inter-prediction is unidirectional inter-prediction or bidirectional inter-prediction. Based on the determination, the device applies the TwU or the TwB to the sub-block width, and applies the ThU or the ThB to the sub-block height. The device then derives a motion vector for the sub-block based on motion vectors for the current block, the sub-block width, and the sub-block height.
1. A method comprising:
selecting, by a processor at an encoder, a current block for sub-block based affine inter-prediction, the current block associated with a plurality of motion vectors;
deriving, by the processor, a sub-block size for a sub-block of the current block, the sub-block size including a sub-block width and a sub-block height;
determining, by the processor, a unidirectional width threshold (TwU), a bidirectional width threshold (TwB), a unidirectional height threshold (ThU), and a bidirectional height threshold (ThB), wherein a total value of TwB and ThB exceeds a total value of TwU and ThU, and wherein the total value of TwB and ThB is greater than the total value of TwU and ThU to ensure a minimum sub-block size for bidirectional inter-prediction exceeds a minimum sub-block size for unidirectional inter-prediction;
determining, by the processor, that the affine inter-prediction is unidirectional inter-prediction or bidirectional inter-prediction;
based on the determination of unidirectional inter-prediction or bidirectional inter-prediction, applying, by the processor, the TwU or the TwB to the sub-block width;
based on the determination, applying, by the processor, the ThU or the ThB to the sub-block height;
deriving, by the processor, a motion vector for the sub-block based on motion vectors for the current block, the sub-block width, and the sub-block height;
encoding, by the processor, prediction information for the sub-block in a bitstream, the prediction information based on the motion vector for the sub-block; and
transmitting, by a transmitter coupled to the processor, the bitstream toward a decoder for decoding and display.
2. The method of claim 1 , wherein a value of TwB, a value of ThB, a value of TwU, and a value of ThU are predefined in a memory coupled to the processor.
3. The method of claim 1 , further comprising encoding a value of TwB, a value of ThB, a value of TwU, a value of ThU, or combinations thereof, in a parameter set in the bitstream.
4. The method of claim 3 , wherein the parameter set is a sequence parameter set (SPS), a picture parameter set (PPS), a slice header, or combinations thereof.
5. The method of claim 1 , further comprising rounding the motion vector for the sub-block to a precision of one sixteenths accuracy.
6. The method of claim 1 , wherein the sub-block size for the sub-block of the current block is derived according to:
{
M
=
clip
3
(
Tw
,
w
,
w
×
MvPre
max
(
abs
(
v
1
x
-
v
0
x
)
,
abs
(
v
1
y
-
v
0
y
)
)
)
N
=
clip
3
(
Th
,
h
,
h
×
MvPre
max
(
abs
(
v
2
x
-
v
0
x
)
,
abs
(
v
2
y
-
v
0
y
)
)
)
where M is the sub-block width, N is the sub-block height, clip3 is a three parameter function that sets a first parameter as a lower threshold for a computation, a second parameter for an upper threshold for the computation, and a third parameter for the computation, Tw applies TwU or TwB, Th applies ThU or ThB, w is a width of the current block, h is a height of the current block, MvPre is a precision of the motion vector for the sub-block, max is a maximum function, abs is an absolute value function, and v 0x , v 1x , v 2x , v 0y , v 1y , and v 2y are motion vector components in a motion vector field derived based on the motion vectors for the current block.
7. An apparatus comprising:
a receiver to receive a bitstream;
a processor coupled to the receiver and configured to:
select a current block from the bitstream for decoding by sub-block based affine inter-prediction;
obtain a plurality of motion vectors for the current block from the bitstream;
derive a sub-block size for a sub-block of the current block, the sub-block size including a sub-block width and a sub-block height;
determine a unidirectional width threshold (TwU), a bidirectional width threshold (TwB), a unidirectional height threshold (ThU), and a bidirectional height threshold (ThB), wherein a total value of TwB and ThB exceeds a total value of TwU and ThU, and wherein the total value of TwB and ThB is greater than the total value of TwU and ThU to ensure a minimum sub-block size for bidirectional inter-prediction exceeds a minimum sub-block size for unidirectional inter-prediction;
determine that the affine inter-prediction is unidirectional inter-prediction or bidirectional inter-prediction based on the bitstream;
based on the determination of unidirectional inter-prediction or bidirectional inter-prediction, apply the TwU or the TwB to the sub-block width;
based on the determination, apply the ThU or the ThB to the sub-block height;
derive a motion vector for the sub-block based on motion vectors for the current block, the sub-block width, and the sub-block height;
employ the motion vector for the sub-block to reconstruct the sub-block based on a reference frame; and
forward a video sequence for display, the video sequence including the sub-block in a current frame.
