On geometric partition mode with intra and inter prediction
Coded information of a coding unit (CU) is received. The CU is partitioned into a first partition and a second partition based on a partition index. The partition index indicates a geometric partition mode (GPM) in which the CU is partitioned into the first partition and the second partition by a straight partition line. Based on a location of the straight partition line in the CU, (i) whether a first prediction mode of the first partition of the CU is one of an intra mode and an inter mode and (ii) whether a second prediction mode of the second partition of the CU is one of the intra mode and the inter mode are determined. The first partition of the CU is reconstructed based on the determined first prediction mode and the second partition of the CU is reconstructed based on the determined second prediction mode.
1. A method of video decoding performed in a video decoder, the method comprising:
receiving coded information of a coding unit (CU) in a current picture from a coded video bitstream;
partitioning the CU into a first partition and a second partition based on a partition index in the coded information, the partition index indicating a geometric partition mode (GPM) in which the CU is partitioned into the first partition and the second partition by a straight partition line;
determining a first prediction mode of the first partition of the CU as a first mode of an intra mode and an inter mode and a second prediction mode of the second partition of the CU as a second mode of the intra mode and the inter mode based on whether a width of the CU is equal to or larger than a height of the CU, the second mode being different from the first mode; and
reconstructing the first partition of the CU based on the determined first prediction mode and reconstructing the second partition of the CU based on the determined second prediction mode.
2. The method of claim 1 , wherein the determining further comprises:
when the straight partition line intersects a top left corner of the CU, determining that the first prediction mode of the first partition is the intra mode and the second prediction mode of the second partition is the inter mode based on the width of the CU being equal to or larger than the height of the CU.
3. The method of claim 1 , wherein the determining further comprises:
when the straight partition line intersects a top left corner of the CU,
determining that the first prediction mode of the first partition is the inter mode and the second prediction mode of the second partition is the intra mode based on the width of the CU being smaller than the height of the CU.
4. The method of claim 1 , further comprising:
when the straight partition line intersects a top left corner of the CU,
determining the first prediction mode of the first partition and the second prediction mode of the second partition based on which of a top reference sample and a left reference sample adjacent to the CU is available for intra prediction.
5. The method of claim 4 , wherein the determining further comprises:
based on (i) the left reference sample being adjacent to the first partition of the CU and available for the intra prediction and (ii) the top reference sample being unavailable for the intra prediction, determining that the first prediction mode of the first partition is the intra mode and the second prediction mode of the second partition is the inter mode.
6. The method of claim 1 , further comprising:
determining the first prediction mode of the first partition and the second prediction mode of the second partition based on which one of the first partition and the second partition of the CU contains a larger portion of a top corner pixel of the CU.
7. The method of claim 6 , wherein the determining further comprises:
based on the first partition of the CU containing the larger portion of the top corner pixel of the CU,
determining that the first prediction mode of the first partition is the intra mode and the second prediction mode of the second partition is the inter mode.
8. The method of claim 6 , wherein the determining further comprises:
when a first portion of the top corner pixel contained in the first partition is equal to a second portion of the top corner pixel contained in the second partition,
determining the first prediction mode and the second prediction mode based on which of a top reference sample and a left reference sample is adjacent to one of the first partition and the second partition.
9. The method of claim 1 , further comprising:
determining whether the first partition of the CU is intra coded or inter coded based on a first index;
determining whether the second partition of the CU is intra coded or inter coded based on a second index; and
when one of the first partition and the second partition is intra coded and another one of the first partition and the second partition is inter coded,
determining the intra mode from an intra mode list for the one of the first partition and the second partition that is intra coded based on an intra mode index included in the coded information; and
determining a merge candidate for the inter mode from a merge candidate list for the other one of the first partition and the second partition that is inter coded based on a merge index included in the coded information.
10. The method of claim 1 , further comprising:
determining whether one of the first partition and the second partition of the CU is intra coded and another one of the first partition and the second partition of the CU is inter coded based on a flag included in the coded information; and
when the one of the first partition and the second partition of the CU is intra coded and the other one of the first partition and the second partition of the CU is inter coded,
determining the intra mode from an intra mode list for the one of the first partition and the second partition that is intra coded based on an intra mode index included in the coded information; and
determining a merge candidate for the inter mode from a merge candidate list for the other one of the first partition and the second partition that is inter coded based on a merge index included in the coded information.
