Selection and Signaling of Motion Vector (MV) Precisions

Abstract

A decoder including a receiver and a processor is disclosed. The receiver is configured to receive an encoded bitstream containing a motion vector (MV) candidate index for a current block. The processor is coupled to the receiver, and configured to obtain precisions of candidate motion vectors (MVs) corresponding to neighboring blocks of the current block, perform first rounding of the precisions based on a rounding scheme, perform second rounding of the candidate MVs based on the first rounding, perform pruning of the candidate MVs, generate a candidate list based on the second rounding and the pruning, and select one of the candidate MVs from the candidate list for decoding the current block based on the MV candidate index.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS

The present application is a continuation of International Application No. PCT/CN2018/090181 filed Jun. 7, 2018, by Huawei Technologies Co., Ltd., and titled “Selection and Signaling of Motion Vector (MV) Precisions,” which claims priority to U.S. non-provisional patent application Ser. No. 15/982,865 filed on May 17, 2018 by Futurewei Technologies, Inc. and titled “Motion Vector Prediction and Merge Candidate Selection,” which claims priority to U.S. provisional patent application No. 62/518,402 filed on Jun. 12, 2017 by Futurewei Technologies, Inc. and titled “Motion Vector Prediction and Merge Candidate Selection,” each of which is incorporated herein by reference.

STATEMENT REGARDING FEDERALLY SPONSORED RESEARCH OR DEVELOPMENT

Not applicable.

REFERENCE TO A MICROFICHE APPENDIX

Not applicable.

BACKGROUND

Videos use a relatively large amount of data, so communication of videos uses a relatively large amount of bandwidth. However, many networks operate at or near their bandwidth capacities. In addition, customers demand high video qualities, which require using even more data. There is therefore a desire to both reduce the amount of data videos use and improve the video qualities. One solution is to compress videos during an encoding process and decompress the videos during a decoding process. Improving compression and decompression techniques is a focus of research and development.

SUMMARY

According to one aspect of the present disclosure, there is provided a receiver configured to receive an encoded bitstream containing a motion vector (MV) candidate index for a current block; and a processor coupled to the receiver, the processor configured to: obtain candidate motion vectors (MVs) corresponding to neighboring blocks of the current block, the neighboring blocks neighbor the current block; obtain precisions of candidate MVs corresponding to neighboring blocks of the current block; perform first rounding of the precisions based on a rounding scheme; perform second rounding of the candidate MVs based on the first rounding; perform pruning of the candidate MVs; generate a candidate list based on the second rounding and the pruning; and select one of the candidate MVs from the candidate list for decoding the current block based on the MV candidate index.

Optionally, in any of the preceding aspects, another implementation of the aspect provides that the neighboring blocks comprise a first neighboring block and a second neighboring block, and wherein the precisions comprise a first precision of the first neighboring block and a second precision of the second neighboring block.

Optionally, in any of the preceding aspects, another implementation of the aspect provides that the rounding scheme comprises: rounding the first precision to a target precision of the current block; and rounding the second precision to the target precision.

Optionally, in any of the preceding aspects, another implementation of the aspect provides that the rounding scheme comprises: determining a lowest precision from among any combination of the first precision, the second precision, and a target precision of the current block; and rounding at least one of the first precision, the second precision, or the target precision to the lowest precision.

Optionally, in any of the preceding aspects, another implementation of the aspect provides that the rounding scheme comprises: determining a median precision from among any combination of the first precision, the second precision, and a target precision of the current block; and rounding at least one of the first precision, the second precision, or the target precision to the median precision.

Optionally, in any of the preceding aspects, another implementation of the aspect provides that the rounding scheme comprises: determining a highest precision from among any combination of the first precision, the second precision, and a target precision of the current block; and rounding at least one of the first precision, the second precision, or the target precision to the highest precision.

Optionally, in any of the preceding aspects, another implementation of the aspect provides that the rounding scheme comprises: determining a default precision; and rounding at least one of the first precision, the second precision, or a target precision of the current block to the default precision.

Optionally, in any of the preceding aspects, another implementation of the aspect provides that the pruning comprises discarding identical candidate MVs or sub-identical candidate MVs.

Optionally, in any of the preceding aspects, another implementation of the aspect provides that the pruning further comprises adding zero MVs to fill the candidate list.

Optionally, in any of the preceding aspects, another implementation of the aspect provides that the processor is further configured to further perform the first rounding and the second rounding before the pruning.

Optionally, in any of the preceding aspects, another implementation of the aspect provides that the processor is further configured to further perform the first rounding and the second rounding after the pruning.

Optionally, in any of the preceding aspects, another implementation of the aspect provides that the candidate list is an AMVP candidate list.

Optionally, in any of the preceding aspects, another implementation of the aspect provides that the candidate list is a merge candidate list.

Optionally, in any of the preceding aspects, another implementation of the aspect provides that the processor is configured to perform inter-prediction using the one of the candidate MVs selected to generate a reconstructed block.

Optionally, in any of the preceding aspects, another implementation of the aspect provides that the reconstructed block is used to generate an image for display on a display screen of an electronic device.

According to another aspect of the present disclosure, there is provided a method comprising: receiving, by a receiver of the video decoder, an encoded bitstream containing a motion vector (MV) candidate index for a current block; obtaining, by a processor of the video decoder, candidate motion vectors (MVs) corresponding to neighboring blocks of the current block, the neighboring blocks neighbor the current block; obtaining, by the processor, precisions of candidate MVs corresponding to neighboring blocks of the current block; performing, by the processor, first rounding of the precisions based on a rounding scheme; performing, by the processor, second rounding of the candidate MVs based on the first rounding; performing, by the processor, pruning of the candidate MVs; generating, by the processor, a candidate list based on the second rounding and the pruning; and selecting, by the processor, one of the candidate MVs from the candidate list for decoding the current block based on the MV candidate index.

