WO2013155665A1 - Improvement of implicit tu split mode in hevc - Google Patents

Abstract

To improve the implicit TU split mode in high efficiency video coding (HEVC) so that the intended principle that TU cannot cross PU's boundary can be fulfilled all the time, the related syntax and semantics is revised. This improvement can be seen as a bug fix which makes the implicit TU split mode more reasonable.

Description

IMPROVEMENT OF IMPLICIT TU SPLIT MODE IN HEVC

TECHNICAL FIELD [0001] The invention relates generally to video coding. In particular, the present invention relates to methods and apparatuses for transform unit (TU) split method in high efficiency video coding (HEVC).

BACKGROUND [0002] HEVC is an emerging video coding standard being developed by ITU-T and MPEG. In current HEVC (text specification draft 6), when the syntax element max transform hierarchy depth inter is equal to 0, the implicit TU split mode is used.

[0003] When the implicit TU split mode is used, the Coding Unit (CU) with non-square inter prediction unit (PU) partition is forced to split once without the signaling of TU split flag. The principle of this implicit TU split mode is to force TU not to cross PU's boundary.

[0004] However, since the asymmetric PU partition (AMP) may be used and the largest TU size may be not equal to CU size, so there are still cases that violate the principle, i.e., TU crosses PU's boundary. Fig. 1 shows an example, the CU size is 64x64, largest one-dimension TU size is 32, and 2NxnU AMP is used. It can be seen from the right figure that the above two TUs (TUo and TUi) still cross PU's boundary. SUMMARY

[0005] In this invention, the implicit TU split mode is improved to address the issue mentioned above. Two methods are proposed. The first method proposes to further split TU in the problem cases to ensure TU not to cross PU's boundary. The second one proposes to directly remove the problem cases.

BRIEF DESCRIPTION OF DRAWINGS

[0006] The invention relates generally to video coding. In particular, the presented invention can be more fully understood by reading the subsequent detailed description and examples with references made to the accompanying drawings, wherein:

[0007] Fig. 1 shows the implicit TU split mode in current HEVC.

[0008] Fig. 2 shows an embodiment of the proposed implicit TU split mode.

DETAILED DESCRIPTION [0009] The following description is of the best-contemplated mode of carrying out the invention. This description is made for the purpose of illustrating the general principles of the invention and should not be taken in a limiting sense. The scope of the invention is best determined by reference to the appended claims.

[0010] As previously mentioned, in current HEVC, there are some cases in implicit TU split mode that TU crosses PU's boundary. To address this issue, two methods are proposed as follows:

[0011] The first method proposes to further split TU in the problem cases to ensure TU not to cross PU's boundary. For example, Fig. 2 shows the solution for the case shown in Fig. 1. Generally, related part of transform tree syntax in the HEVC text specification has been modified as shown in Table 1. The modified parts are highlighted.

[0012] Table 1. Transform tree syntax modification transform_tree( xO, yO, xC, yC, log2CbSize, log2Trafo Width, log2TrafoHeight, Descrip trafoDepth, blkldx ) { tor if( trafoDepth = = 0 && IntraSplitFlag = = 0 && PredMode != MODE INTPvA &&

!(PartMode = = PART 2Nx2N && merge flag|x0^"||y01) )

no_residual_data_flag ae(v) if( !no residual data flag ) {

log2TrafoSize = ( log2Trafo Width + log2TrafoHeight ) » 1

intraSplitFlag = ( IntraSplitFlag && trafoDepth = = 0 ? 1 : 0 )

interSplitFlag = ( max transform hierarchy depth inter = =0 &&

PredMode = = MODE INTER && PartMode != PART_2Nx2N &&

(trafoDepth = = 0 || (trafoDepth= =1 && log2CbSize == log2MaxTrafoSize+l && (blkldx<=l && PartMode= =PART 2NxnU || blkldx>=2 &&

PartMode= =PART 2NxnD || blkldx%2==0 && PartMode==PART nLx2N || blkldx%2==l

&& PartMode==PART_nRx2N)))

maxDepth = ( PredMode = = MODE INTRA ?

