HK1190255A - Parallelization friendly merge candidates for video coding - Google Patents

Description

Parallelization-friendly merge candidates for video coding

This application claims the following applications: united states provisional application No. 61/499,112, 2011, 6/20, 61/543,043, 2011, 10/4, 61/543,059, 2011, 11/7, 61/556,761, 2011, 11/21, 2011, 61/562,387, and 2011, 11/22, 61/562,953, which are hereby incorporated by reference in their entirety.

Technical Field

This disclosure relates to video coding and, more particularly, to techniques for determining a set of motion vector prediction candidates in a motion vector prediction process.

Background

Digital video capabilities can be incorporated into a wide range of devices, including digital televisions, digital direct broadcast systems, wireless broadcast systems, Personal Digital Assistants (PDAs), laptop or desktop computers, digital cameras, digital recording devices, digital media players, video gaming devices, video game consoles, cellular or satellite radio telephones, video teleconferencing devices, and the like. Digital video devices implement video compression techniques to transmit, receive, and store digital video information more efficiently, such as the video compression techniques described in the following standards: standards defined by MPEG-2, MPEG-4, ITU-T H.263, ITU-T H.264/MPEG-4 (part 10, Advanced Video Coding (AVC)), the High Efficiency Video Coding (HEVC) standard currently under development, and extensions of such standards.

Video compression techniques include spatial prediction and/or temporal prediction to reduce or remove redundancy inherent in video sequences. For block-based video coding, a video frame or slice may be partitioned into multiple blocks. Each block may be further partitioned. Blocks in an intra-coded (I) frame or slice are encoded using spatial prediction with respect to reference samples in neighboring blocks in the same frame or slice. Blocks in an inter-coded (P or B) frame or slice may use spatial prediction with respect to reference samples in neighboring blocks in the same frame or slice or temporal prediction with respect to reference samples in other reference frames. Spatial prediction or temporal prediction results in predictive block coding of a block. The residual data represents pixel differences between the original block to be coded and the predictive block.

An inter-coded block is encoded from a motion vector that points to a block of reference samples that forms a predictive block, and residual data that indicates differences between pixel values in the coded block and the reference samples in the predictive block. The intra-coded block is encoded according to an intra-coding mode and residual data. For further compression, the residual data may be transformed from the pixel domain to a transform domain, producing residual transform coefficients that may then be quantized. The quantized transform coefficients, initially arranged in a two-dimensional array, may be scanned in a particular order to generate a one-dimensional vector of transform coefficients for entropy coding.

Disclosure of Invention

In general, techniques are described for coding video data. This disclosure describes techniques for determining a set of merge candidates in a merge mode motion vector prediction process.

In some examples, this disclosure proposes generating a merge candidate set for a current prediction unit of a coding unit without comparing motion information of any merge candidate to motion information of any other prediction unit in the same coding unit. In this way, merge candidate sets for multiple prediction units of a coding unit may be generated in parallel, as generation of a particular merge candidate does not rely on comparison with motion vector information in other prediction units that may or may not have been determined.

This disclosure further proposes to remove a merge candidate included within another prediction unit of the same coding unit from the merge candidate set for the current prediction unit. In this way, there is a limited likelihood that all prediction units of one coding unit will use the same motion vector information, thereby preserving the advantages of partitioning a coding unit into multiple prediction units.

In one example of this disclosure, a method of coding video data includes determining a merge candidate set for a current prediction unit of a current coding unit, wherein the merge candidate set is determined without comparing motion information of merge candidates in the merge candidate set to motion information of any other prediction units, and performing a merge motion vector prediction process for the current prediction unit using the merge candidate set. The method may further include excluding a merge candidate within another prediction unit of the current coding unit from the merge candidate set.

In another example of this disclosure, an apparatus configured to code video data comprises a video coder configured to determine a merge candidate set for a current prediction unit of a current coding unit, wherein the merge candidate set is determined without comparing motion information of merge candidates in the merge candidate set to motion information of any other prediction units, and perform a merge motion vector prediction process for the current prediction unit using the merge candidate set. The video coder may be further configured to remove a merge candidate within another prediction unit of the current coding unit from the merge candidate set.

In another example of this disclosure, an apparatus configured to code video data comprises means for determining a merge candidate set for a current prediction unit of a current coding unit, wherein the merge candidate set is determined without comparing motion information of a merge candidate in the merge candidate set to motion information of any other prediction units, and means for performing a merge motion vector prediction process for the current prediction unit using the merge candidate set. The apparatus may further include means for excluding a merge candidate within another prediction unit of the current coding unit from the merge candidate set.

In another example of this disclosure, a computer-readable storage medium having stored thereon instructions that, when executed, cause one or more processors configured to code video data to operate is proposed. The instructions may cause the one or more processors to determine a merge candidate set for a current prediction unit of a current coding unit, wherein the merge candidate set is determined without comparing motion information of merge candidates in the merge candidate set to motion information of any other prediction units, and perform a merge motion vector prediction process for the current prediction unit using the merge candidate set. The instructions may further cause the one or more processors to remove a merge candidate that is within another prediction unit of the current coding unit from the merge candidate set.

The details of one or more examples are set forth in the accompanying drawings and the description below. Other features, objects, and advantages will be apparent from the description and drawings, and from the claims.

Drawings

Fig. 1 is a conceptual diagram illustrating candidate blocks for motion vector prediction according to a merge mode.

Fig. 2 is a conceptual diagram illustrating an example type of segmentation.

Fig. 3A is a conceptual diagram illustrating candidate blocks for merge mode motion vector prediction for nx 2N partitioning of a coding unit.

Fig. 3B is a conceptual diagram illustrating candidate blocks for merge mode motion vector prediction for 2 nxn partitioning of a coding unit.

Fig. 4A is a conceptual diagram illustrating candidate blocks for merge mode motion vector prediction for nxn partitioning of a coding unit.

Fig. 4B is a conceptual diagram illustrating another example of a candidate block for merge mode motion vector prediction for nxn partitioning of a coding unit.

Fig. 5 is a block diagram illustrating an example video coding system.

Fig. 6 is a block diagram illustrating an example video encoder.

Fig. 7 is a block diagram illustrating an example video decoder.

Fig. 8 is a flow diagram illustrating an example method of encoding video in accordance with the techniques of this disclosure.

Fig. 9 is a flow diagram illustrating an example method of decoding video in accordance with the techniques of this disclosure.

Detailed Description

In general, techniques are described for coding video data. This disclosure describes techniques for determining a merge candidate set in a merge mode motion vector prediction process.

Digital video devices implement video compression techniques to more efficiently encode and decode digital video information. Video compression may apply spatial (intra) prediction and/or temporal (inter) prediction techniques to reduce or remove redundancy inherent in video sequences.

There is a new video coding standard, High Efficiency Video Coding (HEVC), which was developed by the joint collaborative group of video coding (JCT-VC) of the ITU-T Video Coding Experts Group (VCEG) and the ISO/IEC Moving Picture Experts Group (MPEG). A recent draft of the HEVC standard (which is referred to as "HEVC working draft 6" or "WD 6") is described in the following document JCTVC-H1003 by Brass et al: "High Efficiency Video Coding (HEVC) text specification draft 6", video coding joint collaboration group (JCT-VC) of ITU-TSG16WP3 and ISO/IEC JTC1/SC29/WG11, conference 8: san jose, california, usa, month 2 2012, said document being available from http: int-evry. fr/jct/doc _ end _ user/documents/8_ San% 20Jose/wg11/JCTVC-H1003-v22.zip download.

For video coding according to the High Efficiency Video Coding (HEVC) standard currently being developed by the joint collaborative team on video coding (JCT-VC), a video frame may be partitioned into multiple coding units. A Coding Unit (CU) generally refers to a region of an image that serves as a basic unit to which various coding tools are applied to achieve video compression. A CU typically has a luma component, which may be represented as Y, and two chroma components, which may be represented as U and V. Depending on the video sampling format, the size of the U and V components, in terms of the number of samples, may be the same as or different from the size of the Y component. A CU is typically square and may be considered similar to a so-called macroblock under other video coding standards, such as ITU-T h.264, for example.

To achieve better coding efficiency, coding units may have variable sizes that depend on the video content. In addition, a coding unit may be split into multiple smaller blocks for prediction or transform. In particular, each coding unit may be further partitioned into multiple Prediction Units (PUs) and Transform Units (TUs). The prediction unit may be considered similar to so-called partitioning under other video coding standards, such as h.264. A Transform Unit (TU) refers to a block of residual data to which a transform is applied to generate transform coefficients.

For purposes of illustration, coding in accordance with some of the presently proposed aspects of the developing HEVC standard will be described in this application. However, the techniques described in this disclosure may be used for other video coding processes, such as video coding processes defined according to h.264 or other standards or proprietary video coding processes, and so on.

HEVC standardization efforts are based on a model of the video coding device known as the HEVC test model (HM). The HM assumes several capabilities of video coding devices over devices in accordance with, for example, ITU-T H.264/AVC. For example, whereas h.264 provides nine intra-prediction encoding modes, the HM provides up to thirty-four intra-prediction encoding modes.

According to the HM, a CU may include one or more Prediction Units (PUs) and/or one or more Transform Units (TUs). Syntax data within the bitstream may define a Largest Coding Unit (LCU), which is the largest CU in terms of the number of pixels. In general, a CU has a similar purpose to a macroblock of h.264, except that the CU has no size distribution. Thus, a CU may be split into multiple sub-CUs. In general, references to a CU in this disclosure may refer to a largest coding unit of a picture or a sub-CU of an LCU. The LCU may be split into a plurality of sub-CUs, and each sub-CU may be further split into a plurality of sub-CUs. Syntax data for a bitstream may define a maximum number of times an LCU may be split, referred to as CU depth. Thus, the bitstream may also define a minimum coding unit (SCU). This disclosure also uses the term "block" or "portion" to refer to any of a CU, PU, or TU. In general, a "portion" may refer to any subset of a video frame.