Optionally, in any of the preceding aspects, another implementation of the aspect provides that the rounding scheme comprises rounding all of the precisions to a target precision, and wherein the second rounding comprises rounding each of the candidate MVs to a multiple of the target precision.

Optionally, in any of the preceding aspects, another implementation of the aspect provides that the target precision is obtained from a slice header of the encoded bitstream.

According to another aspect of the present disclosure, there is provided a non-transitory computer readable medium comprising a computer program product for use by a video coding device, the computer program product comprising computer executable instructions stored on the non-transitory computer readable medium such that when executed by a processor cause the video coding device to: receive an encoded bitstream containing a motion vector (MV) candidate index for a current block; obtain candidate motion vectors (MVs) corresponding to neighboring blocks of the current block, the neighboring blocks neighbor the current block; obtain precisions of candidate MVs corresponding to neighboring blocks of the current block; perform first rounding of the precisions based on a rounding scheme; perform second rounding of the candidate MVs based on the first rounding; perform pruning of the candidate MVs; generate a candidate list based on the second rounding and the pruning; and select one of the candidate MVs from the candidate list for decoding the current block based on the MV candidate index.

Optionally, in any of the preceding aspects, another implementation of the aspect provides that the computer executable instructions further cause the video coding device to: perform inter-prediction using the one of the candidate MVs selected to generate a reconstructed block; and generate an image on a display of an electronic device using the reconstructed block.

Any of the above embodiments may be combined with any of the other above embodiments to create a new embodiment. These and other features will be more clearly understood from the following detailed description taken in conjunction with the accompanying drawings and claims.

BRIEF DESCRIPTION OF THE DRAWINGS

For a more complete understanding of this disclosure, reference is now made to the following brief description, taken in connection with the accompanying drawings and detailed description, wherein like reference numerals represent like parts.

FIG. 1 is a schematic diagram of a coding system.

FIG. 2 is a diagram of a portion of a video frame.

FIG. 3 is a flowchart illustrating a method of candidate list generation according to an embodiment of the disclosure.

FIG. 4 is a flowchart illustrating a method of decoding according to an embodiment of the disclosure.

FIG. 5 is a schematic diagram of an apparatus according to an embodiment of the disclosure.

DETAILED DESCRIPTION

It should be understood at the outset that, although an illustrative implementation of one or more embodiments are provided below, the disclosed systems and/or methods may be implemented using any number of techniques, whether currently known or in existence. The disclosure should in no way be limited to the illustrative implementations, drawings, and techniques illustrated below, including the exemplary designs and implementations illustrated and described herein, but may be modified within the scope of the appended claims along with their full scope of equivalents.

The following abbreviations and initialisms apply:

AMVP: advanced motion vector prediction

ASIC: application-specific integrated circuit

CPU: central processing unit

CTU: coding tree unit

DSP: digital signal processor

EO: electrical-to-optical

FPGA: field-programmable gate array

ITU: International Telecommunication Union

ITU-T: ITU Telecommunication Standardization Sector

LCD: liquid crystal display

MV: motion vector

OE: optical-to-electrical

PPS: picture parameter set

RAM: random-access memory

RF: radio frequency

ROM: read-only memory

RX: receiver unit

SPS: sequence parameter set

SRAM: static RAM

TCAM: ternary content-addressable memory

TX: transmitter unit.

FIG. 1 is a schematic diagram of a coding system 100. The coding system 100 comprises a source device 110, a medium 150, and a destination device 160. The source device 110 and the destination device 160 are mobile phones, tablet computers, desktop computers, notebook computers, or other suitable devices. The medium 150 is a local network, a radio network, the Internet, or another suitable medium.

The source device 110 comprises a video generator 120, an encoder 130, and an output interface 140. The video generator 120 is a camera or another device suitable for generating video. The encoder 130 may be referred to as a codec. The encoder 130 performs encoding according to a set of rules, for instance as described in “High Efficiency Video Coding,” ITU-T H.265, December 2016 (“H.265”), which is incorporated by reference. The output interface 140 is an antenna or another component suitable for transmitting data to the destination device 160. Alternatively, the video generator 120, the encoder 130, and the output interface 140 are in any suitable combination of devices.

The destination device 160 comprises an input interface 170, a decoder 180, and a display 190. The input interface 170 is an antenna or another component suitable for receiving data from the source device 110. The decoder 180 may also be referred to as a codec. The decoder 180 performs decoding according to a set of rules, for instance as described in H.265. The display 190 is an LCD screen or another component suitable for displaying videos. Alternatively, the input interface 170, the decoder 180, and the display 190 are in any suitable combination of devices.

In operation, in the source device 110, the video generator 120 captures a video, the encoder 130 encodes the video to create an encoded video, and the output interface 140 transmits the encoded video over the medium 150 and towards the destination device 160. The source device 110 may locally store the video or the encoded video, or the source device 110 may instruct storage of the video or the encoded video on another device. The encoded video comprises data defined at various levels, including slices and blocks. A slice is a spatially distinct region of a video frame that the encoder 130 encodes separately from any other region in the video frame. A block is a group of pixels arranged in a rectangle, for instance an 8 pixel×8 pixel square. Blocks may also be referred to as units or coding units. Other levels include regions, CTUs, and tiles. In the destination device 160, the input interface 170 receives the encoded video from the source device 110, the decoder 180 decodes the encoded video to obtain a decoded video, and the display 190 displays the decoded video. The decoder 180 may decode the encoded video in a reverse manner compared to how the encoder 130 encodes the video. The destination device 160 locally stores the encoded video or the decoded video, or the destination device 160 instructs storage of the encoded video or the decoded video on another device.

Together, encoding and decoding are referred to as coding. Coding comprises intra coding and inter coding, which may be referred to as intra prediction and inter prediction, respectively. Intra prediction implements spatial prediction to reduce spatial redundancy within a video frame. Inter prediction implements temporal prediction to reduce temporal redundancy between successive video frames.

