1. A video encoding method, the method comprising: determining a scanning method based on a partition mode of a prediction unit; and encoding information on the scanning method, wherein the scanning method is determined based on rate-distortion optimizatio5 n (RDO) among candidate scanning methods derived in view of a partition shape of the partition mode.

2. The method of claim 1, wherein the determining of the scanning method 10 derives horizontal scanning and zigzag scanning as the candidate scanning methods when the partition mode has a vertically oriented partition shape and derives vertical scanning and zigzag scanning as the candidate scanning methods when the partition mode has a horizontally oriented partition shape. 15 3. The method of claim 2, wherein the partition mode comprises an N x 2N mode, a 2N x N mode, a 2N x 2N mode, an N x N mode, a 2N x nU mode, a 2N x nD mode, an nL x 2N mode and an nR x 2N mode based on a size of the prediction unit for which inter prediction has been performed. 20 4. The method of claim 3, wherein the partition mode having the vertically oriented partition shape comprises the N x 2N mode, the nL x 2N mode and the nR x 2N mode, wherein the partition mode having the vertically oriented partition shape is a 1/2N x 2N mode of a left partition with a smaller partition size for the nL x 2N mode and the 25 partition mode having the vertically oriented partition shape is a 1/2N x 2N mode of a 41 right partition with a smaller partition size for the nR x 2N mode.

5. The method of claim 3, wherein the partition mode having the horizontally oriented partition shape comprises the 2N x N mode, the 2N x nU mode and the 2N x nD mode5 , wherein the partition mode having the horizontally oriented partition shape is a 2N x 1/2N mode of an upper partition with a smaller partition size for the 2N x nU mode and the partition mode having the horizontally oriented partition shape is es a 2N x 1/2N mode of a lower partition with a smaller partition size for the 2N x nD mode. 10

6. The method of claim 3, wherein the determining of the scanning method determines zigzag scanning as the scanning method when the partition mode is the 2N x 2N mode, the N x N mode, a 3/2N x 2N mode of a right partition with a larger partition size in the nL x 2N mode, a 3/2N x 2N mode of a left partition with a larger partition size 15 in the nR x 2N mode, a 2N x 3/2N mode of a lower partition with a larger partition size in the 2N x nU mode or a 2N x 3/2N mode of an upper partition with a larger partition size in the 2N x nD mode.

7. The method of claim 1, wherein the information on the scanning method 20 is indicated using a flag, and the flag indicates whether zigzag scanning is used.

8. A video encoding method, the method comprising: determining a scanning method based on a partition mode of a prediction unit for which short distance intra prediction (SDIP) has been performed; and 25 encoding information on the scanning method, 42 wherein the scanning method is determined based on rate-distortion optimization (RDO) among candidate scanning methods derived in view of a partition shape of the partition mode.

9. The method of claim 8, wherein the determining of the scanning metho5 d derives horizontal scanning and zigzag scanning as the candidate scanning methods when the partition mode has a vertically oriented partition shape and derives vertical scanning and zigzag scanning as the candidate scanning methods when the partition mode has a horizontally oriented partition shape. 10

10. The method of claim 9, wherein the partition mode comprises a 1/2N x 2N mode, a 2N x 1/2N mode, an N x N mode and a 2N x 2N mode based on a size of the prediction unit for which SDIP has been performed. 15 11. The method of claim 10, wherein the partition mode having the vertically oriented partition shape comprises is the 1/2N x 2N mode, and the partition mode having the horizontally oriented partition shape comprises the 2N x 1/2N mode.

12. The method of claim 10, wherein the determining of the scanning method 20 determines zigzag scanning as the scanning method when the partition mode is the N x N mode or the 2N x 2N mode.

13. The method of claim 8, wherein the information on the scanning method is indicated using a flag, and the flag indicates whether zigzag scanning is used. 25 43

14. A video decoding method, the method comprising: determining a scanning method based on a partition mode of a prediction unit; and inverse-scanning a transform coefficient according to the scanning method, wherein the scanning method is determined based on the partition mode using information signaled from an encoding apparatus, an5 d wherein the signaled information is a flag indicates whether zigzag scanning is used.

15. The method of claim 14, wherein the determining of the scanning method 10 decodes the flag indicating whether zigzag scanning is used and determines the scanning method based on a value of the decoded flag when the partition mode has a vertically oriented partition shape or horizontally oriented partition shape, wherein one of zigzag scanning and horizontal scanning is selected based on the value of the decoded flag when the partition mode has the vertically oriented partition 15 shape, and wherein one of zigzag scanning and vertical scanning is selected based on the value of the decoded flag when the partition mode has the horizontally oriented partition shape.

16. The method of claim 15, wherein the partition mode comprises an N x 2N 20 mode, a 2N x N mode, a 2N x 2N mode, an N x N mode, a 2N x nU mode, a 2N x nD mode, an nL x 2N mode and an nR x 2N mode based on a size of the prediction unit for which inter prediction is performed.

17. The method of claim 16, wherein the partition mode having the vertically 25 oriented partition shape comprises the N x 2N mode, the nL x 2N mode and the nR x 2N 44 mode, wherein the partition mode having the vertically oriented partition shape is a 1/2N x 2N mode of a left partition with a smaller partition size for the nL x 2N mode and the partition mode having the vertically oriented partition shape is a 1/2N x 2N mode of a right partition with a smaller partition size for the nR x 2N mode, and wherein the partition mode having the horizontally oriented partition shap5 e comprises the 2N x N mode, the 2N x nU mode and the 2N x nD mode, wherein s the partition mode having the horizontally oriented partition shape is a 2N x 1/2N mode of an upper partition with a smaller partition size for the 2N x nU mode and the partition mode having the horizontally oriented partition shape is a 2N x 1/2N mode of a lower partition 10 with a smaller partition size for the 2N x nD mode.

18. The method of claim 16, wherein the determining of the scanning method determines zigzag scanning as the scanning method when the partition mode is the 2N x 2N mode, the N x N mode, a 3/2N x 2N mode of a right partition with a larger partition 15 size in the nL x 2N mode, a 3/2N x 2N mode of a left partition with a larger partition size in the nR x 2N mode, a 2N x 3/2N mode of a lower partition with a larger partition size in the 2N x nU mode or a 2N x 3/2N mode of an upper partition with a larger partition size in the 2N x nD mode. 20 19. The method of claim 15, wherein the partition mode comprises a 1/2N x 2N mode, a 2N x 1/2N mode, an N x N mode and a 2N x 2N mode based on a size of the prediction unit in short distance intra prediction (SDIP), the partition mode having the vertically oriented partition shape comprising is the 1/2N x 2N mode and the partition mode having the horizontally oriented partition shape comprising the 2N x 1/2N mode. 25 45

20. The method of claim 19, wherein the determining of the scanning method determines zigzag scanning as the scanning method when the partition mode is the N x N mode or the 2N x 2N mode.