The LCU may be associated with a quadtree data structure. In general, the quadtree data structure includes one node per CU, where the root node corresponds to an LCU. If a CU is split into four sub-CUs, the node corresponding to the CU includes four leaf nodes, each of the four leaf nodes corresponding to one of the sub-CUs. Each node of the quadtree data structure may provide syntax data for the corresponding CU. For example, a node in a quadtree may include a split flag that indicates whether a CU corresponding to the node is split into multiple sub-CUs. Syntax elements for a CU may be defined in a recursive manner, and may depend on whether the CU is split into multiple sub-CUs. If a CU is not further split, then the CU is referred to as a leaf CU.

Furthermore, the TUs of a leaf-CU may also be associated with respective quadtree data structures. That is, a leaf-CU may include a quadtree that indicates how to partition the leaf-CU into multiple TUs. This disclosure refers to a quadtree indicating how an LCU is partitioned into CU quadtrees, and a quadtree indicating how a leaf-CU is partitioned into TUs as TU quadtrees. The root nodes of the TU quadtree generally correspond to leaf CUs, while the root nodes of the CU quadtree generally correspond to LCUs. TUs of a TU quadtree that is not split are referred to as leaf-TUs.

A leaf-CU may include one or more Prediction Units (PUs). In general, a PU represents all or a portion of a corresponding CU, and may include data for retrieving a reference sample for the PU. For example, when the PU is inter-mode encoded, the PU may include data defining a motion vector for the PU. The data defining a motion vector may describe, for example, a horizontal component of the motion vector, a vertical component of the motion vector, a resolution for the motion vector (e.g., one-quarter pixel precision or one-eighth pixel precision), a reference frame to which the motion vector points, and/or a reference list for the motion vector (e.g., list 0 or list 1). The data for defining a leaf-CU of a PU may also describe, for example, partitioning the CU into one or more PUs. The partition mode may be different depending on whether the CU is unpredictably coded, intra-prediction mode encoded, or inter-prediction mode encoded. For intra coding, a PU may be considered the same as the transform unit described below.

To code a block, e.g., a Prediction Unit (PU) of video data, a predictor for the block is first derived. The predictor may be derived via intra (I) prediction (i.e., spatial prediction) or inter (P or B) prediction (i.e., temporal prediction). Thus, some prediction units may be inter-coded (I) using spatial prediction with respect to neighboring reference blocks in the same frame, and other prediction units may be inter-coded (P or B) with respect to reference blocks in other frames.

After the predictor is identified, the difference between the original block of video data and its predictor is calculated. This difference is also referred to as a prediction residual, and refers to the pixel difference (i.e., predictor) between the pixels of the block to be coded and the corresponding reference samples of the reference block, which may be integer-precision pixels or interpolated fractional-precision pixels. To achieve better compression, the prediction residual (i.e., the pixel difference array) is generally transformed, for example, using a Discrete Cosine Transform (DCT), an integer transform, a Karhunen-Loeve (K-L) transform, or other transform.

Coding a PU using inter prediction involves calculating a motion vector between a current block and a block in a reference frame. Motion vectors are computed via a process called motion estimation (or motion search). A motion vector, for example, may indicate the displacement of a prediction unit in a current frame relative to reference samples of a reference frame. The reference samples may be blocks that find portions that closely match the CU including the coded PU in terms of pixel differences, which may be determined by Sum of Absolute Differences (SAD), Sum of Squared Differences (SSD), or other difference metrics. The reference sample may occur anywhere within the reference frame or reference slice. In some examples, the reference sample may be interpolated in whole or in part, and the reference sample occurs at a fractional pixel position. After finding a portion of the reference frame that best matches the current portion, the encoder determines the current motion vector for the current portion as the difference in position from the current portion to the matching portion in the reference frame (e.g., from the center of the current portion to the center of the matching portion).

In some examples, the encoder may signal motion vectors for each portion in the encoded video bitstream. The signaled motion vectors are used by a decoder to perform motion compensation in order to decode the video data. However, signaling the original motion vectors directly may result in less efficient coding because a large number of bits are typically required to convey the information.

In some cases, rather than signaling the original motion vectors directly, the encoder may predict the motion vectors for each PU. In this disclosure, the term "block" may be used generically to refer to a CU, PU, or TU. In performing motion vector prediction, the encoder may select a set of candidate motion vectors determined for spatially neighboring blocks in the same frame as the current PU or candidate motion vectors determined for collocated PUs in a reference frame. The encoder may perform motion vector prediction to select a particular candidate motion vector, and signal a syntax element indicating the selected motion vector candidate to reduce the bit rate in the signaling, if needed. Candidate motion vectors from spatially neighboring blocks may be referred to as spatial MVP candidates, while candidate motion vectors from collocated blocks in another reference frame may be referred to as temporal MVP candidates.

The techniques of this disclosure are directed to a "merge" mode of motion vector prediction. In merge mode, the video encoder signals, through the bitstream of the prediction syntax, the decoder to copy the motion vectors, the reference indices (which identify the reference frames in a given reference picture list to which the motion vectors point), and the motion prediction direction (which identifies the reference picture list (list 0 or list 1), i.e., whether the reference frame temporally precedes or succeeds the current frame) (from the selected candidate motion vectors for the current portion of the frame). This is done by signaling an index into the bitstream to the set of candidate motion vectors that identifies the selected candidate motion vector (i.e., the particular spatial MVP candidate or temporal MVP candidate). The set of candidate motion vectors may be derived from preset settings or inferred from certain coding parameters. Thus, for merge mode, the prediction syntax may include a flag that identifies the mode (in this case, "merge" mode), as well as an index that identifies the selected candidate motion vector. In some cases, the candidate motion vector will be in a PU that references the representative cause of the current PU. I.e. the candidate motion vectors will have been decoded by the decoder. Thus, the decoder has received and/or determined the motion vector, reference index, and motion prediction direction for the PU representing the cause. As such, the decoder may retrieve only the motion vector, reference index, and motion prediction direction associated with the PU representing the cause from memory and copy these values for the current PU. To decode a block in merge mode, the decoder obtains a predictor block using motion vector prediction, and adds residual data to the predictor block to reconstruct the coded block.

Once motion estimation is performed to determine a motion vector for the current portion, the encoder compares the matching portion in the reference frame to the current portion. This comparison typically involves subtracting the portion in the reference frame (which is commonly referred to as the "reference sample") from the current portion and generating so-called residual data, as mentioned above. The residual data indicates pixel differences between the current portion and the reference sample. The encoder then transforms this residual data from the spatial domain to a transform domain, such as the frequency domain. Typically, the encoder applies a Discrete Cosine Transform (DCT) to the residual data to accomplish this transform. The encoder performs this transform in order to facilitate compression of the residual data, since the resulting transform coefficients represent different frequencies, with most of the energy typically concentrated on a few low frequency coefficients.

Typically, the resulting transform coefficients are grouped together in a manner that enables variable length coding, especially if the transform coefficients are first quantized (rounded). The encoder performs this run-length encoding of the quantized transform coefficients and then performs statistical lossless (or so-called "entropy") encoding to further compress the run-length coded quantized transform coefficients.

After performing lossless entropy coding, the encoder generates a bitstream that includes the encoded video data. This bitstream also includes, in some cases, a number of prediction syntax elements that specify, for example, whether to perform motion vector prediction, a motion vector mode, and a Motion Vector Predictor (MVP) index (i.e., an index of the candidate portion having the selected motion vector). The MVP index may also be referred to as its syntax element variable name "MVP _ idx".

Fig. 1 shows a set 90 of candidate motion vectors currently proposed in the HEVC standard for use in merge mode. The merge mode uses six merge candidates from the following spatial and temporal blocks: a lower left (BL) block 91, a left (L) block 92, an upper Right (RA) block 93, an upper (a) block 94, an upper Left (LA) block 95, and a time block (T) 96. The candidate motion vectors associated with these blocks are used to determine the motion vector predictor in merge mode.

The time block 96 may be at a different time than the current PU (e.g., T)₂) Is different from the current PU108 (e.g., T) within or adjacent to the collocated block in the frame₁) In the frame of (2) are collocated blocks. The locations of the spatial candidate blocks (i.e., BL, L, LA, a, and RA) depicted in fig. 1 are not absolute locations, but are based on the relative locations with respect to the current PU98 as generally defined below. It should be noted that the candidate block need not be the most recent possible block that satisfies the definition below, but may be any PU that satisfies the definition. L isThe a candidate block 95 is located above the top line defining the current PU and to the left of the left line defining the current PU. The L candidate block 92 is located to the left of the left line defining the current PU, and is located above the bottom line defining the current PU and below the top line defining the current PU. The BL candidate block 91 is located below the bottom line defining the current PU and to the left of the left line defining the current PU. The a candidate block 94 is located above the top line defining the current PU and is located to the right of the left line defining the current PU and to the left of the right line defining the current PU. The RA candidate block 93 is located to the right of the right line defining the current PU and above the top line defining the current PU.

Each PU generates multiple merge candidates. That is, each PU has its own set of merge candidates. This case includes PUs partitioned from larger CUs. The example of fig. 1 is for a 2 nx 2N partitioned PU (e.g., a square PU). Fig. 2 shows other examples of prediction units having different partition types. As shown in fig. 2, the 2N × 2N partition is a square partition. Basically, it is a PU from an undivided CU. The 2 nxn partition is made by dividing a square CU into two horizontally oriented PUs, with PU0 above PU 1. The nx2N partition is made by dividing a square CU into two vertically oriented PUs, with PU0 to the left of PU 1. An nxn partition is made by dividing a square CU into four equally sized PUs. In nxn partitioning, PU0 is above-left of the CU, PU1 is above-right of the CU, PU2 is below-left of the CU, and PU3 is below-right of the CU.