FIG. 2 is a diagram of a portion 200 of a video frame. The portion 200 comprises a current block 210 and neighboring blocks A₀ 220, A₁ 230, B₂ 240, B₁ 250, and B₀ 260. The current block 210 is called a current block because the encoder 130 is currently encoding it. The neighboring blocks 220-260 are called neighboring blocks because they neighbor the current block 210. The neighboring block A₀ 220 is in a bottom-left corner position, the neighboring block A₁ 230 is in a left-bottom side position, the neighboring block B₂ 240 is in a top-left corner position, the neighboring block B₁ 250 is in a top-right side position, and the neighboring block B₀ 260 is in a top-right corner position. In inter prediction, the encoder 130 determines an MV of the current block 210 based on MVs of the neighboring blocks 220-260. The MVs of the neighboring blocks 220-260 are already known when the encoder 130 is encoding the current block 210.

AMVP and merge are two coding modes the encoder 130 may use to determine the MV of the current block 210. In AMVP, the encoder 130 determines an MV candidate list by sequentially scanning the neighboring blocks 220-260 as follows: A₀ 220, A₁ 230, a scaled version of A₀ 220, a scaled version of A₁ 230, B₀ 260, B₁ 250, B₂ 240, a scaled version of B₀ 260, a scaled version of B₁ 250, a scaled version of B₂ 240, a temporal bottom right collocated candidate, a temporal central collocated candidate, and a zero MV. From that candidate list, the encoder 130 selects the best candidate and indicates the best candidate in a signal to the destination device 160. In merge, the encoder 130 determines a candidate list of MVs by sequentially scanning the neighboring blocks as follows: A₁ 230, B₁ 250, B₀ 260, A₀ 220, B₂ 240, temporal bottom right collocated candidate, temporal central collocated candidate, and a zero MV. From that candidate list, the encoder 130 selects the best candidate and indicates the best candidate in a signal to the destination device 160. Though AMVP and merge scan the neighboring blocks 220-260, other coding modes may scan different neighboring blocks that are not shown.

Both AMVP and merge define precisions for MVs. Precisions indicate a distance between pixels that serves as a multiplier to define an MV. In the context of precisions, pixels may be referred to as pels. Thus, precisions include quarter-pel precision, half-pel precision, integer-pel precision, and four-pel precision. Quarter-pel precision defines an MV in terms of a quarter of a distance between pixels so that the MV may be 0.25, 0.5, 0.75, 1.0, 1.25, and so on. The encoder 130 encodes precisions as flags in slice headers. A slice header is a header at the beginning of a slice signal that indicates data applicable to the entire slice. Alternatively, the encoder 130 encodes precisions as flags or other data in block headers.

Both AMVP and merge define rules for determining qualified candidates for their candidate lists. For instance, if two candidate MVs are the same, then the encoder 130 removes one of those candidate MVs. The encoder 130 continues applying those rules until it fills the candidate lists with a maximum number of allowed candidates. Those rules are well defined when the candidate MVs have the same precision. However, it is desirable to provide coding-efficient rules for candidate MVs with different precisions.

Disclosed herein are embodiments for selection and signaling of MV precisions. The embodiments provide for obtaining MVs with different precisions, rounding those precisions in a uniform manner, rounding MVs, and generating candidate lists. The embodiments further provide for signaling precisions at the slice, region, CTU, or tile level. The embodiments apply to AMVP, merge, and other suitable coding modes.

AMVP Candidate List Generation

In a first embodiment of AMVP candidate list generation, the encoder 130 considers three precisions: a first precision of a first neighboring block 220-260, a second precision of a second neighboring block 220-260, and a third precision of the current block 210. For instance, the encoder 130 considers precisions of the neighboring block 220, the neighboring block 230, and the current block 210. The third precision may be referred to as a target precision because it is associated with the current block 210, which is a target of encoding. Using the three precisions, the encoder 130 executes the following pseudo code:

i = 0
if(availableFlagLXA) {
mvpListLX[ i++] = mvLXA
if(availableFlagLXB && (
RoundMvPrecisionToTarget(mvLXA)
RoundMvPrecisionToTarget(mvLXB)
mvLXA! = mvLXB)))
mvpListLX[ i++ ] = mvLXB
} else if(availableFlagLXB)
mvpListLX[i++] = mvLXB
if( i < 2 && availableFlagLXCol)
mvpListLX[ i++] = mvLXCol
while( i < 2 ) {
mvpListLX[i][0] = 0
mvpListLX[i][1] = 0
i++
}

The pseudo code implements the following rules: if the target precision uses a highest precision, then the encoder 130 rounds up the first precision and the second precision to the target precision. If the target precision uses a lowest precision, then the encoder 130 rounds down the first precision and the second precision to the target precision. Otherwise, the encoder 130 rounds up either the first precision or the second precision, whichever is lower, to the target precision, and the encoder 130 rounds down either the first precision or the second precision, whichever is higher, to the target precision. Lower means numerically lower, and higher means numerically higher. Thus, a quarter-pel precision of 0.25 is lower than an integer-pel precision of 1.0.

After rounding the precisions, the encoder 130 either rounds up or rounds down the candidate MVs based on those precisions. For instance, if a candidate MV is 1.25 and has a native precision of 0.25, and if the target precision is 0.5, then the encoder 130 rounds up the candidate MV to 1.5. A native precision is a precision of an MV before the encoder 130 rounds up or rounds down that precision per the pseudo code above. Finally, for each candidate MV, the encoder 130 determines whether an identical candidate MV or a sub-identical candidate MV is already in a candidate list. If not, then the encoder 130 adds the candidate MV to the candidate list. If so, then the encoder 130 discards the candidate MV. Sub-identical means identical within a pre-defined threshold.