max transform hierarchy depth intra + IntraSplitFlag :

max transform hierarchy depth inter + InterSplitFlag +

(InterSplitFlag && trafoDepth==l))

xBase = xO - ( xO & ( 1 « log2Trafo Width ) )

yBase = yO - ( yO & ( 1 « log2TrafoHeight ) )

if( log2TrafoSize <= Log2MaxTrafoSize &&

log2TrafoSize > Log2MinTrafoSize &&

trafoDepth < maxDepth && !intraSplitFlag && !interSplitFlag )

split_transform_flag[ xO ][ yO ][ trafoDepth ] ae(v) if( log2TrafoSize <= Log2MaxTrafoSize ) {

firstChromaCbf = ( log2TrafoSize = = Log2MaxTrafoSize

trafoDepth = = 0 ) ? 1 : 0

if( firstChromaCbf | | log2TrafoSize > Log2MinTrafoSize ) {

if( firstChromaCbf cbf_cb[ xBase ][ yBase ][ trafoDepth - 1 ] ) {

readCbf = TRUE

if( blkldx = = 3 && log2TrafoSize < Log2MaxTrafoSize )

readCbf = cbf_cb[ xBase ][ yBase ][ trafoDepth ]

cbf_cb[ xBase + ( 1 « log2Trafo Width ) ][ yBase ][ trafoDepth ]

I I

1 1

cbf cb[ xBase 1[ yBase + ( 1 « log2TrafoHeight ) ^"|[ trafoDepth 1 if( !readCbf )

cbf_cb[ xO ][ yO ][ trafoDepth ] = 1

else

cbf_cb[ xO ][ yO ][ trafoDepth ] ae(v)

}

if( firstChromaCbf cbf_cr[ xBase ] [ yBase ] [ trafoDepth - 1 ] ) {

readCbf = TRUE

if( blkldx = = 3 && log2TrafoSize < Log2MaxTrafoSize )

readCbf = cbf_cr[ xBase ] [ yBase ] [ trafoDepth ]

cbf_cr[ xBase + ( 1 « log2Trafo Width ) ][ yBase ][ trafoDepth ]

I I

1 1

cbf cr[ xBase ] [ yBase + ( 1 « log2TrafoHeight ) ] [ trafoDepth ] if( !readCbf )

cbf_cr[ xO ][ yO ][ trafoDepth ] = 1

else

cbf_cr[ xO ][ yO ][ trafoDepth ] ae(v) }

}

}

if( split transform flagf xO ][ yO ][ trafoDepth ] ) {

if( InterTUSplitDirection = = 2 ) {

xl = xO + ( ( 1 « log2Trafo Width ) » 1 )

yl = yO

x2 = x0

y2 = yO + ( ( 1 « log2TrafoHeight ) » 1 )

x3 = xl

y3 = y2

} else {

xl = xO + ( ( 1 « log2Trafo Width ) » 2 ) * InterTUSplitDirection

yl = yO + ( ( 1 « log2TrafoHeight) » 2 ) * ( 1 - InterTUSplitDirection ) x2 = xl + ( ( 1 « log2Trafo Width ) » 2 ) * InterTUSplitDirection

y2 = yl + ( ( 1 « log2TrafoHeight) » 2 ) * ( 1 - InterTUSplitDirection ) x3 = x2 + ( ( 1 « log2Trafo Width ) » 2 ) * InterTUSplitDirection

y3 = y2 + ( ( 1 « log2TrafoHeight) » 2 ) * ( 1 - InterTUSplitDirection ) log2TrafoHeight = log2TrafoHeight + 2 * InterTUSplitDirection - 1 log2Trafo Width = log2Trafo Width - 2 * InterTUSplitDirection + 1

}

transform tree( xO, yO, xO, yO,

log2CbSize, log2Trafo Width - 1, log2TrafoHeight - 1,

trafoDepth + 1, 0 )

transform tree( xl, yl, xO, yO,

log2CbSize, log2Trafo Width - 1, log2TrafoHeight - 1,

trafoDepth + 1, 1 )

transform tree( x2, y2, xO, yO,

log2CbSize, log2Trafo Width - 1, log2TrafoHeight - 1,

trafoDepth + 1, 2 )

transform tree( x3, y3, xO, yO,

log2CbSize, log2Trafo Width - 1, log2TrafoHeight - 1,

trafoDepth + 1, 3 )