FIG. 2 shows an additional type of "non-square" partitioning. The 2 nxnd partition is a horizontally-oriented non-square partition type, where the lower PU (PU1) has a smaller size (i.e., a quarter size of the CU size) than the upper PU (PU 0). The 2 nxnu partition is a horizontally-oriented non-square partition type, where the lower PU (PU1) has a larger size (i.e., three-quarters of the CU size) than the upper PU (PU 0). The nL × 2N partition is a vertically-oriented non-square partition type, where the left-side PU (PU0) has a smaller size (i.e., a quarter size of the CU size) than the right-side PU (PU 1). nR × 2N partitioning is a vertically oriented non-square partitioning type, where the left-side PU (PU0) has a larger size (i.e., three-quarters of the CU size) than the right-side PU (PU 1). These segmentation examples are sometimes referred to as asymmetric motion segmentation (AMP).

A CU is partitioned according to one of the partition types to provide more accurate inter prediction (temporal prediction). Motion information is signaled separately for each partition. In the case of finer partitions (e.g., 2N × N partitions finer than 2N × 2N partitions), it may be possible to derive better quality predictors for each partition. On the other hand, since motion information is signaled separately for each partition, the signaling overhead of a CU in the case of finer partitions is also relatively high. In practice, determining the partition type for the current CU is often based on rate-distortion optimization. The type of segmentation chosen is a trade-off between prediction accuracy and signaling overhead. Current proposals for the HEVC standard implement techniques to avoid using redundant merge candidates for PUs of the same CU. A redundant merge candidate is a merge candidate that has the same motion information as another PU in the same CU. For a particular partition type, one of the merge candidates for PU1 (or PU1, PU2, and PU3 for N × N partitions) is compared to the motion information of PU0 (or PU0, PU1, and PU2 for N × N partitions) to avoid the entire CU using the same motion information. If each PU in a CU uses the same motion information, the result will be a duplicate of the 2N × 2N partition type (i.e., no partition). Therefore, the advantage of partitioning CUs for more accurate inter prediction will be negated.

To avoid using redundant merge candidates, one proposal for HEVC compares the motion information of each merge candidate in the merge candidate set with the motion information of other PUs of the same CU. Any merge candidate that has the same motion information as the previously coded PU is removed from the merge candidate set to avoid using the entire CU of the same motion information.

According to this technique, the process for generating a merge candidate set for a PU is as follows:

1. checking motion information for next candidate block

2. Comparing candidate motion information to motion information of previously coded PUs in the same CU

3. If the candidate motion information of the candidate block is the same as the motion information of the previously coded PU, go to step 1; otherwise, go to step 4

4. Adding candidate blocks to a merge candidate set

5. If all candidate blocks have been checked, the process ends; otherwise, go to step 1

In general, the restriction on merging candidates in this process case yields the following results for 2 nx N, N × 2N and nxn partition types:

1)2 NXN/NX2N case: a merge candidate for the second PU (PU1) is set to unavailable if the merge candidate has the same motion information as the motion information of the first PU (PU 0).

2) N × N case:

pu0 and PU1 have the same motion information. Set a merge candidate for PU3 as unavailable if the merge candidate has the same motion information as PU 2;

PU0 and PU2 have the same motion information. If the merge candidate of PU3 has the same motion information as PU1, the merge candidate is set as unavailable.

While this process does eliminate redundant merge candidates, this process requires that all PUs be encoded/decoded before developing a merge candidate set for a subsequent PU. Thus, the ability to process multiple PUs of a CU in parallel is limited. The encoder/decoder must determine the final motion information of all previous PUs in the same CU before constructing the merge candidate set for the current PU. Furthermore, the comparison operation for each candidate block may increase the computational complexity of the encoder/decoder.

In one example, the present disclosure proposes to remove the compare-check operation during the generation of the merge candidate set, thereby making the merge candidate generation parallelization friendly. The disclosed techniques eliminate the need to compare motion information between candidate blocks and other PUs of a CU. Thus, the merge candidate sets for all PUs of a CU may be generated in parallel. The disclosed processes may also reduce the computational complexity of encoding and decoding.

The proposed process for each PU is

1. Check the next candidate block

2. Adding a candidate block to a candidate set

3. If all the neighboring blocks have been checked, the process ends; otherwise, go to step 1

This process provides a unified solution for all PUs, without considering the indices of the prediction units (e.g., PU0, PU1) and without making a comparison of the motion information of the candidate block to the motion information of the previously coded PU. The motion information comparison step described above may be eliminated. Although described repeatedly, the steps of the method may be performed in parallel. For example, a first thread of a parallel process may include instructions to perform a first instance of steps 1 and 2, and a second, different thread of the parallel process may include instructions to perform a second instance of steps 1 and 2. Additional threads may also be provided.

Based on the proposed techniques, merge candidates that are inside a previous PU may be included into a merge candidate set. However, this may cause the entire CU to use the same motion information. As such, a partitioned CU may end up with the same coded motion vectors as a 2 nx 2N partition, and the benefits of partitioning for inter prediction may be limited. Additionally, including such redundant candidate blocks in the merge candidate set may cause some performance degradation because the extra bits are used to signal redundant merge candidates. Thus, the invention also proposes: in case the merge candidate is located inside another PU of the same CU, the merge candidate is removed from the merge candidate set.

Fig. 3A is a conceptual diagram illustrating candidate blocks of a merge mode for nx2N partitioning of a CU. It should be noted that the technique shown in fig. 3A is equally applicable to nL × 2N or nR × 2N asymmetric partition types. The merge candidate set 100 shows the merge candidates for PU0 of an nx 2N partitioned CU. Since no merge candidate in the merge candidate set 100 is within another PU of the same CU, all merge candidates may remain in the merge candidate set 100. The merge candidate set 102 shows the merge candidates for PU1 of an nx 2N partitioned CU. As can be seen, for the merge set 102 for PU1, the merge candidate L is from PU0 of the same CU. Thus, the merge candidate L may be removed/excluded from the merge candidate set 102. In this context, a removed merge candidate may be considered a candidate removed from the predefined list of merge candidates. The excluded merge candidate may be a merge candidate that is excluded from the merge candidate list when deriving the merge candidate list, regardless of whether the list is predefined. In general, a removed/excluded merge candidate is any merge candidate that is not used in the final merge candidate list.

It should be noted that merge candidate L need not be located in the exact location in PU0 (e.g., if PU0 is further partitioned), but may be excluded if it is located in any portion of PU 0. It should also be noted that each of the merge candidate sets 100 and 102 also has a temporal merge candidate T, as shown in fig. 1.

Fig. 3B is a conceptual diagram illustrating candidate blocks of a merge mode for 2 nxn partitioning of a CU. It should be noted that the technique shown in fig. 3B is equally applicable to 2N × nU or 2N × nD asymmetric partition types. The merge candidate set 104 shows the merge candidates for PU0 of a 2 nxn partitioned CU. Since no merge candidate in the merge candidate set 104 is within another PU of the same CU, all merge candidates may remain in the merge candidate set 104. The merge candidate set 106 shows the merge candidates for PU1 of a 2 nxn partitioned CU. As can be seen, for the merge set 106 for PU1, merge candidate a is from PU0 of the same CU. Thus, merge candidate a may be removed/excluded from the merge candidate set 106. It should be noted that merge candidate a need not be located in the exact location in PU0 (as shown), e.g., if PU0 is further partitioned, but may be excluded if it is located in any portion of PU 0. It should also be noted that each of the merge candidate sets 104 and 106 also has a temporal merge candidate T, as shown in fig. 1.

Fig. 4A is a conceptual diagram illustrating candidate blocks of a merge mode for nxn partitioning of a CU. The merge candidate set 108 shows the merge candidates for PU0 of an nxn partitioned CU. Since no merge candidate in the merge candidate set 108 is within another PU of the same CU, all merge candidates may remain in the merge candidate set 108.

The merge candidate set 110 shows the merge candidates for PU1 of an nxn partitioned CU. As can be seen, for the merge set 110 for PU1, the merge candidates L and BL are from PU0 and PU2, respectively, of the same CU. Thus, the merge candidates L and BL may be removed/excluded from the merge candidate set 110. It should be noted that the merge candidates L and BL need not be located in the exact positions in PU0 and PU2 (as shown), e.g., if PU0 or PU2 is further partitioned, but may be excluded if the merge candidates L and/or BL are located in any portion of PU0 and/or PU 2.

The merge candidate set 112 shows the merge candidates for PU2 of an nxn partitioned CU. As can be seen, for the merge set 112 for PU2, the merge candidates a and RA are from PU0 and PU1, respectively, of the same CU. Thus, merge candidates a and RA may be removed/excluded from the merge candidate set 112. It should be noted that merge candidates a and RA need not be located in exact positions in PU0 and PU1 (as shown), e.g., if PU0 or PU1 is further partitioned, but may be excluded if they are located in any portion of PU0 and/or PU 1.

The merge candidate set 114 shows the merge candidates for PU3 of an nxn partitioned CU. As can be seen, for the merge set 114 for PU3, the merge candidates LA, a, and L are from PU0, PU1, and PU2, respectively, of the same CU. Thus, the merge candidates LA, a, and L may be removed/excluded from the merge candidate set 114. It should be noted that merge candidates LA, a, and L need not be located in the exact locations in PU0, PU1, and PU2 (as shown) (e.g., where PU0, PU1, or PU2 is further partitioned), but may be excluded if they are located in any portion of PU0, PU1, and/or PU 2.

It should be noted that each of the merge candidate sets 108, 110, 112, and 114 also has a temporal merge candidate T, as shown in fig. 1.

Although the examples described above only consider 2 nx N, N x2N and nxn segmentation types, other segmentation types (e.g., AMP, geometric motion segmentation (GMP), etc.) may also benefit from the disclosed techniques. In general, the proposed techniques determine a merge candidate set without comparing motion information of the merge candidate to motion information of any other prediction units. Furthermore, if a merge candidate within the merge candidate set for the current prediction unit is within another prediction unit of the same coding unit, the merge candidate may be removed/excluded.

