Method of depth based block partitioning

1. A method of depth-based block partitioning (DBBP) for multi-view video coding or three-dimensional (3D) video coding, the method comprising:

receiving input data associated with a current texture block in a dependent view;

determining a corresponding depth block or a reference texture block in a reference view for the current texture block;

deriving a unified representative value based on the corresponding depth block or the reference texture block;

generating a current segmentation mask from the corresponding depth block or the reference texture block using the unified representative value;

selecting a current block partition from block partition candidates based on the corresponding depth block or the reference texture block and the unified representative value; and

applying DBBP coding to the current texture block according to the current segmentation mask generated and the current block partition selected,

wherein the unified representative value is calculated from partial samples of the corresponding depth block or the reference texture block, and the partial samples correspond to only four corner samples of the corresponding depth block or the reference texture block, and the current texture block corresponds to a coding tree unit (CTU), a coding tree block (CTB), a coding unit (CU), or a prediction unit (PU).

2. The method of claim 1 , wherein the unified representative value corresponds to a mean, an average or a sum of all samples or partial samples of the corresponding depth block or the reference texture block.

3. The method of claim 1 , wherein said deriving the unified representative value is performed during said generating the current segmentation mask and information regarding the unified representative value is provided to said selecting the current block partition.

4. The method of claim 1 , wherein said deriving the unified representative value is performed during said selecting the current block partition and information regarding the unified representative value is provided to said generating the current segmentation mask.

5. The method of claim 1 , wherein said selecting the current block partition comprises comparing selected samples at multiple fixed positions of the block partition candidates in the corresponding depth block or the reference texture block.

6. The method of claim 5 , wherein the multiple fixed positions correspond to upper-left, upper-right, lower-left and lower-right corner samples of the corresponding depth block or the reference texture block.

7. The method of claim 5 , wherein the multiple fixed positions correspond to upper-left, upper-right, lower-left and lower-right corner samples of each partitioned block of each block partition candidate corresponding to the corresponding depth block or the reference texture block.

8. The method of claim 1 , wherein the unified representative value is calculated as an average of selected samples corresponding to sub-sampled positions of the corresponding depth block or the reference texture block.

9. The method of claim 8 , wherein the unified representative value is calculated for each partitioned block of each block partition candidate corresponding to the corresponding depth block or the reference texture block.

10. The method of claim 1 , wherein the unified representative value is calculated from an average of all samples in the corresponding depth block or the reference texture block.

11. The method of claim 1 , wherein said selecting the current block partition from block partition candidates comprises determining, for each block partition candidate, absolute difference between a first sum of first samples at first fixed-positions of a first partitioned block and a second sum of second samples at second fixed-positions of a second partitioned block for each block partition candidate, and selecting the block partition candidate having a largest absolute difference from the block partition candidates as the current block partition.

12. The method of claim 1 , wherein said selecting the current block partition from block partition candidates comprises determining, for each block partition candidate, absolute difference between a first sum of first samples of a first partitioned block and a second sum of second samples of a second partitioned block for each block partition candidate, and selecting the block partition candidate having a largest absolute difference from the block partition candidates as the current block partition.

13. The method of claim 1 , wherein said selecting the current block partition from block partition candidates comprises signaling one or more first flags in a video bitstream to indicate available block partition candidates.

14. The method of claim 13 , wherein said selecting the current block partition from block partition candidates comprises signaling one or more second flags in the video bitstream to indicate the block partition candidate selected as the current block partition.

15. The method of claim 13 , wherein said selecting the current block partition from block partition candidates comprises signaling one or more second flags in the video bitstream to indicate a partition direction of the block partition candidate selected as the current block partition.

16. The method of claim 1 , wherein the block partition candidates exclude AMP (asymmetric motion partitions) when AMP is not available for a current picture, current slice or current coding unit containing the current block.

17. A method of depth-based block partitioning (DBBP) for multi-view video coding or three-dimensional (3D) video coding, the method comprising:

receiving input data associated with a current texture block in a dependent view;

determining a corresponding depth block or a reference texture block in a reference view for the current texture block;

generating a current segmentation mask from the corresponding depth block or the reference texture block using a first representative value of the corresponding depth block or the reference texture block;

selecting a current block partition from block partition candidates based on the corresponding depth block or the reference texture block and a second representative value of the corresponding depth block or the reference texture block; and

applying DBBP coding to the current texture block according to the current segmentation mask generated and the current block partition selected; and

wherein the first representative value, the second representative value, or both are calculated from partial samples of the corresponding depth block or the reference texture block,

wherein the partial samples correspond to only four corner samples of the corresponding depth block or the reference texture block, and the current texture block corresponds to a coding tree unit (CTU), a coding tree block (CTB), a coding unit (CU), or a prediction unit (PU).

18. The method of claim 17 , wherein the partial samples correspond to four corner samples of the corresponding depth block or the reference texture block, and the current texture block corresponds to a CTU (coding tree unit), a CTB (coding tree block), a CU (coding unit), or a PU (prediction unit).

19. A method of depth-based block partitioning (DBBP) for multi-view video coding or three-dimensional (3D) video coding, the method comprising:

receiving input data associated with a current texture block in a dependent view;

determining a corresponding depth block or a reference texture block in a reference view for the current texture block;

determining a first representative value from first four corner samples of the corresponding depth block or the reference texture block;

generating a current segmentation mask from the corresponding depth block or the reference texture block using the first representative value;

determining a second representative value for each partitioned block of a block partition candidate corresponding to the corresponding depth block or the reference texture block based on second four corner samples of each partitioned block;

selecting a current block partition from block partition candidates based on the corresponding depth block or the reference texture block and the second representative values associated with the block partition candidates; and

applying DBBP coding to the current texture block according to the current segmentation mask generated and the current block partition selected,

wherein the first representative value, the second representative value, or both are calculated from partial samples of the corresponding depth block or the reference texture block, and the partial samples correspond to only four corner samples of the corresponding depth block or the reference texture block, and the current texture block corresponds to a coding tree unit (CTU), a coding tree block (CTB), a coding unit (CU), or a prediction unit (PU).

20. A method of depth-based block partitioning (DBBP) for multi-view video coding or three-dimensional (3D) video coding, the method comprising:

receiving input data associated with a current texture block in a dependent view;

determining a corresponding depth block or a reference texture block in a reference view for the current texture block;

generating a current segmentation mask from the corresponding depth block or the reference texture block;

determining a first representative value for first samples in a first partitioned block of the corresponding depth block or the reference texture block for each of block partition candidates, and a second representative value for second samples in a second partitioned block of the corresponding depth block or the reference texture block for each of the block partition candidates;

selecting a current block partition based on one of from the block partition candidates, wherein the selected current block partition that has a largest absolute difference between the first representative value of the current block partition and the second representative value of the current block partition among the block partition candidates; and

applying DBBP coding to the current texture block according to the current segmentation mask generated and the current block partition selected,

wherein the first representative value, the second representative value, or both are calculated from partial samples of the corresponding depth block or the reference texture block, and the partial samples correspond to only four corner samples of the corresponding depth block or the reference texture block, and the current texture block corresponds to a coding tree unit (CTU), a coding tree block (CTB), a coding unit (CU), or a prediction unit (PU).

21. The method of claim 20 , wherein the first representative value and the second representative value are calculated as a sum, an average, a mean, a weighted sum, or a linear combination of the first samples in the first partitioned block and the second samples in the second partitioned block respectively.