In a first alternative, RoundMvPrecisionToLow( ) replaces RoundMvPrecisionToTarget( ) in the pseudo code so that the encoder 130 determines a lowest precision from among the first precision, the second precision, and the target precision, then rounds down either the first precision, the second precision, or both the first precision and the second precision to that precision. In a second alternative, RoundMvPrecisionToLow( ) replaces RoundMvPrecisionToTarget( ) in the pseudo code so that the encoder 130 determines a lowest precision from among the first precision and the second precision, then rounds down either the first precision or the second precision, whichever remains, to that precision. In a third alternative, RoundMvPrecisionToHigh( ) replaces RoundMvPrecisionToTarget( ) in the pseudo code so that the encoder 130 determines a highest precision from among the first precision, the second precision, and the target precision, then rounds up the first precision, the second precision, or both the first precision and the second precision to that precision. In a fourth alternative, RoundMvPrecisionToHigh( ) replaces RoundMvPrecisionToTarget( ) in the pseudo code so that the encoder 130 determines a highest precision from among the first precision and the second precision, then rounds up either the first precision or the second precision, whichever remains, to that precision.

In a fifth alternative, RoundMvPrecisionToDefault( ) replaces RoundMvPrecisionToTarget( ) in the pseudo code so that the encoder 130 rounds up or rounds down the first precision and the second precision to a default precision. The encoder 130 may first receive the default precision in a signal from another device. In a sixth alternative, RoundMvPrecisionToMedian( ) replaces RoundMvPrecisionToTarget( ) in the pseudo code so that the encoder 130 determines a median precision from among the first precision, the second precision, and the target precision, then rounds up or rounds down the first precision, the second precision, or both the first precision and the second precision to that precision. If the encoder 130 considers an even number of precisions, then the median precision may be an average of the two median precisions.

In a second embodiment of AMVP candidate list generation, the encoder 130 considers the target precision of the current block 210. Using the target precision, the encoder 130 executes the following pseudo code:

i = 0
if(availableFlagLXA) {
mvpListLX[i++] = mvLXA
if(availableFlagLXB && (mvLXA != mvLXB))
mvpListLX[i++] = mvLXB
} else if(availableFlagLXB)
mvpListLX[i++] = mvLXB
if(i < 2 && availableFlagLXCol)
mvpListLX[i++] = mvLXCol
if(i == 2)
if(diffMotion(mvLXM, mvLXN))
i−−
while(i < 2) {
mvpListLX[i][0] = 0
mvpListLX i][1] = 0
i++
}

In the pseudo code, mvLXM and mvLXN in diffMotion( ) may be mvLXA, mvLXB, or mvLXCol.

The pseudo code implements the following rules for diffMotion( ): for each candidate MV, the encoder 130 determines whether an identical candidate MV or a sub-identical candidate MV is already in a candidate list. If not, then the encoder adds the candidate MV to the candidate list. If so, then the encoder 130 discards the candidate MV. Thus, the encoder 130 first considers the candidate MVs in their native precisions.

After removing identical candidate MVs or sub-identical candidate MVs, if a target precision uses a highest precision, then the encoder 130 rounds up candidate precisions to the target precision. The candidate precisions are precisions of the neighboring blocks 220-260. If a target precision uses a lowest precision, then the encoder 130 rounds down the candidate precisions to the target precision. Otherwise, the encoder 130 rounds up lower candidate precisions to the target precision, and the encoder 130 rounds down higher candidate precisions to the target precision.

After rounding the precisions, the encoder 130 either rounds up or rounds down the candidate MVs based on those precisions. For each candidate MV, the encoder 130 determines whether an identical candidate MV or a sub-identical candidate MV is already in a candidate list. If not, then the encoder 130 keeps the candidate MV in the candidate list. If so, then the encoder 130 discards the candidate MV from the candidate list. If diffMotion( ) returns 1, then two candidate MVs are not identical or sub-identical. If diffMotion( ) returns 0, then two candidate MVs are identical or sub-identical. Finally, if the candidate list is not full, then the encoder 130 inserts candidate MVs of a 0 value to fill the missing spots. For instance, if the candidate list is to have 10 MVs, but only 8 MVs are left in the candidate list, then the encoder inserts 2 MVs with a 0 value.

In a first alternative, diffMotion( ) operates so that the encoder 130 determines a lowest precision from among a first precision of a first neighboring block 220-260, a second precision of a second neighboring block 220-260, and the target precision, then rounds down all remaining candidate precisions to that precision. In a second alternative, diffMotion( ) operates so that the encoder 130 determines a highest precision from among the first precision, the second precision, and the target precision, then rounds up all remaining candidate precisions to that precision. In a third alternative, diffMotion( ) operates so that the encoder 130 rounds up or rounds down the candidate precisions to a default precision. The encoder 130 may first receive the default precision in a signal from another device. In a fourth alternative, diffMotion( ) operates so that the encoder 130 determines a median precision from among the first precision, the second precision, and the target precision, then rounds up or rounds down the first precision, the second precision, or both the first precision and the second precision to that precision. If the encoder 130 considers an even number of precisions, then the median precision may be an average of the two median precisions.

In the above embodiments, after the encoder 130 generates an AMVP candidate list, it may select a final candidate MV. The encoder 130 may do so based on various criteria that are known. Once the encoder 130 does so, the encoder 130 may signal the final candidate MV in the encoded video. Instead of comparing a first precision and a second precision, the encoder 130 may compare all candidate precisions.