In another example, for an nxn partition mode, all merge candidates for each prediction unit are used regardless of any merge candidates used by another prediction unit. Fig. 4B is a conceptual diagram illustrating an example candidate block for a merge mode of nxn partitioning of a coding unit, where no merge candidate is removed/excluded regardless of its location in another PU. As shown in fig. 4B, all candidates (including temporal candidate T) are used for each of PU116, PU118, PU120, and PU 122. For other partitioning modes (e.g., 2 nx N, N × 2N, etc.) as well as asymmetric modes (e.g., 2 nxnd, 2 nxnu, nL × 2N, nR × 2N, etc.), the exclusion of certain candidates for a current PU that is located inside another PU of the same CU is applied according to the process described above.

In another example of this disclosure, in the case that prediction unit 0 in a particular partition mode uses a merge candidate that is not a duplicate of the merge candidate to be used in the 2 nx 2N partition mode (i.e., the candidate actually selected in motion vector prediction), a merge candidate set is generated that utilizes all possible merge candidates in the set shown in fig. 1 for prediction unit 1 in the nx 2N and 2 nx N partition modes. In another example, for an nxn partition mode, all merge candidates are used regardless of any merge candidate used by another prediction unit.

The rule for generating the merge candidate set according to this example is as follows:

n × 2N partitioning mode: if the merge index for PU0 (i.e., the actually selected merge candidate) is RA, a, or T, then the left (L) merge candidate of PU1 is used; otherwise, L is not used (in contrast to the previous example of the present invention, in which L was not used for PU 1).

2.2N × N partition mode: if the merge index for PU0 is BL, L, or T, then the upper (A) merge candidate for PU1 is used; otherwise, a is not used (in contrast to the previous example of the present invention, in which a was not used for PU 1).

N × N partition mode: considering all prediction candidates for all PUs as valid

In the nx 2N example, utilizing the merge candidate L for PU1 does not become a duplicate of the 2 nx 2N partition mode, since the merge candidate location RA, a, or T for PU0 in the nx 2N partition mode will not necessarily be used for the 2 nx 2N partition. Likewise, in the 2 nx N example, utilizing merge candidate a for PU1 does not become a duplicate of the 2 nx 2N partition mode, since merge candidates BL, L, or T for PU0 will not necessarily be used for the 2 nx 2N partition.

Fig. 5 is a block diagram illustrating an example video encoding and decoding system 10 that may be configured to utilize techniques for generating candidate motion vectors in merge mode, according to an example of this disclosure. As shown in fig. 5, system 10 includes a source device 12 that transmits encoded video to a destination device 14 via a communication channel 16. The encoded video data may also be stored on storage medium 34 or file server 36 and may be accessed by destination device 14 when needed. When stored to a storage medium or file server, video encoder 20 may provide the coded video data to another device, such as a network interface, a Compact Disc (CD), blu-ray or Digital Video Disc (DVD) recorder or press facility device, or other device, for storage of the coded video data to the storage medium. Likewise, a device separate from video decoder 30 (e.g., a network interface, CD or DVD reader, etc.) may retrieve coded video data from the storage medium and provide the retrieved data to video decoder 30.

Source device 12 and destination device 14 may comprise any of a wide variety of devices, including mobile devices, desktop computers, notebook computers (i.e., laptop computers), tablet computers, set-top boxes, telephone handsets (e.g., so-called smartphones, etc.), televisions, cameras, display devices, digital media players, video game consoles, and so forth. In many cases, such devices may be equipped for wireless communication. Thus, communication channel 16 may comprise a wireless channel, a wired channel, or a combination of a wireless channel and a wired channel suitable for transmitting encoded video data. Similarly, the file server 36 may be accessed by the destination device 14 over any standard data connection, including an internet connection. This may include a wireless channel (e.g., a Wi-Fi connection), a wired connection (e.g., DSL, cable modem, etc.), or a combination of both, suitable for accessing encoded video data stored on a file server.

The techniques for generating candidate motion vectors in merge mode according to examples of this disclosure may be applicable to support video coding for any of a variety of multimedia applications, such as over-the-air television broadcasts, cable television transmissions, satellite television transmissions, streaming video transmissions (e.g., via the internet), encoding for digital video stored on a data storage medium, decoding of digital video stored on a data storage medium, or other applications. In some examples, system 10 may be configured to support one-way or two-way video transmission to support applications such as video streaming, video playback, video broadcasting, and/or video telephony.

In the example of fig. 5, source device 12 includes a video source 18, a video encoder 20, a modulator/demodulator 22, and a transmitter 24. In source device 12, video source 18 may include sources such as: a video capture device (e.g., a video camera, etc.), a video archive containing previously captured video, a video feed interface that receives video from a video content provider, and/or a computer graphics system for generating computer graphics data as source video, or a combination of such sources. As one example, if video source 18 is a video camera, source device 12 and destination device 14 may form so-called camera phones or video phones. However, the techniques described in this disclosure may be applicable to video coding in general, and may be applicable to wireless and/or wired applications, or applications where encoded video data is stored on a local disk.

The captured video, the pre-captured video, or the computer-generated video may be encoded by video encoder 20. The encoded video information may be modulated by modem 22 according to a communication standard, such as a wireless communication protocol, and transmitted to destination device 14 via transmitter 24. Modem 22 may include various mixers, filters, amplifiers, or other components designed for signal modulation. Transmitter 24 may include circuitry designed for transmitting data, including amplifiers, filters, and one or more antennas.

Captured video, pre-captured video, or computer-generated video encoded by video encoder 20 may also be stored onto storage medium 34 or file server 36 for later consumption. Storage medium 34 may include a blu-ray disc, DVD, CD-ROM, flash memory, or any other suitable digital storage medium for storing encoded video. The encoded video stored on storage medium 34 may then be accessed by destination device 14 for decoding and playback.

File server 36 may be any type of server capable of storing encoded video and transmitting the encoded video to destination device 14. Example file servers include web servers (e.g., for a website), FTP servers, Network Attached Storage (NAS) devices, local disk drives, or any other type of device capable of storing and transmitting encoded video data to a destination device. The transmission of the encoded video data from the file server 36 may be a streaming transmission, a download transmission, or a combination of both. The file server 36 may be accessed by the destination device 14 over any standard data connection, including an internet connection. This may include a wireless channel (e.g., a Wi-Fi connection), a wired connection (e.g., DSL, cable modem, ethernet, USB, etc.), or a combination of both, suitable for accessing encoded video data stored on a file server.

In the example of fig. 5, destination device 14 includes a receiver 26, a modem 28, a video decoder 30, and a display device 32. Receiver 26 of destination device 14 receives the information over channel 16, and modem 28 demodulates the information to generate a demodulated bitstream for video decoder 30. The information communicated over channel 16 may include a variety of syntax information generated by video encoder 20 for use by video decoder 30 in decoding video data. This syntax may also be included in the encoded video data stored on the storage medium 34 or the file server 36. Each of video encoder 20 and video decoder 30 may form part of a respective encoder-decoder (CODEC) capable of encoding or decoding video data.

The display device 32 may be integrated with the destination device 14 or external to the destination device 14. In some examples, destination device 14 may include an integrated display device and also be configured to interface with an external display device. In other examples, destination device 14 may be a display device. In general, display device 32 displays the decoded video data to a user, and may comprise any of a variety of display devices, such as a Liquid Crystal Display (LCD), a plasma display, an Organic Light Emitting Diode (OLED) display, or another type of display device.

In the example of fig. 5, communication channel 16 may comprise any wireless or wired communication medium, such as a Radio Frequency (RF) spectrum or one or more physical transmission lines, or any combination of wireless and wired media. The communication channel 16 may form part of a packet-based network, such as a local area network, a wide area network, or a global network such as the internet, among others. Communication channel 16 generally represents any suitable communication medium or collection of different communication media for transmitting video data from source device 12 to destination device 14, including any suitable connection of wired or wireless media. Communication channel 16 may include a router, switch, base station, or any other equipment that may be used to facilitate communication from source device 12 to destination device 14.

Video encoder 20 and video decoder 30 may operate in accordance with a video compression standard, such as the High Efficiency Video Coding (HEVC) standard currently under development, and may conform to the HEVC test model (HM). Alternatively, video encoder 20 and video decoder 30 may operate in accordance with other proprietary or industry standards, such as the ITU-T h.264 standard, otherwise known as MPEG-4 part 10, Advanced Video Coding (AVC), or an extension of such a standard. However, the techniques of this disclosure are not limited to any particular coding standard. Other examples include MPEG-2 and ITU-T H.263.

Although not shown in fig. 5, in some aspects, video encoder 20 and video decoder 30 may each be integrated with an audio encoder and decoder, and may include appropriate MUX-DEMUX units, or other hardware and software, to handle encoding of both audio and video in a common data stream or separate data streams. Where applicable, in some examples, the MUX-DEMUX unit may conform to the ITU h.223 multiplexer protocol, or other protocols such as the User Datagram Protocol (UDP).

Video encoder 20 and video decoder 30 may each be implemented as any of a variety of suitable encoder circuits, such as one or more microprocessors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field Programmable Gate Arrays (FPGAs), discrete logic, software, hardware, firmware, or any combinations thereof. When the techniques are implemented in part in software, a device may store instructions for the software in a suitable non-transitory computer-readable medium and execute the instructions in hardware using one or more processors to perform the techniques of this disclosure. Each of video encoder 20 and video decoder 30 may be included in one or more encoders or decoders, either of which may be integrated as part of a combined encoder/decoder (CODEC) in the respective device.

Video encoder 20 may implement any or all of the techniques of this disclosure for generating candidate motion vectors in merge mode in a video encoding process. Likewise, video decoder 30 may implement any or all of these techniques for generating candidate motion vectors in merge mode in a video decoding process. As described in this disclosure, a video coder may refer to a video encoder or a video decoder. Similarly, a video coding unit may refer to a video encoder or a video decoder. Likewise, video coding may refer to video encoding or video decoding.

