Back to results
US 9,756,359 B2Granted
Large blocks and depth modeling modes (DMM'S) in 3D video coding
AssigneeQUALCOMM Incorporated
InventorsXin Zhao, Ying Chen, Li Zhang
Life of this patent
Estimating the term…
Abstract
In some examples, a method of decoding depth data in a video coding process includes defining a depth prediction unit (PU) of a size greater than 32×32 within a depth coding unit (CU) and generating one or more partitions of the depth PU. The method also includes obtaining residual data for each of the partitions; obtaining prediction data for each of the partitions; and reconstructing each of the partitions based on the residual data and the prediction data for the respective partitions.
Claims27
A method of decoding depth data in a video coding process, the method comprising:
- defining a depth prediction unit (PU) of a size greater than 32×32 within a depth coding unit (CU);
- generating one or more partitions of the depth PU, wherein generating the one or more partitions of the depth PU comprises:
- upsampling an N×N partition pattern where N is 32 or less;
- storing a pattern list comprising a plurality of different, partition patterns, wherein each respective partition pattern is identified by a respective one of a plurality of partition pattern indexes;
- selecting, based on one of the partition pattern indexes, a partition pattern from the pattern list;
- and
- applying the selected partition pattern to the depth PU to generate the one or more partitions;
- obtaining residual data for each of the partitions;
- obtaining prediction data for each of the partitions;
- and
- reconstructing each of the partitions based on the residual data and the prediction data for the respective partitions.
8 more elements
A device for decoding video data comprising:
- a video data memory that stores a plurality of different, partition patterns;
- and
- one or more processors configured to:
- define a depth prediction unit (PU) of a size greater than 32×32 within a depth coding unit (CU);
- generate one or more partitions of the depth PU, wherein to generate the one or more partitions of the depth PU the one or more processors are configured to:
- upsample an N×N partition pattern where N is 32 or less;
- store a pattern list comprising the different, partition patterns, wherein each respective partition pattern is identified by a respective one of a plurality of partition pattern indexes;
- select, based on one of the partition pattern indexes, a partition pattern from the pattern list;
- and
- apply the selected partition pattern to the depth PU to generate the one or more partitions;
- determine residual data for each of the partitions;
- determine prediction data for each of the partitions;
- and
- reconstruct each of the partitions based on the residual data and the prediction data for the respective partitions.
11 more elements
A non-transitory computer-readable storage medium having stored thereon instructions that, when executed, cause one or more processors of a device for coding video data to:
- define a depth prediction unit (PU) of a size greater than 32×32 within a depth coding unit (CU);
- generate one or more partitions of the depth PU, wherein to generate the one or more partitions of the depth PU, the non-transitory computer-readable storage medium stores instructions that cause the one or more processors to:
- upsample an N×N partition pattern where N is 32 or less;
- store a pattern list comprising a plurality of different, partition patterns, wherein each respective partition pattern is identified by a respective one of a plurality of partition pattern indexes;
- select, based on one of the partition pattern indexes, a partition pattern from the pattern list;
- and
- apply the selected partition pattern to the depth PU to generate the one or more partitions;
- determine residual data for each of the partitions;
- determine prediction data for each of the partitions;
- and
- reconstruct each of the partitions based on the residual data and the prediction data for the respective partitions.
8 more elements
A device for decoding video data comprising:
- a video data memory that stores one or more partition patterns;
- and
- one or more processors configured to:
- define a depth prediction unit (PU) of a size greater than 32×32 within a depth coding unit (CU);
- generate one or more partitions of the depth PU, wherein to generate the one or more partitions of the depth PU, the one or more processors are configured to:
- upsample an N×N partition pattern where N is 32 or less;
- determine an average value of samples in a texture block, the texture block included at a first region of a texture picture, and the first region being co-located with a second region of the depth PU included in a depth map;
- and
- determine for each depth value of the depth map, based on comparing a respective sample of the texture block to the average value of samples, a partition of the one or more partitions to assign the respective depth value;
- and
- assign the respective depth value to the determined partition of the one or more partitions, wherein to assign the respective depth value to the determined partition of the one or more partitions further, the one or more processors are configured to:
- determine a first location of the respective depth value within the depth map;
- determine a second location of the respective sample in the texture block, wherein the first location in the depth PU is co-located with the second location in the texture block;
- and
- responsive to determining that the respective sample is greater than the average value of samples, assign the respective depth value of the first location to a first partition of the one or more partitions;
- obtain residual data for each of the partitions;
- obtain prediction data for each of the partitions;
- and
- reconstruct each of the partitions based on the residual data and the prediction data for the respective partitions.
16 more elements
Dependencies are inferred from the claim text.