8. The apparatus of claim 7 , further comprising a memory coupled to the processor, the memory configured to store a value of TwB, a value of ThB, a value of TwU, and a value of ThU as predefined values.
9. The apparatus of claim 7 , wherein the processor is further configured to obtain a value of TwB, a value of ThB, a value of TwU, a value of ThU, or combinations thereof, from a parameter set in the bitstream.
10. The apparatus of claim 9 , wherein the parameter set is a sequence parameter set (SPS), a picture parameter set (PPS), a slice header, or combinations thereof.
11. The apparatus of claim 7 , wherein the processor is further configured to round the motion vector for the sub-block to a precision of one sixteenths accuracy.
12. The apparatus of claim 7 , wherein the sub-block size for the sub-block of the current block is derived according to:
{
M
=
clip
3
(
Tw
,
w
,
w
×
MvPre
max
(
abs
(
v
1
x
-
v
0
x
)
,
abs
(
v
1
y
-
v
0
y
)
)
)
N
=
clip
3
(
Th
,
h
,
h
×
MvPre
max
(
abs
(
v
2
x
-
v
0
x
)
,
abs
(
v
2
y
-
v
0
y
)
)
)
where M is the sub-block width, N is the sub-block height, clip3 is a three parameter function that sets a first parameter as a lower threshold for a computation, a second parameter for an upper threshold for the computation, and a third parameter for the computation, Tw applies TwU or TwB, Th applies ThU or ThB, w is a width of the current block, h is a height of the current block, MvPre is a precision of the motion vector for the sub-block, max is a maximum function, abs is an absolute value function, and v 0x , v 1x , v 2x , v 0y , v 1y , and v 2y are motion vector components in a motion vector field derived based on the motion vectors for the current block.
13. A method comprising:
receiving, by a receiver, a bitstream;
selecting, by a processor, a current block from the bitstream for decoding by sub-block based affine inter-prediction;
obtaining, by the processor, a plurality of motion vectors for the current block from the bitstream;
deriving, by the processor, a sub-block size for a sub-block of the current block, the sub-block size including a sub-block width and a sub-block height;
determining, by the processor, a unidirectional width threshold (TwU), a bidirectional width threshold (TwB), a unidirectional height threshold (ThU), and a bidirectional height threshold (ThB), wherein a total value of TwB and ThB exceeds a total value of TwU and ThU, and wherein the total value of TwB and ThB is greater than the total value of TwU and ThU to ensure a minimum sub-block size for bidirectional inter-prediction exceeds a minimum sub-block size for unidirectional inter-prediction;
determining, by the processor, that the affine inter-prediction is unidirectional inter-prediction or bidirectional inter-prediction based on the bitstream;
based on the determination of unidirectional inter-prediction or bidirectional inter-prediction, applying, by the processor, the TwU or the TwB to the sub-block width;
based on the determination, applying, by the processor, the ThU or the ThB to the sub-block height;
deriving, by the processor, a motion vector for the sub-block based on motion vectors for the current block, the sub-block width, and the sub-block height;
employing, by the processor, the motion vector for the sub-block to reconstruct the sub-block based on a reference frame; and
forwarding, by the processor, a video sequence for display, the video sequence including the sub-block in a current frame.
14. The method of claim 13 , wherein a value of TwB, a value of ThB, a value of TwU, and a value of ThU are stored as predefined values.
15. The method of claim 13 , further comprising obtaining a value of TwB, a value of ThB, a value of TwU, a value of ThU, or combinations thereof, from a parameter set in the bitstream.
16. The method of claim 13 , further comprising round the motion vector for the sub-block to a precision of one sixteenths accuracy.
17. The method of claim 13 , wherein the sub-block size for the sub-block of the current block is derived according to:
{
M
=
clip
3
(
Tw
,
w
,
w
×
MvPre
max
(
abs
(
v
1
x
-
v
0
x
)
,
abs
(
v
1
y
-
v
0
y
)
)
)
N
=
clip
3
(
Th
,
h
,
h
×
MvPre
max
(
abs
(
v
2
x
-
v
0
x
)
,
abs
(
v
2
y
-
v
0
y
)
)
)
where M is the sub-block width, N is the sub-block height, clip3 is a three parameter function that sets a first parameter as a lower threshold for a computation, a second parameter for an upper threshold for the computation, and a third parameter for the computation, Tw applies TwU or TwB, Th applies ThU or ThB, w is a width of the current block, h is a height of the current block, MvPre is a precision of the motion vector for the sub-block, max is a maximum function, abs is an absolute value function, and v 0x , v 1x , v 2x , v 0y , v 1y , and v 2y are motion vector components in a motion vector field derived based on the motion vectors for the current block.