In one example of this disclosure, video encoder 20 of source device 12 may be configured to determine a merge candidate set for a current prediction unit of a current coding unit, wherein the merge candidate set is determined without comparing motion information of merge candidates in the merge candidate set to motion information of any other prediction units, and perform a merge motion vector prediction process for the current prediction unit using the merge candidate set. Video encoder 20 may be further configured to remove a merge candidate that is within another prediction unit of the current coding unit from the merge candidate set.

In another example of this disclosure, video decoder 30 of source device 12 may be configured to determine a merge candidate set for a current prediction unit of a current coding unit, wherein the merge candidate set is determined without comparing motion information of merge candidates in the merge candidate set to motion information of any other prediction units, and perform a merge motion vector prediction process for the current prediction unit using the merge candidate set. Video decoder 30 may be further configured to remove a merge candidate within another prediction unit of the current coding unit from the merge candidate set.

Fig. 6 is a block diagram illustrating an example of video encoder 20, video encoder 20 may use techniques for generating candidate motion vectors in merge mode as described in this disclosure. For purposes of illustration, video encoder 20 will be described in the context of HEVC coding, but there are no limitations of this disclosure with respect to other coding standards or methods.

Video encoder 20 may perform intra and inter coding of CUs within video frames. Intra-coding relies on spatial prediction to reduce or remove spatial redundancy in video data within a given video frame. Inter-coding relies on temporal prediction to reduce or remove temporal redundancy between a current frame and a previously coded frame of a video sequence. Intra mode (I-mode) may refer to any of a number of spatially based video compression modes. An inter mode, such as uni-directional prediction (P-mode) or bi-directional prediction (B-mode), may refer to any of a number of temporally-based video compression modes.

As shown in fig. 6, video encoder 20 receives a current video block within a video frame to be encoded. In the example of fig. 6, video encoder 20 includes motion compensation unit 44, motion estimation unit 42, intra-prediction unit 46, reference frame buffer 64, summer 50, transform module 52, quantization unit 54, and entropy encoding unit 56. Transform module 52 illustrated in fig. 6 is a structure or apparatus that applies the actual transform or a combination of transforms to a block of residual data, and should not be confused with a block of transform coefficients, which may be referred to as a Transform Unit (TU) of a CU. For video block reconstruction, video encoder 20 also includes inverse quantization unit 58, inverse transform module 60, and summer 62. A deblocking filter (not shown in fig. 6) may also be included to filter block boundaries to remove blockiness artifacts from the reconstructed video. The deblocking filter will typically filter the output of summer 62, if desired.

During the encoding process, video encoder 20 receives a video frame or slice to be coded. A frame or slice may be divided into a plurality of video blocks, e.g., Largest Coding Units (LCUs). Motion estimation unit 42 and motion compensation unit 44 perform inter-predictive coding of the received video block relative to one or more blocks in one or more reference frames to provide temporal compression. Intra-prediction unit 46 may perform intra-predictive coding of the received video block relative to one or more neighboring blocks in the same frame or slice as the block to be coded to provide spatial compression.

Mode select unit 40 may select one of the coding modes (intra-mode or inter-mode), e.g., based on the error (i.e., distortion) result for each mode, and provide the resulting intra-or inter-predicted block (e.g., Prediction Unit (PU)) to summer 50 to generate residual block data and to summer 62 to reconstruct the encoded block for use in the reference frame. Summer 62 combines the predicted block with the inverse quantized inverse transformed data for the block from inverse transform module 60 to reconstruct the encoded block, as described in more detail below. Some video frames may be designated as I-frames, where all blocks in an I-frame are encoded in intra-prediction mode. In some cases, intra-prediction unit 46 may perform intra-prediction encoding of blocks in P-frames or B-frames, for example, when the motion search performed by motion estimation unit 42 does not yield sufficient prediction for the block.

Motion estimation unit 42 and motion compensation unit 44 may be highly integrated, but are illustrated separately for conceptual purposes. Motion estimation (or motion search) is a process that generates motion vectors that estimate the motion of video blocks. A motion vector, for example, may indicate the displacement of a prediction unit in a current frame relative to reference samples of a reference frame. Motion estimation unit 42 calculates motion vectors for prediction units of inter-coded frames by comparing the prediction units to reference samples of reference frames stored in reference frame buffer 64. The reference samples may be blocks that find portions that closely match the CU including the coded PU in terms of pixel differences, which may be determined by Sum of Absolute Differences (SAD), Sum of Squared Differences (SSD), or other difference metrics. The reference sample may occur anywhere within the reference frame or reference slice.

The portion of the reference frame identified by the motion vector may be referred to as a reference sample. Motion compensation unit 44 may calculate a prediction value for the prediction unit of the current CU, e.g., by retrieving a reference sample identified by the motion vector for the PU. In some video coding techniques, motion estimation unit 42 sends the calculated motion vectors, reference frames, and prediction directions (e.g., directions depending on whether the reference frame temporally precedes or succeeds the current frame) to entropy encoding unit 56 and motion compensation unit 44. Other video coding techniques utilize a motion vector prediction process to encode motion vectors. The motion vector prediction process may be selected from among a plurality of modes including the merge mode.

In merge mode, the encoder considers the set of merge candidate blocks and selects the block with the same (or closest match) motion vector, reference frame, and prediction direction as the current block to be coded. This is accomplished, for example, by: each candidate block is checked in turn and the candidate block that yields the best rate-distortion performance once its motion vector, reference frame and prediction direction are copied to the current block is selected. Then, rather than signaling this motion vector information (i.e., motion vectors, reference frames, and prediction directions) in the encoded video bitstream, the encoder signals an index number for the selected motion vector candidate. The index number identifies a selected candidate motion vector from among a set of candidate motion vectors. The decoder may copy the motion vector information from the motion vector candidate for the current block.

In the examples described above, signaling motion vector information in an encoded bitstream does not necessarily require transmitting such elements from the encoder to the decoder in real-time, but means: this information is encoded into the bitstream and made accessible to the decoder in any manner. This may include real-time transmission (e.g., in a video conference) as well as storing the encoded bitstream on a computer-readable medium for future use by the decoder (e.g., streaming, downloading, disk access, card access, DVD, blu-ray disc, etc.).

According to the example of this disclosure described above, for merge mode, a merge candidate set may be generated without comparing the motion information of any merge candidate to the motion information of other PUs within the same CU as the current PU. In addition, the invention also proposes: in case the merge candidate is located inside another PU of the same CU, the merge candidate is removed from the merge candidate set. The generation of merge candidates may be handled by motion compensation unit 44, motion compensation unit 42, or by any other fixed-function or programmable hardware structure of video encoder 20.

As one example, for an nx2N partition of a CU, all merge candidates (e.g., the merge candidates shown in fig. 1) are available to PU 0. For PU1, when merge candidate L is within PU0, merge candidate L is removed/excluded from the merge candidate list (see fig. 3A). As another example, for a 2 nxn partition of a CU, all merge candidates (e.g., the merge candidate shown in fig. 1) are available to PU 0. For PU1, when merge candidate a is within PU0, merge candidate a is removed from the merge candidate list (see fig. 3B).

As another example, for nxn partitioning of a CU, all merge candidates (e.g., the merge candidate shown in fig. 1) are available to PU 0. For PU1, when merge candidates L and BL are within PU0 and PU2, respectively, merge candidates L and BL are removed/excluded from the merge candidate list (see fig. 4A). For PU2, when merge candidate a and RA are within PU0 and PU1, respectively, merge candidate a and RA are removed/excluded from the merge candidate list (see fig. 4A). For PU3, when merge candidates LA, a, and L are within PU0, PU1, and PU2, respectively, merge candidates LA, a, and L are removed/excluded from the merge candidate list (see fig. 4A). Thus, PU0 may use the merge candidates BL, L, LA, a, RA, and T. PU1 may use merge candidates LA, a, RA, and T. PU2 may use merge candidates BL, L, LA, and T. PU3 may use merge candidates BL, RA, and T.

As yet another example, for an nxn partition mode, all merge candidates for each prediction unit are used regardless of any merge candidates used by another prediction unit (see fig. 4B). For other partitioning modes (e.g., 2 nx N, N × 2N, etc.) as well as asymmetric modes (e.g., 2 nxnd, 2 nxnu, nL × 2N, nR × 2N, etc.), the exclusion of certain candidates for a current PU that is located inside another PU of the same CU is applied according to the process described above.

Returning to fig. 6, intra-prediction unit 46 may perform intra-prediction on the received block as an alternative to inter-prediction performed by motion estimation unit 42 and motion compensation unit 44. Intra-prediction unit 46 may predict the received block (assuming left-to-right, top-to-bottom encoding order of the block) relative to neighboring previously coded blocks (e.g., blocks above, above-right, above-left, and left of the current block). Intra-prediction unit 46 may be configured with a variety of different intra-prediction modes. For example, intra-prediction unit 46 may be configured to have a certain number of directional prediction modes, e.g., thirty-four directional prediction modes, based on the size of the CU being encoded.

Intra-prediction unit 46 may select the intra-prediction mode by, for example, calculating prediction error values for various intra-prediction modes and selecting the mode that yields the lowest error value. The directional prediction mode may include functions for combining values of spatially neighboring pixels and applying the combined values to one or more pixel locations in the PU. Once values for all pixel locations in the PU have been calculated, intra-prediction unit 46 may calculate an error value for the prediction mode based on pixel differences between the calculated or predicted values of the PU and the received original block to be encoded. Intra-prediction unit 46 may continue to test intra-prediction modes until an intra-prediction mode is found that yields an acceptable error value. Intra-prediction unit 46 may then send the PU to summer 50.