【DESCRIPTION】
【Technical Field5 】
[1] The present invention relates to video encoding and decoding, and more
particularly, to a method and an apparatus for determining a scan method based on a
partition mode of a prediction unit and encoding/decoding information thereof.
【Background Art】
10 [2] Recently, demands for high-resolution and high-quality videos, such as highdefinition
(HD) and ultrahigh-definition (UHD) videos, are increasing in various fields of
applications. As video data has higher resolution and higher quality, the amount of data
more increases relative to existing usual video data. Accordingly, when video data is
transferred using media such as existing wired and wireless broad band lines or is stored in
15 existing storage media, transfer cost and storage cost increase. In order to solve these
problems occurring with an increase in resolution and quality of video data, highefficiency
video compression techniques may be utilized.
[3] Video compression technology includes various techniques, such as an inter
prediction technique of predicting pixel values included in a current picture from previous
20 or subsequent pictures of the current picture, an intra prediction technique of predicting
pixel values included in a current picture using pixel information in the current picture,
and an entropy encoding technique of assigning a short code to a value with a high
appearance frequency and assigning a long code to a value with a low appearance
frequency. Video data may be effectively compressed and transferred or stored using
25 such video compression techniques.
3
【Disclosure】
【Technical Problem】
[4] An aspect of the present invention is to provide a method of determining a
scan method based on a partition mode of a prediction unit for improving video
encoding/decoding efficiency and encoding/decoding the scan method5 .
[5] Another aspect of the present invention is to provide an apparatus for
determining a scan method based on a partition mode of a prediction unit for improving
video encoding/decoding efficiency and encoding/decoding the scan method.
【Technical Solution】
10 [6] An embodiment of the present invention provides a video encoding method.
The method includes determining a scanning method based on a partition mode of a
prediction unit, and encoding information on the scanning method, wherein the scanning
method is determined based on rate-distortion optimization (RDO) among candidate
scanning methods derived in view of a partition shape of the partition mode.
15 [7] The determining of the scanning method may derive horizontal scanning and
zigzag scanning as the candidate scanning methods when the partition mode has a
vertically oriented partition shape and derive vertical scanning and zigzag scanning as the
candidate scanning methods when the partition mode has a horizontally oriented partition
shape.
20 [8] The partition mode may include an N x 2N mode, a 2N x N mode, a 2N x 2N
mode, an N x N mode, a 2N x nU mode, a 2N x nD mode, an nL x 2N mode and an nR x
2N mode based on a size of the prediction unit for which inter prediction has been
performed.
[9] The partition mode having the vertically oriented partition shape may include
25 the N x 2N mode, the nL x 2N mode and the nR x 2N mode, wherein the partition mode
4
having the vertically oriented partition shape is a 1/2N x 2N mode of a left partition with a
smaller partition size for the nL x 2N mode and the partition mode having the vertically
oriented partition shape is a 1/2N x 2N mode of a right partition with a smaller partition
size for the nR x 2N mode.
[10] The partition mode having the horizontally oriented partition shape ma5 y
include the 2N x N mode, the 2N x nU mode and the 2N x nD mode, wherein the partition
mode having the horizontally oriented partition shape is a 2N x 1/2N mode of an upper
partition with a smaller partition size for the 2N x nU mode and the partition mode having
the horizontally oriented partition shape is es a 2N x 1/2N mode of a lower partition with a
10 smaller partition size for the 2N x nD mode.
[11] The determining of the scanning method may determine zigzag scanning as
the scanning method when the partition mode is the 2N x 2N mode, the N x N mode, a
3/2N x 2N mode of a right partition with a larger partition size in the nL x 2N mode, a
3/2N x 2N mode of a left partition with a larger partition size in the nR x 2N mode, a 2N x
15 3/2N mode of a lower partition with a larger partition size in the 2N x nU mode or a 2N x
3/2N mode of an upper partition with a larger partition size in the 2N x nD mode.
[12] The information on the scanning method may be indicated using a flag, and
the flag may indicate whether zigzag scanning is used.
[13] Another embodiment of the present invention provides a video encoding
20 method. The method includes determining a scanning method based on a partition mode
of a prediction unit for which short distance intra prediction (SDIP) has been performed,
and encoding information on the scanning method, wherein the scanning method is
determined based on RDO among candidate scanning methods derived in view of a
partition shape of the partition mode.
25 [14] The determining of the scanning method may derive horizontal scanning and
5
zigzag scanning as the candidate scanning methods when the partition mode has a
vertically oriented partition shape and derive vertical scanning and zigzag scanning as the
candidate scanning methods when the partition mode has a horizontally oriented partition
shape.
[15] The partition mode may include a 1/2N x 2N mode, a 2N x 1/2N mode, an 5 N
x N mode and a 2N x 2N mode based on a size of the prediction unit for which SDIP has
been performed.
[16] The partition mode having the vertically oriented partition shape may
include is the 1/2N x 2N mode, and the partition mode having the horizontally oriented
10 partition shape may include the 2N x 1/2N mode.
[17] The determining of the scanning method may determine zigzag scanning as
the scanning method when the partition mode is the N x N mode or the 2N x 2N mode.
[18] The information on the scanning method may be indicated using a flag, and
the flag may indicate whether zigzag scanning is used.
15 [19] Still another embodiment of the present invention provides a video encoding
method. The method includes determining a scanning method based on a partition mode
of a prediction unit, and inverse-scanning a transform coefficient according to the
scanning method, wherein the scanning method is determined based on the partition mode
using information signaled from an encoding apparatus, and the signaled information is a
20 flag indicates whether zigzag scanning is used.
[20] The determining of the scanning method may decode the flag indicating
whether zigzag scanning is used and determine the scanning method based on a value of
the decoded flag when the partition mode has a vertically oriented partition shape or
horizontally oriented partition shape, in which one of zigzag scanning and horizontal
25 scanning may be selected based on the value of the decoded flag when the partition mode
6
is the vertically oriented partition shape, and one of zigzag scanning and vertical scanning
may be selected based on the value of the decoded flag when the partition mode has the
horizontally oriented partition shape.
[21] The partition mode may include an N x 2N mode, a 2N x N mode, a 2N x
2N mode, an N x N mode, a 2N x nU mode, a 2N x nD mode, an nL x 2N mode and an n5 R
x 2N mode based on a size of the prediction unit for which inter prediction is performed.
[22] the partition mode having the vertically oriented partition shape comprises
the N x 2N mode, the nL x 2N mode and the nR x 2N mode, wherein the partition mode
having the vertically oriented partition shape is a 1/2N x 2N mode of a left partition with a
10 smaller partition size for the nL x 2N mode and the partition mode having the vertically
oriented partition shape is a 1/2N x 2N mode of a right partition with a smaller partition
size for the nR x 2N mode, and wherein the partition mode having the horizontally
oriented partition shape comprises the 2N x N mode, the 2N x nU mode and the 2N x nD
mode, wherein s the partition mode having the horizontally oriented partition shape is a
15 2N x 1/2N mode of an upper partition with a smaller partition size for the 2N x nU mode
and the partition mode having the horizontally oriented partition shape is a 2N x 1/2N
mode of a lower partition with a smaller partition size for the 2N x nD mode
[23] The determining of the scanning method may determine zigzag scanning as
the scanning method when the partition mode is the 2N x 2N mode, the N x N mode, a
20 3/2N x 2N mode of a right partition with a larger partition size in the nL x 2N mode, a
3/2N x 2N mode of a left partition with a larger partition size in the nR x 2N mode, a 2N x
3/2N mode of a lower partition with a larger partition size in the 2N x nU mode or a 2N x
3/2N mode of an upper partition with a larger partition size in the 2N x nD mode.
[24] The partition mode may include a 1/2N x 2N mode, a 2N x 1/2N mode, an N
25 x N mode and a 2N x 2N mode based on a size of the prediction unit in SDIP, the partition
7
mode having the vertically oriented partition shape including is the 1/2N x 2N mode and
the partition mode having the horizontally oriented partition shape including the 2N x
1/2N mode.
[25] The determining of the scanning method may determine zigzag scanning as
the scanning method when the partition mode is the N x N mode or the 2N x 2N mode5 .
【Advantageous Effects】
[26] According to the present invention, a scanning method for transform
coefficients is determined using a partition mode of a prediction unit, that is, particular
directivity or particular texture of the prediction unit, thereby increasing efficiency in
10 encoding and decoding.
【Description of Drawings】
[27] FIG. 1 is a block diagram illustrating a video encoding apparatus according
to an exemplary embodiment of the present invention.
[28] FIG. 2 is a block diagram illustrating a video decoding apparatus according
15 to an exemplary embodiment of the present invention.
[29] FIG. 3 schematically illustrates a coefficient scanning method according to
the present invention.
[30] FIG. 4 illustrates a method of determining and encoding a scanning method
based on a partition mode of a prediction unit according to an exemplary embodiment of
20 the present invention.
[31] FIG. 5 illustrates a method of determining and encoding a scanning method
in asymmetric motion partition (AMP) according to an exemplary embodiment of the
present invention.
[32] FIG. 6 illustrates a method of determining and encoding a scanning method
25 in short distance intra prediction (SDIP) according to an exemplary embodiment of the
8
present invention.
[33] FIG. 7 is a flowchart illustrating a video encoding process according to the
present invention.
[34] FIG. 8 is a flowchart illustrating a video decoding process according to the
present invention5 .
【Mode for Invention】
[35] The present invention may be changed and modified variously and be
illustrated with reference to different exemplary embodiments, some of which will be
