Method and apparatus for motion vector refinement-based inter prediction in image coding system
An inter prediction method performed by a decoding apparatus includes: receiving information on a motion vector difference (MVD) in units of integer samples; deriving a first motion vector predictor (MVP) in units of fraction samples on the basis of neighboring blocks of a current block; deriving a second MVP in units of integer samples on the basis of the first MVP; determining a first motion vector (MV) in units of integer samples for the current block on the basis of the second MVP and the MVD; determining a second MV in units of fraction samples for the current block on the basis of the first MVP and the first MV; and generating a prediction sample for the current block on the basis of the second MV. The present invention is capable of increasing the accuracy of the MV while reducing the amount of data allocated to the MVD.
1. An inter prediction method performed by a decoder, the method comprising:
receiving information about a motion vector difference (MVD) of an integer sample unit;
deriving a first motion vector predictor (MVP) of a fractional sample unit for a current block based on neighboring blocks to the current block;
deriving a second MVP of an integer sample unit based on the first MVP;
determining a first motion vector (MV) of an integer sample unit for the current block based on the second MVP and the MVD;
determining a second MV of a fractional sample unit for the current block based on the first MVP and the first MV; and
generating a predicted sample for the current block based on the second MV,
wherein determining the second MV comprises:
calculating a complexity factor (CF) based on the first MVP and the first MV; and
applying an offset to the first MV based on a comparison between the CF and a threshold to determine the second MV.
2. The method of claim 1 , wherein the second MV is determined based on a following equation:
mv
MVR
=
{
r
(
mv
)
+
β
,
CF
<
-
α
r
(
mv
)
,
-
α
≤
CF
<
α
r
(
mv
)
-
β
,
α
≤
CF
where mv MVR denotes the second MV, r(mv) denotes the first MV, α denotes the threshold value, and β denotes the offset.
3. The method of claim 1 , wherein the CF is calculated as a difference between the first MV and the first MVP.
4. The method of claim 1 , wherein the threshold and the offset are predetermined.
5. The method of claim 1 , wherein information regarding at least one of the threshold and the offset is received from an encoder.
6. The method of claim 1 , wherein the offset is determined based on a following table:
Index
β
0
0
1
¼
2
2/4
3
¾
where the Index represents a value of information about the offset, and β denotes the offset.
7. The method of claim 1 , wherein the second MVP is derived by applying rounding to the first MVP.
8. The method of claim 1 , further comprising receiving a first flag indicating adaptive motion vector resolution (AMVR) availability for the current block.
9. The method of claim 8 , further comprising receiving a second flag indicating motion vector refinement (MVR) availability for the current block,
wherein the second flag is received when a value of the first flag is 1.
10. The method of claim 9 , wherein when a value of the second flag is 1, the predicted sample for the current block is generated based on the second MV.
11. The method of claim 1 , further comprising:
receiving a residual signal for the current block; and
generating a reconstructed picture based on the predicted sample and the residual signal.
12. A video encoding method performed by an encoder, the method comprising:
determining a first motion vector (MV) of an integer sample unit based on motion estimation;
deriving a first motion vector predictor (MVP) of a fractional sample unit for a current block based on neighboring blocks to the current block;
deriving a second MVP of an integer sample unit based on the first MVP;
calculating a motion vector difference (MVD) based on the first MV and the second MVP;
encoding and transmitting the MVD;
determining a second MV of a fractional sample unit for the current block based on the first MV and the first MVP;
generating a predicted sample for the current block based on the second MV;
generating a reconstructed picture based on the predicted sample and a residual sample for the current block; and
storing the reconstruct picture in a memory,
wherein determining the second MV comprises:
calculating a complexity factor (CF) based on the first MVP and the first MV; and
applying an offset to the first MV based on a comparison between the CF and a threshold to determine the second MV.
13. The method of claim 12 , wherein the second MV is determined based on a following equation:
mv
MVR
=
{
r
(
mv
)
+
β
,
CF
<
-
α
r
(
mv
)
,
-
α
≤
CF
<
α
r
(
mv
)
-
β
,
α
≤
CF
where mv MVR denotes the second MV, r(mv) denotes the first MV, α denotes the threshold value, and β denotes the offset.
14. The method of claim 12 , wherein the CF is calculated as a difference between the first MV and the first MVP.
15. The method of claim 12 , wherein the threshold and the offset are predetermined.
16. The method of claim 12 , wherein the second MVP is derived by applying rounding to the first MVP.
17. The method of claim 12 , further comprising receiving a first flag indicating adaptive motion vector resolution (AMVR) availability for the current block.
18. The method of claim 17 , further comprising receiving a second flag indicating motion vector refinement (MVR) availability for the current block,
wherein the second flag is received when a value of the first flag is 1.
19. The method of claim 18 , wherein when a value of the second flag is 1, the predicted sample for the current block is generated based on the second MV.