Video encoder 20 forms a residual block by subtracting the prediction data calculated by motion compensation unit 44 or intra-prediction unit 46 from the coded original video block, which may include one luma block and two chroma blocks. Summer 50 represents one or more components that perform this subtraction operation. The residual block may correspond to a two-dimensional matrix of pixel difference values, where the number of values in the residual block is the same as the number of pixels in the PU corresponding to the residual block. The values in the residual block may correspond to differences (i.e., errors) between the PU and the values of the collocated pixels in the original block to be coded. This operation is applied to both luma and chroma components, so the difference value may be either a chroma difference or a luma difference depending on the type of block being coded.

Transform module 52 may form one or more Transform Units (TUs) from the residual blocks. The transform module 52 selects a transform from among a plurality of transforms. The transform may be selected based on one or more coding characteristics (e.g., block size, coding mode, etc.). Transform module 52 then applies the selected transform to the TU, producing a video block comprising a two-dimensional array of transform coefficients.

Transform module 52 may send the resulting transform coefficients to quantization unit 54. Quantization unit 54 may then quantize the transform coefficients. Entropy encoding unit 56 may then perform a scan of the quantized transform coefficients in the matrix according to a scan pattern. This disclosure describes entropy encoding unit 56 as performing a scan. However, it should be understood that in other examples, other processing units, such as quantization unit 54, may perform the scanning.

Once the transform coefficients are scanned into a one-dimensional array, entropy encoding unit 56 may apply entropy coding, such as Context Adaptive Variable Length Coding (CAVLC), Context Adaptive Binary Arithmetic Coding (CABAC), syntax-based context adaptive binary arithmetic coding (SBAC), or another entropy coding method, to the coefficients. Entropy coding may also be applied to syntax elements, such as syntax elements used in merge mode.

To perform CAVLC, entropy encoding unit 56 may select a variable length code for a symbol to be transmitted. The code words in VLC may be constructed such that relatively shorter codes correspond to more likely symbols, while longer codes correspond to less likely symbols. In this way, bit savings may be achieved using VLC as compared to, for example, using a constant length codeword for each symbol to be transmitted.

To perform CABAC, entropy encoding unit 56 may select a context model to apply to a certain context to encode a symbol to be transmitted. In the case of transform coefficients, context may relate to, for example, whether neighboring values are non-zero. Entropy encoding unit 56 may also entropy encode syntax elements, such as signals representing the selected transform. In accordance with the techniques of this disclosure, entropy encoding unit 56 may select a context model to encode these syntax elements based on, for example: the direction of intra prediction for an intra prediction mode, the scan position of the coefficients corresponding to the syntax elements, the block type, and/or the transform type, as well as other factors for context model selection.

After entropy coding by entropy encoding unit 56, the resulting encoded video may be transmitted to another device, such as video decoder 30 or the like, or archived for later transmission or retrieval.

In some cases, entropy encoding unit 56 or another unit of video encoder 20 may be configured to perform other coding functions in addition to entropy coding. For example, entropy encoding unit 56 may be configured to determine Coded Block Pattern (CBP) values for the CU and the PU. Also, in some cases, entropy encoding unit 56 may perform run length coding of the coefficients.

Inverse quantization unit 58 and inverse transform module 60 apply inverse quantization and inverse transform, respectively, to reconstruct the residual block in the pixel domain, e.g., for later use in reconstructing the reference block. Motion compensation unit 44 may calculate a reference block by adding the residual block to a predictive block formed by one of a plurality of frames of reference frame buffer 64 (which may also be referred to as a decoded picture buffer). Motion compensation unit 44 may also apply one or more interpolation filters to the reconstructed reference block to calculate sub-integer pixel values for use in motion estimation. Summer 62 adds the reconstructed residual block to the motion compensated prediction block generated by motion compensation unit 44 to generate a reconstructed video block for storage in reference frame buffer 64. The reconstructed video block may be used by motion estimation unit 42 and motion compensation unit 44 as a reference block to inter-code a block in another subsequently coded video frame.

Fig. 7 is a block diagram illustrating an example of video decoder 30 that decodes an encoded video sequence. In the example of fig. 7, video decoder 30 includes an entropy decoding unit 70, a motion compensation unit 72, an intra prediction unit 74, an inverse quantization unit 76, an inverse transform module 78, a reference frame buffer 82, and a summer 80. In some examples, video decoder 30 may perform a decoding pass generally reciprocal to the encoding pass described with respect to video encoder 20 (see fig. 6).

Entropy decoding unit 70 performs an entropy decoding process on the encoded bitstream to retrieve a one-dimensional array of transform coefficients. The entropy decoding process used depends on the entropy coding (e.g., CABAC, CAVLC, etc.) used by video encoder 20. The entropy coding process used by the encoder may be signaled in the encoded bitstream or the entropy coding process may be a predetermined process.

In some examples, entropy decoding unit 70 (or inverse quantization unit 76) may scan the received values using a scan that mirrors the scan pattern used by entropy encoding unit 56 (or quantization unit 54) of video encoder 20. Although scanning of the coefficients may alternatively be performed in inverse quantization unit 76, for purposes of illustration, scanning is described as being performed by entropy decoding unit 70. In addition, although shown as separate functional units for ease of illustration, the structure and functionality of entropy decoding unit 70, inverse quantization unit 76, and other units of video decoder 30 may be highly integrated with one another.

Inverse quantization unit 76 inverse quantizes (i.e., dequantizes) the quantized transform coefficients provided in the bitstream and decoded by entropy decoding unit 70. The inverse quantization process may include conventional processes, e.g., similar to the processes proposed for HEVC or defined by the h.264 decoding standard. The inverse quantization process may include using a quantization parameter QP calculated by video encoder 20 for the CU to determine the degree of quantization and, likewise, the degree of inverse quantization that should be applied. Inverse quantization unit 76 may inverse quantize the transform coefficients before or after converting the coefficients from a one-dimensional array to a two-dimensional array.

The inverse transform module 78 applies an inverse transform to the inverse quantized transform coefficients. In some examples, inverse transform module 78 may determine the inverse transform based on signaling from video encoder 20 or by inferring the transform from one or more coding characteristics (e.g., block size, coding mode, etc.). In some examples, inverse transform module 78 may determine the transform applied to the current block based on the signaled transform at the root node of the quadtree for the LCU that includes the current block. Alternatively, the transform may be signaled at the root of a TU quadtree of a leaf node CU in the LCU quadtree. In some examples, inverse transform module 78 may apply a cascaded inverse transform, where inverse transform module 78 applies two or more inverse transforms to the transform coefficients of the current block being decoded.

Intra-prediction unit 74 may generate prediction data for a current block of a current frame based on the signaled intra-prediction mode and data from previously decoded blocks of the current frame.

According to an example of this disclosure, video decoder 30 may receive, from an encoded bitstream, prediction syntax indicating indices of selected merge candidate blocks from the set of merge candidate blocks for use in a motion vector prediction process. The video decoder is further configured to retrieve a motion vector, a reference frame, and a prediction direction associated with the merge candidate block identified by the received index, and perform inter-prediction decoding for the current block using the retrieved motion vector, reference frame, and prediction direction.

According to the example of this disclosure described above, for merge mode, a merge candidate set may be generated by video decoder 30 without comparing the motion information of any merge candidate to the motion information of other PUs within the same CU as the current PU. In addition, the invention also proposes: in case the merge candidate is located inside another PU of the same CU, the merge candidate is removed from the merge candidate set. The generation of merge candidates may be handled by motion compensation unit 72 or by any other fixed function or programmable hardware structure of video decoder 30. Once video decoder 30 has determined the final merge candidate set, it may retrieve motion information from the merge candidates indicated by the received index.

As one example, for an nx2N partition of a CU, all merge candidates (e.g., the merge candidates shown in fig. 1) are available to PU 0. For PU1, when merge candidate L is within PU0, merge candidate L is removed/excluded from the merge candidate list (see fig. 3A). As another example, for a 2 nxn partition of a CU, all merge candidates (e.g., the merge candidate shown in fig. 1) are available to PU 0. For PU1, when merge candidate a is within PU0, merge candidate a is removed/excluded from the merge candidate list (see fig. 3B).

As another example, for nxn partitioning of a CU, all merge candidates (e.g., the merge candidate shown in fig. 1) are available to PU 0. For PU1, when merge candidates L and BL are within PU0 and PU2, respectively, merge candidates L and BL are removed/excluded from the merge candidate list (see fig. 4A). For PU2, when merge candidate a and RA are within PU0 and PU1, respectively, merge candidate a and RA are removed/excluded from the merge candidate list (see fig. 4A). For PU3, when merge candidates LA, a, and L are within PU0, PU1, and PU2, respectively, merge candidates LA, a, and L are removed/excluded from the merge candidate list (see fig. 4A).

As yet another example, for an nxn partition mode, all merge candidates for each prediction unit are used regardless of any merge candidates used by another prediction unit (see fig. 4B). For other partitioning modes (e.g., 2 nx N, N × 2N, etc.) as well as asymmetric modes (e.g., 2 nxnd, 2 nxnu, nL × 2N, nR × 2N, etc.), the exclusion of certain candidates for the current PU that is located inside the PU of the same CU is applied according to the process described above.

Returning to fig. 7, motion compensation unit 72 may generate motion compensated blocks, possibly performing interpolation filter-based interpolation. An identifier for an interpolation filter to be used for motion estimation of sub-pixel precision may be included in the syntax element. Motion compensation unit 72 may calculate interpolated values for sub-integer pixels of the reference block using interpolation filters as used by video encoder 20 during encoding of the video block. Motion compensation unit 72 may determine the interpolation filters used by video encoder 20 from the received syntax information and use the interpolation filters to generate predictive blocks.

Additionally, in the HEVC example, motion compensation unit 72 and intra-prediction unit 74 may use some of the syntax information (e.g., provided through a quadtree) to determine the size of the LCU used to encode the frame(s) of the encoded video sequence. Motion compensation unit 72 and intra-prediction unit 74 may also use syntax information to determine splitting information that describes how each CU (and, likewise, how sub-CUs) of a frame of an encoded video sequence are split. The syntax information may also include modes indicating how each CU is encoded (e.g., intra-prediction or inter-prediction, and for intra-prediction, intra-prediction encoding modes), one or more reference frames for each inter-encoded PU (and/or a reference list containing identifiers for the reference frames), and other information to decode the encoded video sequence.