described and shown in the drawings. However, these embodiments are not intended for
10 limiting the invention but are construed as including includes all modifications,
equivalents and replacements which belong to the spirit and technical scope of the
invention. Like reference numerals in the drawings refer to like elements throughout.
[36] Although the terms first, second, etc. may be used to describe various
elements, these elements should not be limited by these terms. These terms are used only
15 to distinguish one element from another element. For example, a first element could be
termed a second element and a second element could be termed a first element likewise
without departing from the teachings of the present invention. The term "and/or"
includes any and all combinations of a plurality of associated listed items.
[37] It will be understood that when an element is referred to as being "connected
20 to" or “coupled to" another element, the element can be directly connected or coupled to
another element or intervening elements. On the contrary, when an element is referred to
as being "directly connected to" or "directly coupled to" another element, there are no
intervening elements present.
[38] The terminology used herein is for the purpose of describing particular
25 embodiments only and is not intended to be limiting of the invention. As used herein,
9
the singular forms "a," "an" and "the" are intended to include the plural forms as well,
unless the context clearly indicates otherwise. It will be further understood that the terms
"include" and/or "have," when used in this specification, specify the presence of stated
features, integers, steps, operations, elements, and/or components, but do not preclude the
presence or addition of one or more other features, integers, steps, operations, elements5 ,
components, and/or groups thereof.
[39] Hereinafter, exemplary embodiments of the invention will be described in
detail with reference to the accompanying drawings. Like reference numerals in the
drawings refer to like elements throughout, and redundant descriptions of like elements
10 will be omitted herein.
[40] FIG. 1 is a block diagram illustrating a video encoding apparatus according
to an exemplary embodiment of the present invention.
[41] Referring to FIG. 1, the video encoding apparatus 100 includes a picture
partition module 110, prediction modules 120 and 125, a transform module 130, a
15 quantization module 135, a rearrangement module 160, an entropy encoding module 165,
a dequantization module 140, an inverse transform module 145, a filter module 150 and a
memory 155.
[42] Although elements illustrated in FIG. 1 are independently shown so as to
represent different distinctive functions in the video encoding apparatus, such a
20 configuration does not indicate that each element is constructed by a separate hardware
constituent or software constituent. That is, the elements are independently arranged for
convenience of description, wherein at least two elements may be combined into a single
element, or a single element may be divided into a plurality of elements to perform
functions. It is to be noted that embodiments in which some elements are integrated into
25 one combined element and/or an element is divided into multiple separate elements are
10
included in the scope of the present invention without departing from the essence of the
present invention.
[43] Some elements are not essential to the substantial functions in the invention
and may be optional constituents for merely improving performance. The invention may
be embodied by including only constituents essential to embodiment of the invention5 ,
except for constituents used to merely improve performance. The structure including
only the essential constituents except for the optical constituents used to merely improve
performance belongs to the scope of the invention.
[44] The picture partition module 110 may partition an input picture into at least
10 one process unit. Here, the process unit may be a prediction unit (PU), a transform unit
(TU) or a coding unit (CU). The picture partition module 110 may partition one picture
into a plurality of combinations of CUs, PUs and TUs and select one combination of CUs,
PUs and TUs on the basis of a predetermined criterion (for example, a cost function),
thereby encoding the picture.
15 [45] For example, one picture may be partitioned into a plurality of CUs. A
recursive tree structure, such as a quad tree structure, may be used to partition a picture
into CUs. A CU, for which a picture or a CU of a maximum size may be as root, may be
partitioned into sub-coding units with as many child nodes as the partitioned CUs. A CU
which is not partitioned any more in accordance with a predetermined limitation is a leaf
20 node. That is, assuming that a CU may be partitioned into quadrants only, a single CU
may be partitioned into at most four different CUs.
[46] In the embodiments of the invention, a CU may be used to refer to not only a
unit of encoding but also a unit of decoding.
[47] A PU may be partitioned into at least one square or rectangular form with the
25 same size in a CU. For PUs partitioned from a same CU, a PU may have different shape
11
from other PUs.
[48] When a PU for intra prediction is generated based on a CU and the CU is not
a minimum CU, the CU may be subjected to intra prediction without being partitioned into
plural PUs (N´N).
[49] The prediction modules 120 and 125 may include an inter prediction modul5 e
120 to perform inter prediction and an intra prediction module 125 to perform intra
prediction. The prediction modules 120 and 125 may determine which of inter
prediction and intra prediction should be performed on a PU, and may determine specific
information (for example, an intra prediction mode, a motion vector, and a reference
10 picture) of the determined prediction method. Here, a process unit on which prediction is
performed may be different from a process unit for which a prediction method and
specific information thereon are determined. For example, a prediction method and a
prediction mode may be determined for each PU, while prediction may be performed for
each TU. A residual value (residual block) between a generated predicted block and an
15 original block may be input to the transform module 130. Further, prediction mode
information, motion vector information and the like used for prediction may be encoded
along with the residual value by the entropy encoding module 165 and be transmitted to
the decoding apparatus. When a specific encoding mode is used, the original block may
be encoded and transmitted to the decoding apparatus without generating a prediction
20 block by the prediction modules 120 and 125.
[50] The inter prediction module 120 may predict a PU on the basis of
information on at least one picture among a previous picture of a current picture and a
subsequent picture of a current picture. The inter prediction module 120 may include a
reference picture interpolation module, a motion prediction module, and a motion
25 compensation module.
12
[51] The reference picture interpolation module may be supplied with reference
picture information from the memory 155 and generate pixel information less than an
integer pixel from a reference picture. In the case of luma pixels, a DCT-based 8-tap
interpolation filter with a variable filter coefficient may be used to generate information on
a pixel smaller than an integer pixel in a unit of a 1/4 pixel. In the case of chroma pixels5 ,
a DCT-based 4-tap interpolation filter with a variable filter coefficient may be used to
generate information on a pixel smaller than an integer pixel in a unit of a 1/8 pixel.
[52] The motion prediction module may perform motion prediction on the basis of
the reference picture interpolated by the reference picture interpolation module. Various
10 methods, such as a full search-based block matching algorithm (FBMA), a three-step
search (TSS) algorithm and a new three-step search (NTS) algorithm, may be used to
calculate a motion vector. A motion vector has a motion vector value in the unit of a 1/2
or 1/4 pixel on the basis of an interpolated pixel. The motion prediction module may
predict a current PU using different motion prediction methods. Various methods, such
15 as skip mode, merge mode, and advanced motion vector prediction (AMVP), etc. may be
used as the motion prediction method.
[53] The intra prediction module 125 may generate a PU on the basis of
information on a reference pixel neighboring to a current block. The information on a
reference pixel neighboring to the current block is pixel information in a current picture.
20 When a reference pixel is a pixel for which inter prediction has been performed because a
block, which includes the reference pixel, neighboring to the current PU is a block for
which inter prediction has been performed, information on a reference pixel included in
the block for which inter prediction has been performed may be replaced with information
on a reference pixel in a block for which intra prediction has been performed. That is,
25 when a reference pixel is not available, information on the unavailable reference pixel may
13
be replaced with information on at least one reference pixel of the available reference
pixels.
[54] A prediction mode of intra prediction includes a directional prediction mode
in which reference pixel information is used according to a prediction direction and a nondirectional
prediction mode in which information on direction is not used in performin5 g
prediction. A mode for predicting luma information and a mode for predicting chroma
information may be different from each other. Further, intra prediction mode
information used to obtain luma information or predicted luma signal information may be
used to predict chroma information.
10 [55] If a PU and a TU have the same size when intra prediction is performed, intra
prediction on the PU may be performed based on left pixels, an upper-left pixel and upper
pixels of the PU. On the other hand, if a PU and a TU have different sizes when intra
prediction is performed, intra prediction may be performed using reference pixels based
on the TU. Intra prediction using N´N partitioning may be performed only for a
15 minimum CU.
[56] In the intra prediction method, a predicted block may be generated according
to the prediction mode after an adaptive intra smoothing (AIS) filter being applied.
Different types of AIS filters may be applied to the reference pixels. In the intra
prediction method, the intra prediction mode of a current PU may be predicted from an
20 intra prediction mode of a PU neighboring to the current PU. In predicting the prediction
mode of the current PU using mode information predicted from a neighboring PU, when
the current PU and the neighboring PU have the same intra prediction mode, information
indicating that the current PU and the neighboring PU have the same prediction mode may
be transmitted using predetermined flag information. When the current PU and the
25 neighboring PU have different prediction modes, information on the prediction mode of
14
the current block may be encoded by entropy encoding.