11. A method of video encoding performed in a video encoder, the method comprising:
partitioning a coding unit (CU) in a current picture into a first partition and a second partition based on a geometric partition mode (GPM) in which the CU is partitioned into the first partition and the second partition by a straight partition line;
determining a first prediction mode of the first partition of the CU as a first mode of an intra mode and an inter mode and a second prediction mode of the second partition of the CU as a second mode of the intra mode and the inter mode based on whether a width of the CU is equal to or larger than a height of the CU, the second mode being different from the first mode; and
encoding the first partition of the CU in a bitstream based on the determined first prediction mode and the second partition of the CU in the bitstream based on the determined second prediction mode.
12. The method of claim 11 , wherein the determining further comprises:
when the straight partition line intersects a top left corner of the CU,
determining that the first prediction mode of the first partition is the intra mode and the second prediction mode of the second partition is the inter mode based on the width of the CU being equal to or larger than the height of the CU.
13. The method of claim 11 , wherein the determining further comprises:
when the straight partition line intersects a top left corner of the CU,
determining that the first prediction mode of the first partition is the inter mode and the second prediction mode of the second partition is the intra mode based on the width of the CU being smaller than the height of the CU.
14. The method of claim 11 , further comprising:
when the straight partition line intersects a top left corner of the CU,
determining the first prediction mode of the first partition and the second prediction mode of the second partition based on which of a top reference sample and a left reference sample adjacent to the CU is available for intra prediction.
15. The method of claim 14 , wherein the determining further comprises:
based on (i) the left reference sample being adjacent to the first partition of the CU and available for the intra prediction and (ii) the top reference sample being unavailable for the intra prediction, determining that the first prediction mode of the first partition is the intra mode and the second prediction mode of the second partition is the inter mode.
16. The method of claim 11 , further comprising:
determining the first prediction mode of the first partition and the second prediction mode of the second partition based on which one of the first partition and the second partition of the CU contains a larger portion of a top corner pixel of the CU.
17. The method of claim 16 , wherein the determining further comprises:
based on the first partition of the CU containing the larger portion of the top corner pixel of the CU,
determining that the first prediction mode of the first partition is the intra mode and the second prediction mode of the second partition is the inter mode.
18. The method of claim 16 , wherein the determining further comprises:
when a first portion of the top corner pixel contained in the first partition is equal to a second portion of the top corner pixel contained in the second partition,
determining the first prediction mode and the second prediction mode based on which of a top reference sample and a left reference sample is adjacent to one of the first partition and the second partition.
19. The method of claim 11 , further comprising:
generating a first index that indicates whether the first partition of the CU is intra coded or inter coded based on a first index;
generating a second index that indicates whether the second partition of the CU is intra coded or inter coded; and
when one of the first partition and the second partition is intra coded and another one of the first partition and the second partition is inter coded,
determining the intra mode from an intra mode list for the one of the first partition and the second partition that is intra coded; and
determining a merge candidate for the inter mode from a merge candidate list for the other one of the first partition and the second partition that is inter coded.
20. A method of processing visual media data, the method comprising:
processing a bitstream that includes the visual media data according to a format rule, wherein
the bitstream includes coded information of a coding unit (CU) in a current picture; and
the format rule specifies that
the CU is partitioned into a first partition and a second partition based on a partition index in the coded information, the partition index indicating a geometric partition mode (GPM) in which the CU is partitioned into the first partition and the second partition by a straight partition line,
a first prediction mode of the first partition of the CU is a first mode of an intra mode and an inter mode and a second prediction mode of the second partition of the CU is a second mode of the intra mode and the inter mode based on whether a width of the CU is equal to or larger than a height of the CU, the second mode being different from the first mode, and
the first partition of the CU is processed based on the first prediction mode and the second partition of the CU is processed based on the second prediction mode.