Merge Candidate List Generation

In a first embodiment of merge candidate list generation, the encoder 130 considers three precisions: a first precision of a first neighboring block 220-260, a second precision of a second neighboring block 220-260, and a target precision of the current block 210. The first embodiment of merge candidate list generation is similar to the first embodiment of AMVP candidate list generation. Using the three precisions, the encoder 130 executes the following pseudo code:

i = 0
if(availableFlagA1)
mergeCandList[i++] = RoundMvPrecisionToTarget(A1)
if(availableFlagB1 && diffMotions(RoundMvPrecisionToTarget(B1)))
mergeCandList[i++] = B1
if(availableFlagB0 && diffMotions(RoundMvPrecisionToTarget(B0)))
mergeCandList[i++] = B0
if(availableFlagA0 && diffMotions(RoundMvPrecisionToTarget(A0)))
mergeCandList[i++] = A0
if(availableFlagB2 && diffMotions(RoundMvPrecisionToTarget(B2)))
mergeCandList[i++] = B2
if(availableFlagCol)
mergeCandList[i++] =
diffMotions(RoundMvPrecisionToTarget(Col)))

The pseudo code implements the following rules for diffMotions( ): If the target precision uses a highest precision, then the encoder 130 rounds up the first precision and the second precision to the target precision. If the target precision uses a lowest precision, then the encoder 130 rounds down the first precision and the second precision to the target precision. Otherwise, the encoder 130 rounds up either the first precision or the second precision, whichever is lower, to the target precision, and the encoder 130 rounds down either the first precision or the second precision, whichever is higher, to the target precision.

After rounding the precisions, the encoder 130 either rounds up or rounds down the candidate MVs based on those precisions. Finally, for each candidate MV, the encoder 130 determines whether an identical candidate MV or a sub-identical candidate MV is already in a candidate list. If not, then the encoder 130 adds the candidate MV to the candidate list. If so, then the encoder 130 discards the candidate MV. If diffMotion( ) returns 1, then two candidate MVs are not identical or sub-identical. If diffMotion( ) returns 0, then two candidate MVs are identical or sub-identical.

In a first alternative, RoundMvPrecisionToLow( ) replaces RoundMvPrecisionToTarget( ) in the pseudo code so that the encoder 130 determines a lowest precision from among the first precision, the second precision, and the target precision, then rounds down either the first precision, the second precision, or both the first precision and the second precision to that precision. In a second alternative, RoundMvPrecisionToLow( ) replaces RoundMvPrecisionToTarget( ) in the pseudo code so that the encoder 130 determines a lowest precision from among the first precision and the second precision, then rounds down either the first precision or the second precision, whichever remains, to that precision. In a third alternative, RoundMvPrecisionToHigh( ) replaces RoundMvPrecisionToTarget( ) in the pseudo code so that the encoder 130 determines a highest precision from among the first precision, the second precision, and the target precision, then rounds up the first precision, the second precision, or both the first precision and the second precision to that precision. In a fourth alternative, RoundMvPrecisionToHigh( ) replaces RoundMvPrecisionToTarget( ) in the pseudo code so that the encoder 130 determines a highest precision from among the first precision and the second precision, then rounds up either the first precision or the second precision, whichever remains, to that precision.

In a fifth alternative, RoundMvPrecisionToDefault( ) replaces RoundMvPrecisionToTarget( ) in the pseudo code so that the encoder 130 rounds up or rounds down the first precision and the second precision to a default precision. The encoder 130 may first receive the default precision in a signal from another device. In a sixth alternative, RoundMvPrecisionToMedian( ) replaces RoundMvPrecisionToTarget( ) in the pseudo code so that the encoder 130 determines a median precision from among the first precision, the second precision, and the target precision, then rounds up or rounds down the first precision, the second precision, or both the first precision and the second precision to that precision. If the encoder 130 considers an even number of precisions, then the median precision may be an average of the two median precisions.

In a second embodiment of merge candidate list generation, the encoder 130 considers the target precision of the current block 210. The second embodiment of merge candidate list generation is similar to the second embodiment of AMVP candidate list generation. Using the target precision, the encoder 130 executes the following pseudo code:

i = 0
if(availableFlagA1)
mergeCandList[i++] = A1
if(availableFlagB1)
mergeCandList[i++] = B1
if(availableFlagB0)
mergeCandList[i++] = B0
if(availableFlagA0)
mergeCandList[i++] = A0
if(availableFlagB2)
mergeCandList[i++] = B2
PruneList( );
if(availableFlagCol && i < max_num_candidates)
mergeCandList[i++] = Col

The pseudo code implements the following rules for PruneList( ): If a target precision uses a highest precision, then the encoder 130 rounds up candidate precisions to the target precision. The candidate precisions are precisions of the neighboring blocks 220-260. If a target precision uses a lowest precision, then the encoder 130 rounds down the candidate precisions to the target precision. Otherwise, the encoder 130 rounds up lower candidate precisions to the target precision, and the encoder 130 rounds down higher candidate precisions to the target precision.

After rounding the precisions, the encoder 130 either rounds up or rounds down the candidate MVs based on those precisions. For each candidate MV, the encoder 130 determines whether an identical candidate MV or a sub-identical candidate MV is already in a candidate list. If not, then the encoder 130 adds the candidate MV to the candidate list. If so, then the encoder 130 discards the candidate MV and decrements i by 1. Finally, if the candidate list is not full, then the encoder 130 inserts candidate MVs of a 0 value to fill the missing spots. For instance, if the candidate list is to have 10 MVs, but only 8 MVs are left in the candidate list, then the encoder 130 inserts 2 MVs with a 0 value.

In a first alternative, PruneList( ) operates so that the encoder 130 determines a lowest precision from among a first precision of a first neighboring block 220-260, a second precision of a second neighboring block 220-260, and the target precision, then rounds down all remaining candidate precisions to that precision. In a second alternative, PruneList( ) operates so that the encoder 130 determines a highest precision from among the first precision, the second precision, and the target precision, then rounds up all remaining candidate precisions to that precision. In a third alternative, PruneList( ) operates so that the encoder 130 rounds up or rounds down the candidate precisions to a default precision. The encoder 130 may first receive the default precision in a signal from another device. In a fourth alternative, PruneList( ) operates so that the encoder 130 determines a median precision from among the first precision, the second precision, and the target precision, then rounds up or rounds down the first precision, the second precision, or both the first precision and the second precision to that precision. If the encoder 130 considers an even number of precisions, then the median precision may be an average of the two median precisions.