[57] A residual block including residual information which is a difference
between the original block of the PU and the predicted block of a PU generated based on
the PU generated by the prediction modules 120 and 125, may be generated. The
generated residual block may be input to the transform module 1305 .
[58] The transform module 130 may transform the residual block using a
transform method such as Discrete Cosine Transform (DCT) or Discrete Sine Transform
(DST). The residual block includes information on the residual between the PU
generated by the prediction modules 120 and 125 and the original block. A transform
10 method to be used to transform the residual block may be determined among DCT and
DST on the basis of the information on the intra prediction mode applied to the PU which
is used to generate the residual block.
[59] The quantization module 135 may quantize values transformed into a
frequency domain by the transform module 130. A quantization coefficient may change
15 depending on a block or importance of a picture. Values output from the quantization
module 135 may be provided to the dequantization module 140 and the rearrangement
module 160.
[60] The rearrangement module 160 may rearrange coefficients with respect to
quantized residual values.
20 [61] The rearrangement module 160 may change a two-dimensional (2D) block of
coefficients into a one-dimensional (1D) vector of coefficients through coefficient
scanning. For example, the rearrangement module 125 may change a 2D block of
coefficients into a 1D vector of coefficients by scanning from DC coefficients to
coefficients of a high frequency domain using zigzag scanning. Vertical scanning for
25 scanning a 2D block of coefficients in a vertical direction and horizontal scanning for
15
scanning a 2D block of coefficients in a horizontal direction may be used depending on a
size of a TU and an intra prediction mode, instead of zigzag scanning. That is, a
scanning method for use may be selected based on the size of the TU and the intra
prediction mode among zigzag scanning, vertical scanning, and horizontal scanning.
[62] The entropy encoding module 165 may perform entropy encoding on th5 e
basis of the values obtained by the rearrangement module 160. Various encoding
methods, such as exponential Golomb coding, context-adaptive variable length coding
(CAVLC), or context-adaptive binary arithmetic coding (CABAC), may be used for
entropy encoding.
10 [63] The entropy encoding module 165 may encode a variety of information, such
as residual coefficient information and block type information on a CU, prediction mode
information, partitioning unit information, PU information, transfer unit information,
motion vector information, reference frame information, block interpolation information
and filtering information from the rearrangement module 160 and the prediction modules
15 120 and 125.
[64] The entropy encoding module 165 may entropy-encode coefficients of a CU
input from the rearrangement module 160.
[65] The dequantization module 140 and the inverse transform module 145
dequantize the values which are quantized by the quantization module 135 and inversely
20 transform the values which are transformed by the transform module 130. The residual
values generated by the dequantization module 140 and the inverse transform module 145
may be added to the predicted PU. The predicted PU may be predicted by the motion
vector prediction module, the motion compensation module, and the intra prediction
module of the prediction modules 120 and 125. A reconstructed block may be generated
25 by adding the residual values to the predicted PU (predicted values).
16
[66] The filter module 150 may include at least one of a deblocking filter, an
offset module, and an adaptive loop filter (ALF).
[67] The deblocking filter may remove block distortion generated on boundaries
between blocks in a reconstructed picture. Whether to apply the deblocking filter to a
current block may be determined on the basis of pixels included in several rows o5 r
columns of the block. When the deblocking filter is applied to a block, a strong filter or
a weak filter may be applied depending on a required deblocking filtering strength.
When horizontal filtering and vertical filtering are performed in applying the deblocking
filter, the horizontal filtering and vertical filtering may be performed in parallel.
10 [68] The offset module may apply an offset form the original picture in the unit of
a pixel to the picture for which the deblocking filtering process is completed. , A region to
which the offset may be applied may be determined after partitioning pixels of a picture
into a predetermined number of regions. The offset may be applied to the determined
area in consideration of edge information on each pixel and the method of applying the
15 offset to the determined area..
[69] The ALF may perform filtering based on a comparison result of the filtered
reconstructed picture and the original picture. Pixels included in a picture may be
partitioned into predetermined groups, a filter to be applied to each group may be
determined, and differential filtering may be performed for each group. Information on
20 whether to apply the ALF may be transferred by each coding unit (CU) and a size and
coefficient of an ALF to be applied to each block may vary. The ALF may have various
types and a number of coefficients included in a corresponding filter may vary. Further,
an ALF filter with the same form (fixed form) may be applied to a block regardless of
characteristics of the block.
25 [70] The memory 155 may store a reconstructed block or picture output from the
17
filter module 150, and the stored reconstructed block or picture may be supplied to the
prediction modules 120 and 125 when performing inter prediction.
[71] FIG. 2 is a block diagram illustrating a video decoding apparatus according
an exemplary embodiment of the present invention.
[72] Referring to FIG. 2, the video decoding apparatus 200 may include a5 n
entropy decoding module 210, a rearrangement module 215, a dequantization module 220,
an inverse transform module 225, prediction modules 230 and 235, a filter module 240,
and a memory 245.
[73] When a video bitstream is input from the video encoding apparatus, the input
10 bitstream may be decoded according to an inverse process of the video encoding process
performed in the video encoding apparatus.
[74] The entropy decoding module 210 may perform entropy decoding according
to an inverse process of the entropy encoding process by the entropy encoding module of
the video encoding apparatus. For example, various methods, such as exponential
15 Golomb coding, CAVLC or CABAC, may be used for entropy encoding, corresponding to
the method used by the video encoding apparatus.
[75] The entropy decoding module 210 may decode information associated with
intra prediction and inter prediction performed by the encoding apparatus.
[76] The rearrangement module 215 may perform rearrangement on the bit stream
20 entropy-decoded by the entropy decoding module 210 on the basis of the rearrangement
method of the encoding module. The rearrangement module 215 may reconstruct and
rearrange coefficients in a 1D vector form into coefficients in a 2D block. The
rearrangement module 215 may be provided with information on coefficient scanning
performed by the encoding apparatus and may perform rearrangement using a method of
25 inversely scanning the coefficients on the basis of scanning order in which scanning is
18
performed by the encoding apparatus.
[77] The dequantization module 220 may perform dequantization on the basis of a
quantization parameter provided from the encoding apparatus and the rearranged
coefficients of the block.
[78] The inverse transform module 225 may perform inverse DCT and invers5 e
DST on a result of quantization performed by the video encoding apparatus, having been
subjected to DCT and DST performed by the transform module. Inverse transform may
be performed on the basis of a transfer unit determined by the video encoding apparatus.
The transform module of the video encoding apparatus may selectively perform DCT and
10 DST depending on a plurality of information elements, such as a prediction method, a size
of the current block and a prediction direction, etc. and the inverse transform module 225
of the video decoding apparatus may perform inverse transform on the basis of
information on the transform performed by the transform module of the video encoding
apparatus.
15 [79] The prediction modules 230 and 235 may generate a prediction block
(predicted block) on the basis of prediction block generation information supplied from
the entropy decoding module 210 and information on a previously-decoded block or
picture provided from the memory 245.
[80] Similarly to the operation of the video encoding apparatus as described above,
20 if a PU and a TU have the same size when intra prediction is performed, intra prediction
on the PU is performed based on left pixels, an upper-left pixel and upper pixels of the PU.
On the other hand, if a PU and a TU have different sizes when intra prediction is
performed, intra prediction may be performed using reference pixels based on the TU.
Intra prediction using N´N partitioning may be used only for a minimum CU.
25 [81] The prediction modules 230 and 235 may include a PU determination
19
module, an inter prediction module and an intra prediction module. The PU
determination module may receive a variety of information, such as PU information,
prediction mode information on an intra prediction method and motion prediction-related
information on an inter prediction method, etc. from the entropy decoding module 210,
may determine a PU for a current CU. The PU determination module may determin5 e
which of the inter prediction and the intra prediction is performed on the PU. An inter
prediction module 230 may perform inter prediction on a current PU on the basis of
information on at least one picture among a previous picture and a subsequent picture of a
current picture including the current PU. An inter prediction module 230 may use
10 information necessary for inter prediction for the current PU provided from the video
encoding apparatus.
[82] In order to perform inter prediction, it may be determined on the basis of a
CU whether a motion prediction method for a PU included in the CU is a skip mode, a
merge mode or an AMVP mode.
15 [83] An intra prediction module 235 may generate a prediction block on the basis
of pixel information in a current picture. When a PU is a PU for which intra prediction is
performed, intra prediction may be performed based on intra prediction mode information
on the PU provided from the video encoding apparatus. The intra prediction module 235
may include an AIS filter, a reference pixel interpolation module, and a DC filter. The
20 AIS filter performs filtering on reference pixels of a current block The AIS filter may
decide whether to apply the filter or not depending on a prediction mode for the current
PU. AIS filtering may be performed on the reference pixels of the current block using
the prediction mode for the PU and information on the AIS filter provided from the video