Summer 80 combines the residual block with the corresponding prediction block produced by motion compensation unit 72 or intra-prediction unit 74 to form a decoded block. In effect, the decoded block reconstructs the originally coded block, subject to losses due to quantization or other coding aspects. When desired, deblocking filters may also be applied to filter the decoded blocks in order to remove blocking artifacts. The decoded video blocks are then stored in a reference frame buffer 82, the reference frame buffer 82 providing reference blocks for subsequent motion compensation and also generating decoded video for presentation on a display device (such as display device 32 of fig. 5, etc.).

Fig. 8 is a flow diagram illustrating an example method of encoding video in accordance with the techniques of this disclosure. The method of fig. 8 may be performed by video encoder 20 of fig. 6. Video encoder 20 may be configured to determine a merge candidate set for a current prediction unit of a current coding unit, wherein the merge candidate set is determined without comparing motion information of a merge candidate in the merge candidate set to motion information of another prediction unit 200, and remove a merge candidate within another prediction unit of the current coding unit from the merge candidate set 202. The merge candidate set may include a left above merge candidate, an above merge candidate, a right above merge candidate, a left merge candidate, a below left merge candidate, and a temporal merge candidate.

In the case that the current coding unit has a 2 nxn partition type (including prediction unit 0 positioned above prediction unit 1), video encoder 20 may remove the above merge candidate from the merge candidate set for prediction unit 1. In the case that the current coding unit has an nx 2N partition type (including prediction unit 0 positioned to the left of prediction unit 1), video encoder 20 may remove the left merge candidate from the merge candidate set for prediction unit 1.

In the case that the current coding unit has an nxn partition type (including prediction unit 0 located in an upper-left portion of the current coding unit, prediction unit 1 located in an upper-right portion of the current coding unit, prediction unit 2 located in a lower-left portion of the current coding unit, and prediction unit 3 located in a lower-right portion of the current coding unit), video encoder 20 may remove the left-side merge candidate and the lower-left merge candidate from the merge candidate set for prediction unit 1. In this case, video encoder 20 may further remove the above merge candidate and the upper-right merge candidate from the merge candidate set for prediction unit 2. In this case, video encoder 20 may further remove the above merge candidate, the left merge candidate, and the left above merge candidate from the merge candidate set for prediction unit 3.

In other examples, excluding the merge candidate comprises excluding, for all partition modes other than the nxn partition mode, a merge candidate that is within another prediction unit of the current coding unit from the merge candidate set. In this case, no merge candidates are removed/excluded from the prediction unit from the coding unit having the N × N partition mode.

Video encoder 20 may be further configured to perform a merge motion vector prediction process for the current prediction unit using the merge candidate set to determine a selected merge candidate 204 for the current prediction unit, and signal a syntax element 206 in the encoded video bitstream indicating the selected merge candidate.

Fig. 9 is a flow diagram illustrating an example method of decoding video in accordance with the techniques of this disclosure. The method of fig. 9 may be performed by video decoder 30 of fig. 7. Video decoder 30 may be configured to receive syntax element 220 indicating a selected merge candidate for a current prediction unit, and determine a merge candidate set for the current prediction unit of the current coding unit, wherein the merge candidate set is determined without comparing motion information of the merge candidates in the merge candidate set to motion information 222 of any other prediction units. Video decoder 30 may be further configured to remove 224 a merge candidate that is within another prediction unit of the current coding unit from the merge candidate set. The merge candidate set may include a left above merge candidate, an above merge candidate, a right above merge candidate, a left merge candidate, a below left merge candidate, and a temporal merge candidate.

In the case that the current coding unit has a 2 nxn partition type (including prediction unit 0 positioned above prediction unit 1), video decoder 30 may remove the above merge candidate from the merge candidate set for prediction unit 1. In the case that the current coding unit has an nx 2N partition type (including prediction unit 0 positioned to the left of prediction unit 1), video decoder 30 may remove the left merge candidate from the merge candidate set for prediction unit 1.

In the case that the current coding unit has an nxn partition type (including prediction unit 0 located in an upper-left portion of the current coding unit, prediction unit 1 located in an upper-right portion of the current coding unit, prediction unit 2 located in a lower-left portion of the current coding unit, and prediction unit 3 located in a lower-right portion of the current coding unit), video decoder 30 may remove the left-side merge candidate and the lower-left merge candidate from the merge candidate set for prediction unit 1. In this case, video decoder 30 may further remove the above merge candidate and the right above merge candidate from the merge candidate set for prediction unit 2. In this case, video decoder 30 may further remove the above merge candidate, the left merge candidate, and the left above merge candidate from the merge candidate set for prediction unit 3.

In other examples, excluding the merge candidate comprises excluding, for all partition modes other than the nxn partition mode, a merge candidate that is within another prediction unit of the current coding unit from the merge candidate set. In this case, no merge candidates are removed/excluded from the prediction unit from the coding unit having the N × N partition mode.

Video decoder 30 may be further configured to perform a merge motion vector prediction process for the current prediction unit using the merge candidate set and the received syntax elements to determine motion vectors 226 for the current prediction unit, and decode the current prediction unit 228 using the determined motion vectors.

In one or more examples, the functions described may be implemented in hardware, software, firmware, or any combination thereof. If implemented in software, the functions may be stored on or transmitted over as one or more instructions or code on a computer-readable medium and executed by a hardware-based processing unit. Computer-readable media may include computer-readable storage media (which corresponds to tangible media such as data storage media) or communication media, including any medium that facilitates transfer of a computer program from one place to another, such as in accordance with a communication protocol. In this manner, computer-readable media may generally correspond to (1) tangible computer-readable storage media that is not transitory, or (2) communication media such as signals or carrier waves. A data storage medium may be any available medium that can be accessed by one or more computers or one or more processors to retrieve instructions, code and/or data structures for implementation of the techniques described in this disclosure. The computer program product may include a computer-readable medium.

By way of example, and not limitation, such computer-readable storage media can comprise RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, flash memory, or any other medium that can be used to store desired program code in the form of instructions or data structures and that can be accessed by a computer. Also, any connection is properly termed a computer-readable medium. For example, if the instructions are transmitted from a website, server, or other remote source using a coaxial cable, fiber optic cable, twisted pair, Digital Subscriber Line (DSL), or wireless technologies such as infrared, radio, and microwave, etc., then the coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwave, etc. are included in the definition of medium. It should be understood, however, that computer-readable storage media and data storage media do not include connections, carrier waves, signals, or other transitory media, but rather pertain to non-transitory tangible storage media. Disk and disc, as used herein, includes Compact Disc (CD), laser disc, optical disc, Digital Versatile Disc (DVD), floppy disk and blu-ray disc where disks usually reproduce data magnetically, while discs reproduce data optically with lasers. Combinations of the above should also be included within the scope of computer-readable media.

The instructions may be executed by one or more processors, such as one or more Digital Signal Processors (DSPs), general purpose microprocessors, Application Specific Integrated Circuits (ASICs), field programmable logic arrays (FPGAs), or other equivalent integrated or discrete logic circuitry. Thus, the term "processor," as used herein may refer to any of the foregoing structure or any other structure suitable for implementation of the techniques described herein. In addition, in some aspects, the functionality described herein may be provided within dedicated hardware and/or software modules configured for encoding and decoding, or incorporated in a combined codec. Also, the techniques may be fully implemented in one or more circuits or logic elements.

The techniques of this disclosure may be implemented in a wide variety of devices or apparatuses, including wireless handsets, Integrated Circuits (ICs), or sets of ICs (e.g., chipsets). Various components, modules, or units are described in this disclosure to emphasize functional aspects of devices configured to perform the disclosed techniques, but do not necessarily require realization by different hardware units. More particularly, as described above, the various units may be combined in a codec hardware unit, or provided by a set of interoperability hardware units (including one or more processors as described above) in conjunction with suitable software and/or firmware.

Various examples have been described. These and other examples are within the scope of the following claims.

Claims (40)

1. A method of coding video data, comprising:

determining a merge candidate set for a current prediction unit of a current coding unit, wherein the merge candidate set is determined without comparing motion information of merge candidates in the merge candidate set to motion information of any other prediction units; and

performing a merge motion vector prediction process for the current prediction unit using the merge candidate set.

2. The method of claim 1, further comprising:

excluding merge candidates that are within another prediction unit of the current coding unit from the merge candidate set.

3. The method of claim 2, wherein excluding merge candidates comprises excluding merge candidates from the merge candidate set that are within another prediction unit of the current coding unit for all partition modes other than an nxn partition mode.

4. The method of claim 3, further comprising:

merge candidates are not excluded from prediction units from the coding unit having the NxN partition mode.

5. The method of claim 2, wherein the merge candidate set includes a left above merge candidate, an above merge candidate, a right above merge candidate, a left merge candidate, a below left merge candidate, and a temporal merge candidate.

6. The method of claim 5, wherein the current coding unit has a 2 NxN, 2 NxnU, or 2 NxnD partition type, including prediction unit 0, positioned above prediction unit 1, and wherein excluding a merge candidate from the merge candidate set comprises excluding the above merge candidate from the merge candidate set for prediction unit 1.

7. The method of claim 5, wherein the current coding unit has an nx 2N, nL x2N or an nR x2N partition type, including a prediction unit 0 positioned to the left of prediction unit 1, and wherein excluding a merge candidate from the merge candidate set comprises excluding the left merge candidate from the merge candidate set for prediction unit 1.