In a third embodiment of merge candidate list generation, the encoder 130 considers the target precision of the current block 210. The third embodiment of merge candidate list generation is similar to the second embodiment of merge candidate list generation. However, unlike the second embodiment of merge candidate list generation, which executes PruneList ( ) before checking the candidate list for identical candidate MVs or sub-identical candidate MVs, the third embodiment of merge candidate list generation executes PruneList ( ) after checking a candidate list. Using the target precision, the encoder 130 executes the following pseudo code:

i = 0
if(availableFlagA1)
mergeCandList[i++] = A1
if(availableFlagB1)
mergeCandList[i++] = B1
if(availableFlagB0)
mergeCandList[i++] = B0
if(availableFlagA0)
mergeCandList[i++] = A0
if(availableFlagB2)
mergeCandList[i++] = B2
if(availableFlagCol)
mergeCandList[i++] = Col
PruneList( )

In the above embodiments, after the encoder 130 generates a merge candidate list, it may select a final candidate MV. The encoder 130 may do so based on various criteria that are known. Once the encoder 130 does so, the encoder 130 may signal the final candidate MV in the encoded video.

Precision Signaling

In a first embodiment of precision signaling, the encoder 130 encodes a precision for MVs in a slice header as follows:

slice_segment_header( ) {
coding_mode_idx
mv_precision[ coding_mode_idx ]
}

As shown, the slice header indicates a precision for all MVs for a single coding mode, which may be either AMVP or merge.

In a second embodiment of precision signaling, the encoder 130 encodes a precision for MVs in a slice header as follows:

slice_segment_header( ) {
num_coding_mode
for(i = 0; i < num_coding_mode; i++) {
coding_mode_idx
mv_precision_used_by_coding_mode[coding_mode_idx]
}
}

As shown, the slice header indicates a precision for all MVs for multiple coding modes, which may include AMVP and merge.

In a third embodiment of precision signaling, the encoder 130 encodes a precision for MVs in a slice header as follows:

slice_segment_header( ) {
if( motion_vector_resolution_control_idc = = 3 ) {
num_mv_precision
for( i = 0; i < num_mv_precision; i++ ) {
mv_precision_idx[i]
}
}
}

As shown, the slice header indicates a precision for all MVs for multiple coding modes, which may include AMVP and merge. The third embodiment of signaling is similar to the second embodiment of signaling, but uses a different syntax.

In a fourth embodiment of precision signaling, the encoder 130 encodes a precision for MVs in a coding unit header as follows:

coding_unit( x0, y0, log2CbSize ) {
if( motion_vector_resolution_control_idc = = 3 ) {
mv_precision_idx_used_by_coding_mode
}
}

As shown, the coding unit header indicates a precision for all MVs for a single coding mode, which may be either AMVP or merge.

In a fifth embodiment of precision signaling, the encoder 130 encodes a precision for MVs in a coding unit header as follows:

coding_unit( x0, y0, log2CbSize ) {
if(motion_vector_resolution_control_idc = = 3) {
num_coding_mode
for(i = 0; i < num_coding_mode; i++) {
coding_mode_idx
mv_precision_idx_used_by_coding_mode
[coding_mode_idx]
}
}
}

As shown, the coding unit header indicates a precision for all MVs for multiple coding modes, which may include AMVP and merge.

In a sixth embodiment of precision signaling, the encoder 130 encodes all allowed precisions in an SPS in a form such as {mvr1, mvr2, . . . , mvrN}. For each slice header, the encoder 130 also encodes a slice header index indicating one of those precisions that is to apply to all blocks in the corresponding slice. For instance, the SPS is {quarter-pel, integer-pel, four-pel} and a slice header index is either 00 indicating quarter-pel, 01 indicating integer-pel, or 11 indicating four-pel.

In a seventh embodiment of precision signaling, the encoder 130 encodes all allowed precisions in an SPS in a form such as {mvr1, mvr2, . . . , mvrN}. For each slice header, the encoder 130 also encodes multiple slice header indices indicating which of those precisions are to apply to blocks in the corresponding slice. For instance, the SPS is {quarter-pel, integer-pel, four-pel} and the slice header indices are any combination of 00 indicating quarter-pel, 01 indicating integer-pel, and 11 indicating four-pel.

In an eighth embodiment of precision signaling, the encoder 130 encodes all allowed precisions in an SPS in a form such as {mvr1, mvr2, . . . , mvrN}. For each slice header, the encoder 130 also encodes a slice header index indicating a subset of those precisions that are to apply to blocks in the corresponding slice. For instance, the SPS is {quarter-pel, integer-pel, four-pel} and the slice header index is 00 indicating {quarter-pel, integer-pel}, 01 indicating {quarter-pel, four-pel}, or 10 indicating {integer-pel, four-pel}.

In a ninth embodiment of precision signaling, the encoder 130 encodes all allowed precisions in an SPS in a form such as {mvr1, mvr2, . . . , mvrN}. For each slice header, the encoder 130 also encodes a slice header index indicating one of those precisions, in addition to its immediate finer or coarser precision, that are to apply to blocks in the corresponding slice. For instance, the SPS is {quarter-pel, integer-pel, four-pel}, and the slice header index is 00 indicating {quarter-pel}, 01 indicating {quarter-pel, integer-pel}, or 10 indicating {integer-pel, four-pel} when the slice header indicates a finer precision or the slice header index is 00 indicating {quarter-pel, integer-pel}, 01 indicating {integer-pel, four-pel}, or 10 indicating {four-pel} when the slice header indicates a coarser precision.