encoding apparatus. When the prediction mode for the current block is a mode not
25 performing AIS filtering, the AIS filter may not be applied.
20
[84] When the prediction mode for the PU is a prediction mode of performing
intra prediction on the basis of pixel values obtained by interpolating the reference pixels,
the reference pixel interpolation module may generate reference pixels in a unit of a
fractional pixel less than an integer pixel (i.e. full pixel) by interpolating the reference
pixels. When the prediction mode for the current PU is a prediction mode of generatin5 g
a prediction block without interpolating the reference pixels, the reference pixels may not
be interpolated. The DC filter may generate a prediction block through filtering when
the prediction mode for the current block is the DC mode.
[85] The reconstructed block or picture may be provided to the filter module 240.
10 The filter module 240 includes a deblocking filter, an offset module, and an ALF.
[86] The video encoding apparatus mat provide information on whether the
deblocking filter is applied to a corresponding block or picture, and information on which
of a strong filter and a weak filter is applied when the deblocking filter is used. The
deblocking filter of the video decoding apparatus may be provided with information on the
15 deblocking filter from the video encoding apparatus and may perform deblocking filtering
on a corresponding block.
[87] The offset module may apply offset to the reconstructed picture on the basis
of information on an offset type and offset value applied to the picture in the encoding
process.
20 [88] The ALF may be applied to a CU on the basis of information on whether the
ALF is applied and ALF coefficient information, etc. provided from the encoding
apparatus. The ALF information may be included and provided in a specific parameter
set.
[89] The memory 245 may store the reconstructed picture or block for use as a
25 reference picture or a reference block and may provide the reconstructed picture to an
21
output module.
[90] As described above, in the embodiments of the invention, the term "coding
unit" is used as an encoding unit for a convenience of descriptions. However, the term
“coding unit” may be also used as a unit of decoding.
[91] Hereinafter, scanning methods based on prediction modes and partitio5 n
modes in prediction to be illustrated in FIGS. 3 to 8 according to exemplary embodiments
of the present invention may be achieved in accordance with functions of the modules of
the encoding apparatus and the decoding apparatus described above in FIGS. 1 and 2,
which fall within the scope of the present invention.
10 [92] FIG. 3 schematically illustrates coefficient scanning methods according to
the present invention.
[93] Referring to FIG. 3, the scanning methods may include horizontal scanning
310, vertical scanning 320, and zigzag scanning 330 or upright diagonal scanning 340.
Here, one of these scanning methods shown in FIG. 3 may be used based on a partitioned
15 shape of a PU, and a 2D block of quantized transform coefficients may be changed into a
1D vector of transform coefficients by scanning.
[94] Horizontal scanning 310, which scans transform coefficients in a horizontal
direction, may be applied, for example, when a PU is a partition which is a vertically
oriented block, such as an N x 2N block. The vertically oriented block is highly likely to
20 include a texture of a vertical component, in which the transform coefficients are highly
likely to be distributed in the horizontal direction. Thus, a scanning order shown in the
method 310 of FIG. 3 may be applied to scanning the transform coefficients.
[95] Vertical scanning 320, which scans transform coefficients in a vertical
direction, may be applied, for example, when a PU is a partition which is a horizontally
25 oriented block, such as a 2N x N block. The horizontally oriented block is highly likely
22
to include texture of horizontal component, in which the transform coefficients are highly
likely to be distributed in the vertical direction. Thus, a scanning order shown in the
method 320 of FIG. 3 may be applied to scanning the transform coefficients.
[96] Zigzag scanning 330 or upright diagonal scanning 340 may be applied when
a PU do not have particular directivity or particular component of texture. For example5 ,
zigzag scanning 330 or upright diagonal scanning 340 may be applied to 2N x 2N or N x
N square block.
[97] The scanning methods of FIG. 3 are provided for examples of the present
invention, and the present invention is not limited thereto. Alternative scanning methods
10 performed in different orders may be also used as well as the scanning methods of FIG. 3.
[98] As described above, when a PU is a partition such as an Nx2N block or a
2NxN block, these blocks may be highly likely to have particular component of texture or
strong directionality. Accordingly horizontal scanning or vertical scanning is used
depending on a partition shape of the PU. However, even though a PU is a partition such
15 as an Nx2N block or a 2NxN block, these blocks may have insignificant directionality or
not include particular component of texture. In this case, it may not be efficient to use
the particular scanning methods, for example, horizontal scanning for an Nx2N block and
vertical scanning for a 2NxN block. Thus, a method of effectively scanning and
encoding transform coefficients is needed.
20 [99] FIG. 4 illustrates a method of determining a scanning method and encoding
information thereon based on a partition mode of a PU according to an exemplary
embodiment of the present invention.
[100] Referring to FIG. 4, a single CU of an inter prediction mode may be
partitioned into PUs with the same size or different sizes. For example, the CU may be
25 partitioned into 2N x N block 400, N x 2N block 410, 2N x 2N block 420 or N x N block
23
430. Partition modes PartMode of the PUs may be determined based on sizes of the
partitioned PUs.
[101] Partition modes PartMode of PUs may include a PART_2NxN mode in
which a CU is partitioned into 2N x N 400 blocks, a PART_Nx2N mode in which a CU is
partitioned into N x 2N 410 blocks, a PART_2Nx2N mode in which a CU is partitione5 d
into 2N x 2N 420 blocks, and a PART_NxN mode in which a CU is partitioned into N x N
430 blocks.
[102] In the present embodiment, a scanning method is determined based on a
partition mode of a PU, in which a partitioned shape of the partition mode may be
10 considered. That is, candidate scanning methods may be obtained in view of partitioned
shapes of PUs, among which a scanning method may be determined based on ratedistortion
optimization (RDO).
[103] When the partition mode indicates a horizontally oriented shape, for
example, the partition mode is the PART_2NxN mode in which a CU is partitioned into
15 2N x N 400 blocks, the blocks may be likely to have particular component of texture or
directionality (for example, horizontal component of texture or transform coefficients
distributed in the vertical direction). Vertical scanning may be derived as a candidate
scanning method in view of such a partitioned shape. Also, zigzag scanning (or upright
diagonal scanning) may be derived as a candidate scanning method considering that the
20 blocks are likely not to have particular component of texture or directionality. That is,
for the partition mode of the horizontally oriented shape, a scanning method having
minimum RDO may be selected among the two candidate scanning methods, vertical
scanning and zigzag scanning (or upright diagonal scanning).
[104] Alternatively, when the partition mode indicates a vertically oriented shape,
25 for example, the partition mode is the PART_Nx2N mode in which a CU is partitioned
24
into N x 2N 410 blocks, the blocks may be likely to have particular component of texture
or directionality (for example, vertical component of texture or transform coefficients
distributed in the horizontal direction). Horizontal scanning may be derived as a
candidate scanning method in view of such a partitioned shape. Also, zigzag scanning
(or upright diagonal scanning) may be derived as a candidate scanning metho5 d
considering that the blocks are likely not to have particular component of texture or
directionality. That is, for the partition mode of a vertically oriented shape, a scanning
method having minimum RDO may be selected among the two candidate scanning
methods, horizontal scanning and zigzag scanning (or upright diagonal scanning).
10 [105] Meanwhile, for the partition mode of a square shape, for example, the
PART_2Nx2N mode in which a CU is partitioned into 2N x 2N 420 blocks or the
PART_NxN mode in which a CU is partitioned in N x N 430 blocks, zigzag scanning (or
upright diagonal scanning) may be used.
[106] Table 1 illustrates available scanning methods according to partition modes
15 of PUs according to the exemplary embodiment of the present invention. Here, in the
PART_2NxN mode and the PART_Nx2N mode, one scanning method may be selected in
view of RDO from two candidate scanning methods.
[107] [Table 1]
PU Partition mode Scan pattern
PART_2NxN Vertical scanning/Zigzag scanning
PART_Nx2N Horizontal scanning/Zigzag scanning
PART_2Nx2N Zigzag scanning
PART_NxN Zigzag scanning
[108] When a scanning method is determined based on a partition mode of a PU
20 as described above, transform coefficients may be scanned using the determined scanning
25
method. Information on the determined scanning method may be encoded and
transmitted to the decoding apparatus. The information on the scanning method may be
indicated using a flag, for example, a flag isZigZagScanFlag indicating whether zigzag
scanning is used.
[109] For instance, when the partition mode of the PU is the PART_2NxN mode5 ,
information on a determined scanning method of vertical scanning and zigzag scanning
(or upright diagonal scanning) may be encoded using a flag, and the flag information may
be transmitted to the decoding apparatus. In the PART_2NxN mode, isZigZagScanFlag
may be set to 1 if zigzag scanning is determined to be performed, and isZigZagScanFlag
10 may be set to 0 if vertical scanning is determined to be performed. Alternatively, when
the partition mode of the PU is the PART_Nx2N, information on a determined scanning
method of horizontal scanning and zigzag scanning (or upright diagonal) scanning may be
encoded using a flag, for example, isZigZagScanFlag, and the flag information may be
transmitted to the decoding apparatus.
15 [110] FIG. 5 illustrates a method of determining a scanning method and encoding
the information thereon in asymmetric motion partition (AMP) according to an exemplary
embodiment of the present invention.
[111] As described above, a single CU of an inter prediction mode may be
partitioned into PUs with the same size or different sizes. As shown in FIG. 5, a 64 x 64
20 block may be partitioned into 16 x 64 block, 48 x 64 block, 64 x 16 block or 64 x 48 block
i.e. blocks of different shapes. This partition mode is referred to as AMP. AMP may be
applied to partitioning a CU to enhance coding efficiency when a picture includes
irregular image patterns.