8. The method of claim 2, wherein the current coding unit has an NxN partition type, including prediction unit 0 positioned in an upper left portion of the current coding unit, prediction unit 1 positioned in an upper right portion of the current coding unit, prediction unit 2 positioned in a lower left portion of the current coding unit, and prediction unit 3 positioned in a lower right portion of the current coding unit,

wherein excluding a merge candidate from the merge candidate set comprises excluding the left merge candidate and the bottom-left merge candidate from the merge candidate set for prediction unit 1,

wherein excluding a merge candidate from the merge candidate set comprises excluding the above merge candidate and the above-right merge candidate from the merge candidate set for prediction unit 2, and

wherein excluding a merge candidate from the merge candidate set comprises excluding the above merge candidate, the left merge candidate, and the left above merge candidate from the merge candidate set for prediction unit 3.

9. The method of claim 2, wherein coding video comprises decoding video, and wherein the method further comprises:

receiving a syntax element indicating a selected merge candidate for the current prediction unit;

performing the merge motion vector prediction process for the current prediction unit using the merge candidate set and the received syntax elements to determine a motion vector for the current prediction unit; and

decoding the current prediction unit using the determined motion vector.

10. The method of claim 2, wherein coding video comprises encoding video, and wherein the method further comprises:

performing the motion vector prediction process for the current prediction unit using the merge candidate set to determine a selected merge candidate for the current prediction unit; and

signaling a syntax element indicating the selected merge candidate in an encoded video bitstream.

11. An apparatus configured to code video data, comprising:

a video coder configured to:

determining a merge candidate set for a current prediction unit of a current coding unit, wherein the merge candidate set is determined without comparing motion information of merge candidates in the merge candidate set to motion information of any other prediction units; and

performing a merge motion vector prediction process for the current prediction unit using the merge candidate set.

12. The apparatus of claim 11, wherein the video coder is further configured to:

excluding merge candidates that are within another prediction unit of the current coding unit from the merge candidate set.

13. The apparatus of claim 12, wherein the video coder is further configured to:

excluding merge candidates from the merge candidate set that are within another prediction unit of the current coding unit for all partition modes other than the NxN partition mode.

14. The apparatus of claim 13, wherein the video coder is further configured to:

merge candidates are not excluded from prediction units from the coding unit having the NxN partition mode.

15. The apparatus of claim 12, wherein the merge candidate set includes a left above merge candidate, an above merge candidate, a right above merge candidate, a left merge candidate, a below left merge candidate, and a temporal merge candidate.

16. The apparatus of claim 15, wherein the current coding unit has a 2 nxn, 2 nxnu, or 2 nxnd partition type, including a prediction unit 0 positioned above a prediction unit 1, and wherein the video coder is further configured to exclude the above merge candidate from the merge candidate set for prediction unit 1.

17. The apparatus of claim 15, wherein the current coding unit has an nx 2N, nL x2N or an nrx2N partition type, including a prediction unit 0 positioned to the left of prediction unit 1, and wherein the video coder is further configured to exclude the left merge candidate from the merge candidate set for prediction unit 1.

18. The apparatus of claim 15, wherein the current coding unit has an NxN partition type including prediction unit 0 positioned in an upper left portion of the current coding unit, prediction unit 1 positioned in an upper right portion of the current coding unit, prediction unit 2 positioned in a lower left portion of the current coding unit, and prediction unit 3 positioned in a lower right portion of the current coding unit,

wherein the video coder is further configured to exclude the left merge candidate and the bottom-left merge candidate from the merge candidate set for prediction unit 1,

wherein the video coder is further configured to exclude the above merge candidate and the above-right merge candidate from the merge candidate set for prediction unit 2, and

wherein the video coder is further configured to exclude the above merge candidate, the left merge candidate, and the left above merge candidate from the merge candidate set for prediction unit 3.

19. The apparatus of claim 12, wherein the video coder is a video decoder, and wherein the video decoder is further configured to:

receiving a syntax element indicating a selected merge candidate for the current prediction unit;

performing the merge motion vector prediction process for the current prediction unit using the merge candidate set and the received syntax elements to determine a motion vector for the current prediction unit; and

decoding the current prediction unit using the determined motion vector.

20. The apparatus of claim 12, wherein the video coder is a video encoder, and wherein the video encoder is further configured to:

performing the motion vector prediction process for the current prediction unit using the merge candidate set to determine a selected merge candidate for the current prediction unit; and

signaling a syntax element indicating the selected merge candidate in an encoded video bitstream.

21. The apparatus of claim 11, wherein the video coder is part of a mobile device.

22. An apparatus configured to code video data, comprising:

means for determining a merge candidate set for a current prediction unit of a current coding unit, wherein the merge candidate set is determined without comparing motion information of a merge candidate in the merge candidate set to motion information of any other prediction units; and

means for performing a merge motion vector prediction process for the current prediction unit using the merge candidate set.

23. The apparatus of claim 22, further comprising:

means for excluding a merge candidate within another prediction unit of the current coding unit from the merge candidate set.

24. The apparatus of claim 23, wherein the means for excluding merge candidates comprises means for excluding merge candidates from the merge candidate set that are within another prediction unit of the current coding unit for all partition modes other than an nxn partition mode.

25. The apparatus of claim 23, further comprising:

means for excluding merge candidates from prediction units from the coding unit having the NxN partition mode.

26. The apparatus of claim 23, wherein the merge candidate set includes a left above merge candidate, an above merge candidate, a right above merge candidate, a left merge candidate, a below left merge candidate, and a temporal merge candidate.

27. The apparatus of claim 26, wherein the current coding unit has a 2 nxn, 2 nxnu, or 2 nxnd partition type, including a prediction unit 0 positioned above a prediction unit 1, and wherein the means for excluding a merge candidate from the merge candidate set comprises means for excluding the above merge candidate from the merge candidate set for prediction unit 1.

28. The apparatus of claim 26, wherein the current coding unit has an nx 2N, nL x2N or an nrx2N partition type, including a prediction unit 0 positioned to the left of a prediction unit 1, and wherein the means for excluding a merge candidate from the merge candidate set comprises means for excluding the left merge candidate from the merge candidate set for prediction unit 1.

29. The apparatus of claim 26, wherein the current coding unit has an NxN partition type including prediction unit 0 positioned in an upper left portion of the current coding unit, prediction unit 1 positioned in an upper right portion of the current coding unit, prediction unit 2 positioned in a lower left portion of the current coding unit, and prediction unit 3 positioned in a lower right portion of the current coding unit,

wherein the means for excluding a merge candidate from the merge candidate set comprises means for excluding the left merge candidate and the bottom-left merge candidate from the merge candidate set for prediction unit 1,

wherein the means for excluding a merge candidate from the merge candidate set comprises means for excluding the above merge candidate and the above-right merge candidate from the merge candidate set for prediction unit 2, and

wherein the means for excluding a merge candidate from the merge candidate set comprises means for excluding the above merge candidate, the left merge candidate, and the left above merge candidate from the merge candidate set for prediction unit 3.

30. The apparatus of claim 23, wherein the apparatus is configured to decode video, and wherein the apparatus further comprises:

means for receiving a syntax element indicating a selected merge candidate for the current prediction unit;

means for performing the merge motion vector prediction process for the current prediction unit using the merge candidate set and the received syntax elements to determine a motion vector for the current prediction unit; and

means for decoding the current prediction unit using the determined motion vector.

31. The apparatus of claim 23, wherein the apparatus is configured to encode video, and wherein the apparatus further comprises:

means for performing the motion vector prediction process for the current prediction unit using the merge candidate set to determine a selected merge candidate for the current prediction unit; and

means for signaling a syntax element indicating the selected merge candidate in an encoded video bitstream.

32. The apparatus of claim 22, wherein the apparatus is part of a mobile device.

33. A computer-readable storage medium having stored thereon instructions that, when executed, cause one or more processors configured to code video data to:

determining a merge candidate set for a current prediction unit of a current coding unit, wherein the merge candidate set is determined without comparing motion information of merge candidates in the merge candidate set to motion information of any other prediction units; and

performing a merge motion vector prediction process for the current prediction unit using the merge candidate set.

34. The computer-readable storage medium of claim 33, wherein the instructions further cause the one or more processors to:

excluding merge candidates that are within another prediction unit of the current coding unit from the merge candidate set.

35. The computer-readable storage medium of claim 34, wherein the instructions further cause the one or more processors to:

excluding merge candidates from the merge candidate set that are within another prediction unit of the current coding unit for all partition modes other than the NxN partition mode.

36. The computer-readable storage medium of claim 35, wherein the instructions further cause the one or more processors to:

merge candidates are not excluded from prediction units from the coding unit having the NxN partition mode.

37. The computer-readable storage medium of claim 34, wherein the merge candidate set includes a left above merge candidate, an above merge candidate, a right above merge candidate, a left merge candidate, a below left merge candidate, and a temporal merge candidate.

38. The computer-readable storage medium of claim 37, wherein the current coding unit has a 2 nxn, 2 nxnu, or 2 nxnd partition type, including prediction unit 0 positioned above prediction unit 1, and wherein the instructions further cause the one or more processors to exclude the above merge candidate from the merge candidate set for prediction unit 1.

39. The computer-readable storage medium of claim 37, wherein the current coding unit has an nx 2N, nL x2N or nRx2N partition type, including a prediction unit 0 positioned to the left of a prediction unit 1, and wherein the instructions further cause the one or more processors to exclude the left merge candidate from the merge candidate set for prediction unit 1.

40. The computer-readable storage medium of claim 37, wherein the current coding unit has an NxN partition type including prediction unit 0 positioned in an upper left portion of the current coding unit, prediction unit 1 positioned in an upper right portion of the current coding unit, prediction unit 2 positioned in a lower left portion of the current coding unit, and prediction unit 3 positioned in a lower right portion of the current coding unit,

wherein the instructions further cause the one or more processors to exclude the left-side merge candidate and the bottom-left merge candidate from the merge candidate set for prediction unit 1, wherein the instructions further cause the one or more processors to exclude the above merge candidate and the top-right merge candidate from the merge candidate set for prediction unit 2, and

wherein the instructions further cause the one or more processors to exclude the above merge candidate, the left merge candidate, and the left above merge candidate from the merge candidate set for prediction unit 3.