In a tenth embodiment of precision signaling, the encoder 130 encodes all allowed precisions in an SPS in a form such as {mvr1, mvr2, . . . , mvrN}. For each slice header, the encoder 130 also encodes a slice header index indicating which of those precisions are to apply to blocks in the corresponding slice. For each block header, the encoder 130 also encodes a block header index indicating one of those precisions that is to apply to the corresponding block. For instance, the SPS is {quarter-pel, integer-pel, four-pel}, a first slice header index of 0 indicates {quarter-pel, integer-pel}, a first block header index of 0 in the first slice header indicates quarter-pel, a second block header index of 1 in the first slice header indicates integer-pel, a second slice header index of 1 indicates {integer-pel, four-pel}, a third block header index of 0 in the second slice header indicates integer-pel, and a fourth block header index of 1 in the second slice header indicates four-pel.

In the above embodiments, instead of encoding header indices at the slice level, the encoder 130 may encode the header indices at the region level, CTU level, or tile level. Though the indices are expressed as specific binary numbers, any binary numbers or other numbers may be used. Instead of indicating all precisions in an SPS, the encoder 130 may indicate all the precisions in a PPS. Instead of comparing a first precision and a second precision, the encoder 130 may compare all candidate precisions.

FIG. 3 is a flowchart illustrating a method 300 of candidate list generation according to an embodiment of the disclosure. The encoder 130 performs the method 300. At step 310, a current block of a video is obtained. For instance, the encoder 130 obtains the current block 210. At step 320, candidate MVs corresponding to neighboring blocks of the video are obtained. The neighboring blocks neighbor the current block. For instance, the encoder 130 obtains candidate MVs corresponding to the neighboring blocks 220-260. At step 330, precisions of the candidate MVs are obtained.

At step 340, first rounding of the precisions is performed based on a rounding scheme. For instance, the rounding scheme is one of the rounding schemes described above. Thus, the rounding scheme may comprise rounding to a target precision; rounding to a lowest precision, a median precision, or a highest precision from among precisions of the neighboring blocks 220-260; rounding to a lowest precision, a median precision, or a highest precision from among precisions of the neighboring blocks 220-260 and the target block; or rounding to a default precision. At step 350, second rounding of the candidate MVs is performed based on the first rounding. For instance, the encoder 130 determines a precision from step 340 and either rounds up or rounds down the candidate MVs based on that precision.

At step 360, pruning of the candidate MVs is performed. For instance, the pruning is completed as described above. Thus, the pruning may comprise discarding identical candidate MVs, discarding sub-identical candidate MVs, or adding zero MVs to fill a candidate list. Finally, at step 370, a candidate list is generated based on the second rounding and the pruning. The candidate list may be an AMVP candidate list or a merge candidate list as described above.

FIG. 4 is a flowchart illustrating a method 400 of decoding according to an embodiment of the disclosure. The decoder 180 performs the method 400. At step 410, an encoded bitstream containing an MV candidate index for a current block is received. The MV candidate index identifies which of the MV candidates was used for encoding the current block, as described above.

At step 420, candidate MVs corresponding to neighboring blocks of the video are obtained. The neighboring blocks neighbor the current block. For instance, the decoder 180 obtains candidate MVs corresponding to the neighboring blocks 220-260. At step 430, precisions of candidate MVs are obtained.

At step 440, first rounding of the precisions is performed based on a rounding scheme. For instance, the rounding scheme is one of the rounding schemes described above. Thus, the rounding scheme may comprise rounding to a target precision; rounding to a lowest precision, a median precision, or a highest precision from among precisions of the neighboring blocks 220-260; rounding to a lowest precision, a median precision, or a highest precision from among precisions of the neighboring blocks 220-260 and the target block; or rounding to a default precision. At step 450, second rounding of the candidate MVs is performed based on the first rounding. For instance, the decoder 180 determines a precision from step 440 and either rounds up or rounds down the candidate MVs based on that precision.

At step 460, pruning of the candidate MVs is performed. For instance, the pruning is completed as described above. Thus, the pruning may comprise discarding identical candidate MVs, discarding sub-identical candidate MVs, or adding zero MVs to fill a candidate list. At step 470, a candidate list is generated based on the second rounding and the pruning. The candidate list may be an AMVP candidate list or a merge candidate list as described above.

In step 480, one of the candidate MVs is selected from the candidate list for decoding the current block based on the MV candidate index. In an embodiment, the decoder 180 performs inter-prediction using the candidate MV that was selected to generate a reconstructed block. The decoder 180 can then generate an image on a display of an electronic device using the reconstructed block.

FIG. 5 is a schematic diagram of an apparatus 500 according to an embodiment of the disclosure. The apparatus 500 may implement the disclosed embodiments. The apparatus 500 comprises ingress ports 510 and an RX 520 for receiving data; a processor, logic unit, baseband unit, or CPU 530 to process the data; a TX 540 and egress ports 550 for transmitting the data; and a memory 560 for storing the data. The apparatus 500 may also comprise OE components, EO components, or RF components coupled to the ingress ports 510, the RX 520, the TX 540, and the egress ports 550 for ingress or egress of optical, electrical signals, or RF signals.

The processor 530 is any combination of hardware, middleware, firmware, or software. The processor 530 comprises any combination of one or more CPU chips, cores, FPGAs, ASICs, or DSPs. The processor 530 communicates with the ingress ports 510, the RX 520, the TX 540, the egress ports 550, and the memory 560. The processor 530 comprises a coding component 570, which implements the disclosed embodiments. The inclusion of the coding component 570 therefore provides a substantial improvement to the functionality of the apparatus 500 and effects a transformation of the apparatus 500 to a different state. Alternatively, the memory 560 stores the coding component 570 as instructions, and the processor 530 executes those instructions.

The memory 560 comprises any combination of disks, tape drives, or solid-state drives. The apparatus 500 may use the memory 560 as an over-flow data storage device to store programs when the apparatus 500 selects those programs for execution and to store instructions and data that the apparatus 500 reads during execution of those programs. The memory 560 may be volatile or non-volatile and may be any combination of ROM, RAM, TCAM, or SRAM.