[112] From left of FIG. 5, AMP includes a PART_nLx2N mode in which a CU is
25 partitioned into blocks with size of nL x 2N 500, a PART_nRx2N mode in which a CU is
26
partitioned into blocks with size of nR x 2N 510, a PART_2NxnU mode in which a CU is
partitioned into blocks with size of 2N x nU 520, and a PART_2NxnD mode in which a
CU is partitioned into blocks with size of 2N x nD 530. Here, in the PART_nLx2N
mode and the PART_nRx2N mode, a PU may have a size of 1/2N x 2N 501 and 512 or
size of 3/2N x 2N 502 and 511. In the PART_2NxnU mode and the PART_2NxnD, 5 a
PU may have a size of 2N x 1/2N 521 and 532 or size of 2N x 3/2N 522 and 531.
[113] As described in FIG. 4, according to a embodiment of the present invention,
a scanning method may be determined based on a partition mode, that is, a size of a
partitioned block in AMP. That is, candidate scanning methods may be obtained in view
10 of partitioned shapes of AMP, among which a scanning method may be determined based
on RDO.
[114] For instance, for a vertically oriented block (a block that its height is
longer than its width) in the 1/2N x 2N mode, such as a left block 501 of nL x 2N block
500 and a right block 512 of nR x 2N block 510, horizontal scanning considering
15 particular component of texture or directionality that the vertically oriented block may
have (for example, vertical component of texture and transform coefficients distributed in
the horizontal direction), or zigzag scanning (or upright diagonal scanning) considering
that the vertically oriented block does not have particular component of texture or
directionality, may be derived as candidate scanning methods. Here, a scanning method
20 having minimum RDO may be selected among the two candidate scanning methods.
[115] Alternatively, for a horizontally oriented block (a block that its width is
longer than its height) in the 2N x 1/2N mode, such as an upper block 521 of 2N x nU
block 520 and a lower block 532 of 2N x nD block 530, vertical scanning considering
particular component of texture and directionality that the horizontally oriented block may
25 have (for example, horizontal texture or transform coefficients distributed in the vertical
27
direction), or zigzag scanning (or upright diagonal scanning) considering that the
horizontally oriented block does not have particular component of texture or directionality,
may be derived as candidate scanning methods. Here, a scanning method having
minimum RDO may be selected among the two candidate scanning methods.
[116] Meanwhile, zigzag scanning (or upright diagonal scanning) may be used fo5 r
larger partitioned portions of nL x 2N 500, nR x 2N 510, 2N x nU 520 and 2N x nD 530
(i.e. 3/2N x 2N and 2N x 3/2N modes). That is, zigzag scanning (or upright diagonal
scanning) may be used for a right partition 502 of nL x 2N block 500, a left partition 512
of nR x 2N block 510, a lower partition 522 of 2N x nU block 520 and an upper partition
10 531 of 2N x nD block 530.
[117] When a scanning method is determined based on an AMP mode as
described above, information on the determined scanning method may be encoded. For
example, as described above in FIG. 4, in the PART_nLx2N mode and PART_nRx2N
mode and for the vertically oriented blocks 501 and 512 (i.e. the 1/2N x 2N mode),
15 isZigZagScanFlag may be set to 1 if zigzag scanning is used, and isZigZagScanFlag may
be set to 0 if horizontal scanning is used . In the PART_2NxnU and PART_2NxnD
modes and for the horizontally oriented blocks 521 and 532 (i.e. the 2N x 1/2N mode),
isZigZagScanFlag may be set to 1 if zigzag scanning is used and isZigZagScanFlag may
be set to 0 if vertical scanning is used. Such flag information may be encoded and
20 transmitted to the decoding apparatus.
[118] FIG. 6 illustrates a method of determining a scanning method and encoding
the information thereon in short distance intra prediction (SDIP) according to an
exemplary embodiment of the present invention.
[119] SDIP refers to a method of partitioning a CU into 2N x 2N PU, N x N PU,
25 1/2N x 2N PU or 2N x 1/2N PU and performing intra prediction on the partitioned PU.
28
When SDIP is performed, a distance between a reference pixel for intra prediction and a
prediction target pixel may be reduced as compared with in conventional intra prediction
performed using a square-shaped PU. Thus a residual value that is a differential value
between an original pixel and the prediction target pixel (predicted pixel) decreases,
resulting in an increase in encoding efficiency5 .
[120] Referring to FIG. 6, one CU may be partitioned into PUs with different
sizes depending on features of a picture. For example, a 32 x 32 CU may be partitioned
into four 16 x 16 PUs 610, 620, 630 and 640. A 16 x 16 PU 610 may be additionally
partitioned into four 4 x 16 PUs 611, 612, 613 and 614, among which a 4 x 16 PU 611
10 may be further partitioned into four 1 x 16 PUs 611-1, 611-2, 611-3 and 611-4.
[121] Likewise, a 16 x 16 PU 630 may be additionally partitioned into four 8 x 8
PUs. A 8 x 8 PU 631 may be further partitioned into four 2 x 8 PUs 631-1, 631-2, 631-3
and 631-4. Also, a 8 x 8 PU 632 may be further partitioned into four 4 x 4 PUs, among
which a 4 x 4 PU 632-1 may be further partitioned into four 1 x 4 PUs.
15 [122] As described above with reference to FIGS. 4 and 5, a scanning method is
determined based on a partition mode of a PU in SDIP, that is, a size of the PU in the
present embodiment. That is, candidate scanning methods are obtained in view of
partitioned shapes of PUs, among which a scanning method is determined based on RDO.
[123] For instance, when the partition mode of the PU in SDIP is a 1/2N x 2N
20 mode which has a vertically oriented partition shape such as the 4 x 16 PUs 611, 612, 613
and 614, the 2 x 8 PUs 631-1, 631-2, 631-3 and 631-4 and the 1 x 4 PU, horizontal
scanning and zigzag scanning (or upright diagonal scanning) may be derived as candidate
scanning methods in view of particular component of texture or directionality (for
example, vertical texture and transform coefficients distributed in the horizontal direction).
25 Here, a scanning method having minimum RDO may be selected among the two candidate
29
scanning methods.
[124] Alternatively, when the partition mode of the PU in SDIP is a 2N x 1/2N
mode which has a horizontally oriented partition shape such as 16 x 4 PU, 8 x 2 PU and 4
x 1 PU, vertical scanning and zigzag scanning (or upright diagonal scanning) may be
derived as candidate scanning methods in view of particular component of texture o5 r
directionality (for example, horizontal texture and transform coefficients distributed in the
vertical direction). Here, a scanning method having minimum RDO may be selected
among the two candidate scanning methods.
[125] Information on the determined scanning method may be encoded using a
10 flag, for example, isZigZagScanFlag, and be transmitted to the decoding apparatus, as
described above in FIGS. 4 and 5.
[126] FIG. 7 is a flowchart illustrating a video encoding method according to the
present invention. Each step of FIG. 7 may be performed by corresponding modules of
the video encoding apparatus of FIG. 1.
15 [127] Referring to FIG. 7, a CU of a current picture is input to the encoding
apparatus (S700). When the input CU is a CU of an inter prediction mode, the CU of
the inter prediction mode ("inter CU") may include a plurality of PUs of the inter
prediction mode ("inter PU") and have one of two prediction modes PreMode, a skip
mode ("MODE_SKIP") and an inter mode ("MODE_INTER").
20 [128] A CU in MODE_SKIP is not partitioned into smaller PUs any longer and
allocated motion information on a PU with a partition mode PartMode of PART_2Nx2N.
[129] A CU in MODE_INTER may be partitioned into four types of PUs, in
which information indicating that a prediction mode is MODE_INTER
(PredMode==MODE_INTER) and information indicating which is a partition mode
25 among PART_2Nx2N, PART_2NxN, PART_Nx2N and PART_NxN (i.e. information
30
such as PartMode==PART_2Nx2N, PartMode==PART_2NxN,
PartMode==PART_Nx2N, or PartMode==PART_NxN) may be transmitted to the
decoding apparatus through a syntax in a CU level.
[130] The encoding apparatus performs motion prediction for a current inter PU
(S710). When the CU is partitioned into a plurality of PUs, a PU to be currently encode5 d
("current PU") is input. The encoding apparatus may perform motion prediction for the
current PU using previous frame, subsequent frame or previous and subsequent frames of
a current frame. Motion information on the current PU, such as a motion vector, a
reference picture index and a prediction direction index, may be obtained through motion
10 prediction.
[131] The encoding apparatus may derive a motion prediction value of the current
PU in the inter prediction mode (S720). The motion information on the current PU is not
transmitted to the decoding apparatus as it is but differential values from predicated values
obtained from temporally and spatially neighboring blocks are transmitted to the decoding
15 apparatus so as to enhance compression efficiency. Motion prediction method may
include a merge mode and an AMVP mode, which may be used to derive a motion
prediction value.
[132] In the merge mode, merging candidates are obtained from motion
information on blocks temporally and spatially neighboring to the current PU. When a
20 candidate having the same motion information as the current PU is present among the
candidates, the encoding apparatus may transmit a flag Merge_Flag indicating that the
merge mode is used and an index of the candidate having the same motion information as
the current PU to the decoding apparatus. In detail, the encoding apparatus derives an
available temporal motion vector predictor (MVP) value using a reference picture index
25 refIdxLX indicating a reference picture obtained in motion prediction and makes a
31
merging candidate list MergeCandList. When a candidate having the same motion
information as the current PU is present on the merging candidate list, the encoding
apparatus sets Merge_Flag to 1 and encodes an index Merge_Idx of the candidate.
[133] In the AMVP mode, the encoding apparatus derives AMVP candidates from
motion information on blocks temporally and spatially neighboring to the current PU5 .
That is, the encoding apparatus derives a motion vector predictor value mvpLX of a luma
component. In detail, the encoding apparatus derives spatial motion vector candidates
(MVPs) from neighboring PUs to the current PU. The encoding apparatus derives a
temporal motion vector candidate of a collocated block using a reference picture index
10 RefIdxLX obtained in motion prediction. The encoding apparatus makes an MVP list
mvpListLX based on the spatial motion vector candidates and the temporal motion vector
candidate. When a plurality of motion vectors has the same value on the MVP list, the
encoding apparatus removes motion vectors other than a motion vector having a highest
priority from the MVP list. Here, motion vectors may have priories in order of motion
15 vectors (mvLXA) of left neighboring blocks to the current PU, motion vectors (mvLXB)
of upper neighboring blocks to the current PU and a motion vector (mvLXCol) of a