} else {

if( PredMode = = MODE INTRA | | trafoDepth I = 0 | |

cbf cb[ xO ][ yO ][ trafoDepth ] | |

cbf_cr[ xO ] [ yO ] [ trafoDepth ] ) {

readCbf = TRUE

if( blkldx = = 3 && PredMode != MODE INTRA &&

( ( log2CbSize <= Log2MaxTrafoSize+l ) | | ( log2TrafoSize <

Log2MaxTrafoSize ) )

readCbf = cbf_luma[ xBase ][ yBase ][ trafoDepth ]

cbf_luma[ xBase + ( 1 « log2Trafo Width ) ][ yBase ] [ trafoDepth ]

I I

1 1

cbf_luma[ xBase ] [ yBase + ( 1 « log2TrafoHeight ) ][ trafoDepth ]

I I

1 1

cbf_cb[ xBase ] [ yBase ] [ trafoDepth - 1]

cbf_cr[ xBase ][ yBase ][ trafoDepth - 1]

max transform hierarchy depth inter = =0 && trafoDepth==2 if( !readCbf )

cbf_luma[ xO ] [ yO ] [ trafoDepth ] = 1 Else

cbf_luma[ xO ][ yO ][ trafoDepth ] ae(v)

}

transform unit (xO, yO, xC,

yC, log2Trafo Width, log2TrafoHeight, trafoDepth, blkldx)

}

}

}

[0013] The second method proposes to directly remove the problem cases, i.e., disable the AMP for CU sizes larger than the largest TU size. The related modification of coding unit semantics in HEVC text specification are shown below (the modified parts are highlighted): [0014] The value of part mode is restricted as follows.

[0015] If PredMode is equal to MODE INTRA, part mode shall be equal to 0 or 1.

[0016] Otherwise (PredMode is equal to MODE INTER), the following applies

[0017] If log2CbSize is greater than Log2MinCbSize and smaller than Log2MaxTrafoSize+l, part mode shall be in the range of 0 to 2, inclusive and in the range of 4 to 7, inclusive.

[0018] Otherwise if log2CbSize is greater than Log2MinCbSize, part mode shall be in the range of 0 to 2, inclusive.

[0019] Embodiments of the TU split method, according to the present invention as described above may be implemented in various hardware, software codes, or a combination of both. For example, an embodiment of the present invention can be a circuit integrated into a video compression chip or program codes integrated into video compression software to perform the processing described herein. An embodiment of the present invention may also be program codes to be executed on a Digital Signal Processor (DSP) to perform the processing described herein. The invention may also involve a number of functions to be performed by a computer processor, a digital signal processor, a microprocessor, or field programmable gate array (FPGA). These processors can be configured to perform particular tasks according to the invention, by executing machine-readable software code or firmware code that defines the particular methods embodied by the invention. The software code or firmware codes may be developed in different programming languages and different format or style. The software code may also be compiled for different target platforms. However, different code formats, styles and languages of software codes and other means of configuring code to perform the tasks in accordance with the invention will not depart from the spirit and scope of the invention.

[0020] The invention may be embodied in other specific forms without departing from its spirit or essential characteristics. The described examples are to be considered in all respects only as illustrative and not restrictive. The scope of the invention is, therefore, indicated by the appended claims rather than by the foregoing description. All changes which come within the meaning and range of equivalency of the claims are to be embraced within their scope.

Claims

1. A method of splitting Transform Unit (TU) implicitly, comprising:

determining if a TU crosses a boundary of a Prediction Unit (PU); and

further splitting the TU if the TU crosses the boundary of the PU to ensure the TU not crossing the boundary of the PU.

2. A method of limit the asymmetric PU partition (AMP), comprising:

determining a Coding Unit (CU) size; and

disabling the AMP for the CU size if the CU size is larger than a largest Transform Unit (TU) size.