In an example embodiment, an apparatus comprises: a memory element; and a processor element coupled to the memory element and configured to: obtain a current block of a video frame, obtain candidate MVs corresponding to neighboring blocks of the video frame, the neighboring blocks neighbor the current block, obtain precisions of the candidate MVs, perform first rounding of the precisions based on a rounding scheme, perform second rounding of the candidate MVs based on the first rounding, perform pruning of the candidate MVs, and generate a candidate list based on the second rounding and the pruning.

While several embodiments have been provided in the present disclosure, it may be understood that the disclosed systems and methods might be embodied in many other specific forms without departing from the spirit or scope of the present disclosure. The present examples are to be considered as illustrative and not restrictive, and the intention is not to be limited to the details given herein. For example, the various elements or components may be combined or integrated in another system or certain features may be omitted, or not implemented.

In addition, techniques, systems, subsystems, and methods described and illustrated in the various embodiments as discrete or separate may be combined or integrated with other systems, components, techniques, or methods without departing from the scope of the present disclosure. Other items shown or discussed as coupled may be directly coupled or may be indirectly coupled or communicating through some interface, device, or intermediate component whether electrically, mechanically, or otherwise. Other examples of changes, substitutions, and alterations are ascertainable by one skilled in the art and may be made without departing from the spirit and scope disclosed herein.

Claims

1. A decoder comprising:

a receiver configured to receive an encoded bitstream containing a motion vector (MV) candidate index for a current block; and

a processor coupled to the receiver, the processor configured to: obtain candidate motion vectors (MVs) corresponding to neighboring blocks of the current block, the neighboring blocks neighbor the current block; obtain precisions of the candidate MVs; perform first rounding of the precisions based on a rounding scheme; perform second rounding of the candidate MVs based on the first rounding; perform pruning of the candidate MVs; generate a candidate list based on the second rounding and the pruning; and select one of the candidate MVs from the candidate list for decoding the current block based on the MV candidate index.

2. The decoder of claim 1, wherein the neighboring blocks comprise a first neighboring block and a second neighboring block, and wherein the precisions comprise a first precision of the first neighboring block and a second precision of the second neighboring block.

3. The decoder of claim 2, wherein the rounding scheme comprises:

rounding the first precision to a target precision of the current block; and

rounding the second precision to the target precision.

4. The decoder of claim 2, wherein the rounding scheme comprises:

determining a lowest precision from among any combination of the first precision, the second precision, and a target precision of the current block; and

rounding at least one of the first precision, the second precision, or the target precision to the lowest precision.

5. The decoder of claim 2, wherein the rounding scheme comprises:

determining a median precision from among any combination of the first precision, the second precision, and a target precision of the current block; and

rounding at least one of the first precision, the second precision, or the target precision to the median precision.

6. The decoder of claim 2, wherein the rounding scheme comprises:

determining a highest precision from among any combination of the first precision, the second precision, and a target precision of the current block; and

rounding at least one of the first precision, the second precision, or the target precision to the highest precision.

7. The decoder of claim 2, wherein the rounding scheme comprises:

determining a default precision; and

rounding at least one of the first precision, the second precision, or a target precision of the current block to the default precision.

8. The decoder of claim 1, wherein the pruning comprises discarding identical candidate MVs or sub-identical candidate MVs.

9. The decoder of claim 8, wherein the pruning further comprises adding zero MVs to fill the candidate list.

10. The decoder of claim 1, wherein the processor is further configured to further perform the first rounding and the second rounding before the pruning.

11. The decoder of claim 1, wherein the processor is further configured to further perform the first rounding and the second rounding after the pruning.

12. The decoder of claim 1, wherein the candidate list is an advanced motion vector prediction (AMVP) candidate list.

13. The decoder of claim 1, wherein the candidate list is a merge candidate list.

14. The decoder of claim 1, wherein the processor is configured to perform inter-prediction using the one of the candidate MVs selected to generate a reconstructed block.

15. The decoder of claim 14, wherein the reconstructed block is used to generate an image for display on a display screen of an electronic device.

16. A method implemented in a video decoder, comprising:

receiving, by a receiver of the video decoder, an encoded bitstream containing a motion vector (MV) candidate index for a current block;

obtaining, by a processor of the video decoder, candidate motion vectors (MVs) corresponding to neighboring blocks of the current block, the neighboring blocks neighbor the current block;

obtaining, by the processor, precisions of candidate MVs;

performing, by the processor, first rounding of the precisions based on a rounding scheme;

performing, by the processor, second rounding of the candidate MVs based on the first rounding;

performing, by the processor, pruning of the candidate MVs;

generating, by the processor, a candidate list based on the second rounding and the pruning; and

selecting, by the processor, one of the candidate MVs from the candidate list for decoding the current block based on the MV candidate index.

17. The method of claim 16, wherein the rounding scheme comprises rounding all of the precisions to a target precision, and wherein the second rounding comprises rounding each of the candidate MVs to a multiple of the target precision.

18. The method of claim 17, wherein the target precision is obtained from a slice header of the encoded bitstream.

19. A non-transitory computer readable medium comprising a computer program product for use by a video coding device, the computer program product comprising computer executable instructions stored on the non-transitory computer readable medium such that when executed by a processor cause the video coding device to:

receive an encoded bitstream containing a motion vector (MV) candidate index for a current block;

obtain candidate motion vectors (MVs) corresponding to neighboring blocks of the current block, the neighboring blocks neighbor the current block;

obtain precisions of candidate MVs corresponding to neighboring blocks of the current block;

perform first rounding of the precisions based on a rounding scheme;

perform second rounding of the candidate MVs based on the first rounding;

perform pruning of the candidate MVs;

generate a candidate list based on the second rounding and the pruning; and

select one of the candidate MVs from the candidate list for decoding the current block based on the MV candidate index.

20. The non-transitory computer readable medium of claim 19, wherein the computer executable instructions further cause the video coding device to:

perform inter-prediction using the one of the candidate MVs selected to generate a reconstructed block; and

generate an image on a display of an electronic device using the reconstructed block.