temporal collocated block, which are available. A motion vector of a best predictor
among the motion vector candidates on the MVP list is selected as a motion vector
predictive value mvpLX. The best predictor is a candidate block minimizing a rate20
distortion (RD) cost function, for example, JMotSAD considering bit cost and sum of
absolute difference (SAD).
[134] The encoding apparatus encodes the motion information on the current PU
(S730). When the merge mode is used for motion prediction of the current PU, if a
candidate having the same motion information as the current PU is present among
25 merging candidates, the encoding apparatus indicates that the merge mode is applied to
32
the current PU, and encodes and transmits a flag Merge_Flag indicating that the merge
mode is used and an index Merge_Idx of the candidate having the same motion
information as the current PU to the decoding apparatus.
[135] When the AMVP mode is used for motion prediction of the current PU, the
encoding apparatus determines a candidate minimizing a cost function among AMV5 P
candidates by comparing motion vector information on the AMVP candidates with motion
vector information on the current PU. The encoding apparatus performs motion
compensation using the candidate minimizing the cost function and a differential value
between the motion information on candidate minimizing the cost function and the motion
10 information on current PU, thereby obtaining a residual signal. That is, the encoding
apparatus may entropy-encode a motion vector difference between a motion vector of the
current PU and the motion vector of the best predictor.
[136] The encoding apparatus obtains the residual signal by deriving a difference
by pixels between a pixel value of the current block and a pixel value of the prediction
15 block through motion compensation (S740) and transforms the residual signal (S750).
[137] The residual signal is encoded via transformation, in which a transcoding
kernel may be used for transformation. A transcoding kernel may have a shape of 2x2,
4x4, 8x8, 16x16, 32x32 or 64x64, among which a kernel to be used for transformation
may be determined in advance. Here, transform coefficients are generated by
20 transformation and form a 2D block. For example, transform coefficients C for an n x n
block may be derived by Equation 1.
[138] [Equation 1]
[139] Here, C(n, n) is an n x n matrix of transform coefficients, T(n, n) is an n x n
33
transformation kernel matrix, and B(n, n) is an n x n matrix of a residual block.
[140] The transform coefficients calculated by Equation 1 are quantized.
[141] The encoding apparatus determines based on RDO which to transmit among
the residual signal and the transform coefficients (S760). When prediction is properly
done, the residual signal may be transmitted as it is, without transcoding. Here, th5 e
encoding apparatus may compare cost functions before/after transcoding and may select a
method having minimum costs.
[142] The encoding apparatus may transmit a type of a signal to transmit (e.g. the
residual signal or the transform coefficients), with respect to the current block and
10 transmit the signal to the decoding apparatus. For example, if transmitting the residual
signal as it is without transcoding involves minimum cost, the encoding apparatus may
signal the residual signal with respect to the current block. If transmitting the transform
coefficients involves minimum cost, the encoding apparatus may signal the transform
coefficients with respect to the current block.
15 [143] The encoding apparatus scans the transform coefficients (S770). The
encoding apparatus changes quantized transform coefficients of a 2D block form into
transform coefficients of a 1D vector form by scanning. Here, one of horizontal
scanning, vertical scanning and zigzag scanning (or upright diagonal scanning) may be
selected based on a size of a PU, that is, a partition mode of the PU, to scan the transform
20 coefficients.
[144] In detail, candidate scanning modes (methods) may be derived based on
partition shapes of PUs, among which a scanning mode is determined based on RDO. If
the partition mode of the PU has a vertically oriented partition shape, horizontal scanning
and zigzag scanning (or upright diagonal scanning) are derived as candidate scanning
25 modes. If the partition mode of the PU has a horizontally oriented partition shape,
34
vertical scanning and zigzag scanning (or upright diagonal scanning) are derived as
candidate scanning modes. Then, a scanning mode having minimum RDO is selected
among the candidate scanning modes.
[145] Here, as described above in FIGS. 4 and 6, such scanning modes may be
applied to the partition modes of the PU in inter prediction, for example, the N x 2N block5 ,
2N x N block, 2N x 2N block, N x N block, 2N x nU block, 2N x nD block, nL x 2N
block and nR x 2N block modes, and the partition modes of the PU in an intra prediction
mode (e.g. short distance intra prediction: SDIP), for example, the 1/2N x 2N block, 2N x
1/2N block, N x N block and 2N x 2N block modes. As for this, descriptions thereof are
10 omitted herein since already fully described before.
[146] The encoding apparatus may entropy-encode the information to be
transmitted (S780). That is, the encoding apparatus may entropy-encode the scanned
transform coefficients and information on the prediction mode. The encoded information
may form a compressed bitstream and be stored or transmitted in a network abstraction
15 layer (NAL).
[147] FIG. 8 is a flowchart illustrating a video decoding method according to the
present invention. Each step of FIG. 8 may be performed by corresponding modules of
the video decoding apparatus of FIG. 2.
[148] Referring to FIG. 8, the decoding apparatus may entropy-decode a received
20 bitstream (S800). The decoding apparatus may identify a block type from a variable
length coding (VLC) table to recognize a prediction mode of a current block. Further,
the decoding apparatus may identify information on whether transmitted information on
the current block is a residual signal or transform coefficients. Depending on a result, the
decoding apparatus may obtain the residual signal or transform coefficients for the current
25 block.
35
[149] The decoding apparatus may determine a scanning method (S810). That is,
the decoding apparatus determines a scanning method based on a partition mode of a PU
using information signaled from the encoding apparatus. The signaled information may
be a flag indicating whether zigzag scanning is used (for example, isZigZagScanFlag).
[150] Specifically, when the partition mode of the PU has a vertically oriente5 d
partitioned shape or horizontally oriented partitioned shape, the decoding apparatus
decodes a flag indicating whether zigzag scanning is used and determines a scanning
method based on a value of the decoded flag. When the partition mode of the PU has the
vertically oriented partition shape, the decoding apparatus selects one of zigzag scanning
10 and horizontal scanning based on the value of the decoded flag. When the partition
mode of the PU has the horizontally oriented partition shape, the decoding apparatus
selects one of zigzag scanning and vertical scanning based on the value of the decoded
flag. For instance, zigzag scanning is used when isZigZagScanFlag is 1, while horizontal
scanning (for the partition mode having the vertically oriented partition shape)/vertical
15 scanning (for the partition mode having the horizontally oriented partition shape) may be
used when isZigZagScanFlag is 0.
[151] Here, as described above in FIGS. 4 and 6, such scanning methods may be
applied to the partition modes of the PU in inter prediction, for example, the N x 2N block,
2N x N block, 2N x 2N block, N x N block, 2N x nU block, 2N x nD block, nL x 2N
20 block and nR x 2N block modes, and the partition modes of the PU in intra prediction (e.g.
short distance intra prediction: SDIP), for example, the 1/2N x 2N block, 2N x 1/2N block,
N x N block and 2N x 2N block modes. As for this, descriptions thereof are omitted herein
since already fully described before.
[152] Meanwhile, zigzag scanning may be used for a partition mode having a
25 square shape, such as 2N x 2N block and N x N block modes, or for a right partition of nL
36
x 2N block mode, a left partition of nR x 2N block mode, a lower partition of 2N x nU
block mode or an upper partition of 2N x nD block mode as larger partitioned portions in
AMP.
[153] The decoding apparatus may inverse-scan the entropy decoded residual
signal or transform coefficients (S820). The decoding apparatus may generate a residua5 l
block by inverse-scanning for the case of the residual signal and may generate a 2D
transform block by inverse-scanning for the case of the transform coefficients. When the
transform block is generated, the decoding apparatus may dequantize and inversetransform
the transform block, thereby obtaining a residual block. A process of
10 obtaining a residual block by inverse-transforming a transform block is expressed by
Equation 2.
[154] [Equation 2]
[155] Here, B(n, n) is an n x n matrix of a residual block, T(n, n) is an n x n
15 transformation kernel matrix, and C(n, n) is an n x n matrix of transform coefficients.
[156] The decoding apparatus may perform inter prediction (S830). The
decoding apparatus may decode information on the prediction mode and perform inter
prediction according to the prediction mode.
[157] For example, when the prediction mode PredMode is the merge mode (for
20 example, PredMode== MODE_SKIP && Merge_Flag==1), the decoding apparatus may
derive a motion vector mvLX of a luma component and a reference picture index
refIdxLX for the merge mode. To this end, the decoding apparatus may derive merging
candidates from partitions of PUs (i.e. prediction blocks) spatially neighboring to the
current PU. The decoding apparatus may derive the reference picture index refIdxLX so
37
as to obtain a temporal merging candidate for the current PU. The decoding apparatus
may derive an available temporal motion vector predictor value (MVP) using the derived
reference picture index. When a number NumMergeCand of candidates on a merging
candidate list MergeCandList made based on the spatial merging candidates and the
temporal merging candidate is 1, the decoding apparatus sets a merging candidate inde5 x
(Merge_Idx) to1. Otherwise, the decoding apparatus may set the merging candidate
index to a received merge index. The decoding apparatus derives a motion vector
(mvLX) of a merging candidate indicated by the received merge index and the reference
picture index (refIdxLX). The decoding apparatus may use the derived motion vector
10 and the derived reference picture index for motion compensation.
[158] When the prediction mode PredMode is the AMVP mode, the decoding
apparatus may derive a reference picture index (refIdxLX) for the current PU and may
derive a motion vector predictor value (mvpLX) of a luma component using the reference
picture index. In detail, the decoding apparatus may derive spatial motion vector
15 candidates (MVPs) from neighboring PUs to the current PU and may derive a temporal
motion vector candidate (MVP) of a collocated block indicated by the reference picture
index. The decoding apparatus may generate an MVP list mvpListLX based on the
derived spatial motion vector candidates and the derived temporal motion vector candidate.
When a plurality of motion vectors has the same value on the MVP list, the decoding
20 apparatus may remove motion vectors other than a motion vector having a highest priority
from the MVP list. Here, as described above, motion vectors have priories in order of
motion vector (mvLXA) of left neighboring block to the current PU, motion vector
(mvLXB) of upper neighboring block to the current PU and a motion vector (mvLXCol)
of a temporal collocated block, which are available. When a number
25 NumMVPCand(LX) of MVP candidates on the MVP list is 1, the decoding apparatus may
38
set an MPV candidate index mvpIdx to 0. When the number of MVP candidates is 2 or
more, the decoding apparatus may set the MPV candidate index mvpIdx equal to a
received index value. The decoding apparatus may determine a motion vector indicated
by mvpIdx among the MVP candidates on the MVP list mvpListLX as a motion vector
predictor value mvpLX. The decoding apparatus may derive a motion vector mvL5 X
using a motion vector predictor value mvpLX and Equation 3.
[159] [Equation 3]
[160] Here, mvLX[0], mvdLX[0] and mvpLX[0] are x components of an LX
10 motion vector information (i.e. x components of mvLX, mvdLX and mvpLX), and
mvLX[1], mvdLX[1] and mvpLX[1] are y components of the LX motion vector
information (i.e. y components of mvLX, mvdLX and mvpLX).
[161] The decoding apparatus may derive a reconstructed signal (S840). For
instance, the decoding apparatus may add the residual signal to a signal of a previous
15 frame (i.e. predicted signal) to generate the reconstructed signal. The decoding apparatus
may add a prediction signal of the previous frame obtained by motion compensation using
a derived motion vector and the decoded residual signal for the current PU, thereby
generating the reconstructed signal.
[162] Although the methods have been described with a series of stages or blocks
20 based on the flowcharts in the aforementioned embodiments, the present invention is not
limited to the foregoing sequence of the stages. Some stages may be carried out in
different order from described above or at the same time. Also, it will be understood by
those skilled in the art that the stages illustrated in the flowcharts are not exclusive,
39
additional stages may be included in the flowchart, or one or more stages may be deleted
from the flowcharts without affecting the scope of the present invention.
[163] While a few exemplary embodiments have been shown and described with
reference to the accompanying drawings, it will be apparent to those skilled in the art that
various modifications and variations can be made from the foregoing descriptions withou5 t
departing from the essence of the present invention. The exemplary embodiments are
provided not to restrict the concept of the present invention but to illustrate the present
invention and do not limit the scope of the present invention. The scope of the invention
is defined by the appended claims, and all differences within the scope will be construed
10 as being included within the appended claims of the present invention.
40
I/We claim:
1. A video encoding method, the method comprising:
determining a scanning method based on a partition mode of a prediction unit; and
encoding information on the scanning method,
wherein the scanning method is determined based on rate-distortion optimizatio5 n
(RDO) among candidate scanning methods derived in view of a partition shape of the
partition mode.
2. The method of claim 1, wherein the determining of the scanning method
10 derives horizontal scanning and zigzag scanning as the candidate scanning methods when
the partition mode has a vertically oriented partition shape and derives vertical scanning
and zigzag scanning as the candidate scanning methods when the partition mode has a
horizontally oriented partition shape.
15 3. The method of claim 2, wherein the partition mode comprises an N x 2N
mode, a 2N x N mode, a 2N x 2N mode, an N x N mode, a 2N x nU mode, a 2N x nD
mode, an nL x 2N mode and an nR x 2N mode based on a size of the prediction unit for
which inter prediction has been performed.
20 4. The method of claim 3, wherein the partition mode having the vertically
oriented partition shape comprises the N x 2N mode, the nL x 2N mode and the nR x 2N
mode,
wherein the partition mode having the vertically oriented partition shape is a 1/2N
x 2N mode of a left partition with a smaller partition size for the nL x 2N mode and the
25 partition mode having the vertically oriented partition shape is a 1/2N x 2N mode of a
41
right partition with a smaller partition size for the nR x 2N mode.
5. The method of claim 3, wherein the partition mode having the
horizontally oriented partition shape comprises the 2N x N mode, the 2N x nU mode and
the 2N x nD mode5 ,
wherein the partition mode having the horizontally oriented partition shape is a
2N x 1/2N mode of an upper partition with a smaller partition size for the 2N x nU mode
and the partition mode having the horizontally oriented partition shape is es a 2N x 1/2N
mode of a lower partition with a smaller partition size for the 2N x nD mode.
10
6. The method of claim 3, wherein the determining of the scanning method
determines zigzag scanning as the scanning method when the partition mode is the 2N x
2N mode, the N x N mode, a 3/2N x 2N mode of a right partition with a larger partition
size in the nL x 2N mode, a 3/2N x 2N mode of a left partition with a larger partition size
15 in the nR x 2N mode, a 2N x 3/2N mode of a lower partition with a larger partition size in
the 2N x nU mode or a 2N x 3/2N mode of an upper partition with a larger partition size in
the 2N x nD mode.
7. The method of claim 1, wherein the information on the scanning method
20 is indicated using a flag, and the flag indicates whether zigzag scanning is used.
8. A video encoding method, the method comprising:
determining a scanning method based on a partition mode of a prediction unit for
which short distance intra prediction (SDIP) has been performed; and
25 encoding information on the scanning method,
42
wherein the scanning method is determined based on rate-distortion optimization
(RDO) among candidate scanning methods derived in view of a partition shape of the
partition mode.
9. The method of claim 8, wherein the determining of the scanning metho5 d
derives horizontal scanning and zigzag scanning as the candidate scanning methods when
the partition mode has a vertically oriented partition shape and derives vertical scanning
and zigzag scanning as the candidate scanning methods when the partition mode has a
horizontally oriented partition shape.
10
10. The method of claim 9, wherein the partition mode comprises a 1/2N x
2N mode, a 2N x 1/2N mode, an N x N mode and a 2N x 2N mode based on a size of the
prediction unit for which SDIP has been performed.
15 11. The method of claim 10, wherein the partition mode having the vertically
oriented partition shape comprises is the 1/2N x 2N mode, and the partition mode having
the horizontally oriented partition shape comprises the 2N x 1/2N mode.
12. The method of claim 10, wherein the determining of the scanning method
20 determines zigzag scanning as the scanning method when the partition mode is the N x N
mode or the 2N x 2N mode.
13. The method of claim 8, wherein the information on the scanning method
is indicated using a flag, and the flag indicates whether zigzag scanning is used.
25
43
14. A video decoding method, the method comprising:
determining a scanning method based on a partition mode of a prediction unit; and
inverse-scanning a transform coefficient according to the scanning method,
wherein the scanning method is determined based on the partition mode using
information signaled from an encoding apparatus, an5 d
wherein the signaled information is a flag indicates whether zigzag scanning is
used.
15. The method of claim 14, wherein the determining of the scanning method
10 decodes the flag indicating whether zigzag scanning is used and determines the scanning
method based on a value of the decoded flag when the partition mode has a vertically
oriented partition shape or horizontally oriented partition shape,
wherein one of zigzag scanning and horizontal scanning is selected based on the
value of the decoded flag when the partition mode has the vertically oriented partition
15 shape, and
wherein one of zigzag scanning and vertical scanning is selected based on the value
of the decoded flag when the partition mode has the horizontally oriented partition shape.
16. The method of claim 15, wherein the partition mode comprises an N x 2N
20 mode, a 2N x N mode, a 2N x 2N mode, an N x N mode, a 2N x nU mode, a 2N x nD
mode, an nL x 2N mode and an nR x 2N mode based on a size of the prediction unit for
which inter prediction is performed.
17. The method of claim 16, wherein the partition mode having the vertically
25 oriented partition shape comprises the N x 2N mode, the nL x 2N mode and the nR x 2N
44
mode, wherein the partition mode having the vertically oriented partition shape is a 1/2N x
2N mode of a left partition with a smaller partition size for the nL x 2N mode and the
partition mode having the vertically oriented partition shape is a 1/2N x 2N mode of a
right partition with a smaller partition size for the nR x 2N mode, and
wherein the partition mode having the horizontally oriented partition shap5 e
comprises the 2N x N mode, the 2N x nU mode and the 2N x nD mode, wherein s the
partition mode having the horizontally oriented partition shape is a 2N x 1/2N mode of an
upper partition with a smaller partition size for the 2N x nU mode and the partition mode
having the horizontally oriented partition shape is a 2N x 1/2N mode of a lower partition
10 with a smaller partition size for the 2N x nD mode.
18. The method of claim 16, wherein the determining of the scanning method
determines zigzag scanning as the scanning method when the partition mode is the 2N x
2N mode, the N x N mode, a 3/2N x 2N mode of a right partition with a larger partition
15 size in the nL x 2N mode, a 3/2N x 2N mode of a left partition with a larger partition size
in the nR x 2N mode, a 2N x 3/2N mode of a lower partition with a larger partition size in
the 2N x nU mode or a 2N x 3/2N mode of an upper partition with a larger partition size in
the 2N x nD mode.
20 19. The method of claim 15, wherein the partition mode comprises a 1/2N x
2N mode, a 2N x 1/2N mode, an N x N mode and a 2N x 2N mode based on a size of the
prediction unit in short distance intra prediction (SDIP), the partition mode having the
vertically oriented partition shape comprising is the 1/2N x 2N mode and the partition
mode having the horizontally oriented partition shape comprising the 2N x 1/2N mode.
25
45
20. The method of claim 19, wherein the determining of the scanning method
determines zigzag scanning as the scanning method when the partition mode is the N x N
mode or the 2N x 2N mode.