17
`
`EP 2 600 613 A2
`
`18
`
`block having the same size of the transform unit accard-
`ing to the scan pattem.
`{0103} The orediction mode decoder 230 reconstructs
`the intra prediction mode of the current prediction mode
`on the basis of the intra prediction information and the
`size information on the current prediction unit received
`fromthe entropy decoder 210. The received intra predic-
`tion information is restored through an inverse process
`of the process shown in FIG. 4.
`{0408] The reference pixel generator 240 generates
`unavailable reference pixels of the current prediction unit,
`and adaptively filters the reference pixels according to
`the intra prediction mode of the current prediction unit
`received from the prediction mode decoder 236. A meth-
`od of generating the reference pixels and a method of
`fillering the reference pixels are the same as those of the
`reference pixel generator 142 in the intra predictor 146
`of FIG. 2.
`
`Specifically, itis determined whether reference
`{6105]
`pixels of the current prediction unit are available. The
`reference pixels of the current prediction unit used for
`intra prediction consist of a corner reference pixel posi-
`tioned at (x =-1, y =-1), 2L upper reference pixels posi-
`tioned at (x =0,--, and 2L-1, y =-1), and 2M leftreference
`pixels positioned at (x = 0, y=6,--, and 2M-1). Here, L
`is a width of the current orediction unit, and M is a height
`of the current prediction unit.
`{6106] When the reference pixels for generating a pre-
`diction block are unavailable or insufficient, reference pix-
`els are generated.
`{@107] When ail the reference pixels are unavailable,
`reference pixels are generated with a predetermined val-
`ue.
`
`{61G8] When sameofthe reference pixels are unavail
`able, itis determined whether the unavailable reference
`pixels exist in only one direction from available pixels or
`betweentheavailable pixels.
`[6109] When the unavailable reference pixels exist in
`only one direction from the available pixels, reference
`pixels are generated by copying the value of an available
`pixel closest to the unavailable pixel. For example, when
`the currant prediction unit
`is positioned at an upper
`boundary of a picture or a slice, the corner referencepixel
`and the upper referencepixels are unavailable. Thus, in
`this case, the corner reference pixel and the upper ref-
`erence pixels can be generated by copying a reference
`pixel positioned at (x =-1, y = 0). Alternatively, reference
`pixels may be generated using an available reference
`pixel of the closest position and one or more available
`reference pixels. Forexample, when the corner reference
`pixel having a position (x = -1, y = -1) and referencepixels
`at positions (x = 0, ---, and L-1, y =-4) are available and
`reference pixels at positions (x =L,--, 2L-1, v = -1) are
`unavailable, reference pixels at the unavailable positions
`can be generated using change in a difference between
`a reference pixel at a position (x = L-1, y = -1) and a
`comer reference pixel value or another refererice pixel
`value.
`
`20
`
`Nhon
`
`40
`
`oor
`
`6116] Whenthe unavailable reference pixels exist be-
`tween the available pixels, reference pixels are generat-
`ed using two available pixels 9 and gq adiacent to the
`unavailable reference pixels. For example, when the cor-
`ner reference pixel and the L upper reference pixels po-
`sitioned at (x = 0, --, and L-1, y = -4) are unavailable,
`reference pixels existing between the reference pixel p
`at a position (x = -1, y = G) and the reference pixel q ata
`position (x =1L, y =-1} can be generated using the refer-
`ence pixels p and q.
`fO144] The generated reference pixel values may be
`obtained by rounding off an average of the reference pixel
`p and the reference pixel q. Also, the reference pixel val-
`ues may be generated using change in difference be-
`tween pixels values of the reference pixel po and the ref-
`erence pixel a.
`in this case, the refererice pixel values
`may be generated by linear interpolation according to
`positions corresponding to the generated pixel values or
`using a weighted average of the two reference pixels.
`fa442] Meanwhile, when a plurality of prediction units
`are on an upper side of the current prediction uni, there
`is a high possibility that a difference between boundary
`pixels present on both sides of a boundary between two
`of the upper prediction units will be higherthan a differ-
`ence between adjacent pixels in each upper prediction
`unit. This results from an error caused by a quantization
`parameter, itis highly probable that the error will occur
`in directional intra prediction modes in which the predic-
`tion block is generated using two adjacent reference pix-
`els.
`
`in particular, modes (mode numbers 3, 6 and
`{0143}
`3) having a direction of 45° with reference to a horizontal
`or vertical direction of FiG. 3 are rnost seriously affected.
`In the vertical and horizontal
`intra prediction modes
`{mode numbers 0 and 1), one pixel is used to generate
`the prediction block, and thus the vertical and horizontal
`intra prediction modes areslightly affected.
`fO144]
`For this reason, afilter (smoothing filter) is ap-
`plied to the reference pixels in the directional intra pre-
`diction modes 3, 6 and 9, and is not applied to the refer-
`ence pixels in the vertical and horizontal intra oredictian
`modes. In the DC mode out of non-directional modes,
`the filter is not applied either. For these modes, whether
`4 not to apply the filler may be determined regardless
`of a size of the current prediction unit.
`{0448}
`in directional intra prediction modes existing be-
`tween the intra prediction mode 3, § or 9 and the hori-
`zontal or vertical intra prediction made, thefilter (smooth-
`ing filter} can be adaptively applied to reference pixels.
`it is preferable to increase a probability of applying the
`filter (smoothingfiller) as the direction of the directional
`intra prediction mode is relatively closer to the direction
`of the intra prediction mode having the direction of 45°.
`Specifically, when a first directional intra prediction mode
`is closer in directivity to the intra prediction mode having
`the direction of 45° than a second directional intra pre-
`diction mode, if a filteris applied to the second directional
`intra prediction made, the filter is also applied to thefirst
`
`IPR2021-00827
`Unified EX1002 Page 925
`
`IPR2021-00827
`Unified EX1002 Page 925
`
`

`

`13
`
`EP 2 600 613 A2
`
`20
`
`directional intra prediction mode. On the other hand, if
`the filter is apolied to the first directional intra prediction
`mode, the filter may or may not be applied to the second
`directional intra prediction made.
`{6176} There is a high possibility that a change in dif-
`ference between pixels in a prediction unit of a large size
`will be less than a change in difference between pixels
`in a prediction unit of a small size. Thus, the number of
`directional modes in which the filler is applied may be
`increase orthe strongerfiller may be applied asihe size
`of the prediction unit increases. On the other hand, when
`a prediction unit becomes smaller than a specific size
`the filter may not be applied.
`{8147]
`For one example, in the intra prediction mode
`3, 8 or 9 having the direction of 45°, a first filler may be
`applied to a prediction unit having a size equal fo or small-
`er than a first size, and a secondfilter that is stronger
`than the first filter may be applied to a prediction unit
`having a size larger than the first size. The first size may
`vary according to directional prediction modes.
`{6148]
`For another example,
`in the intra prediction
`mode 5 existing between the vertical
`intra prediction
`modeand the intra prediction mode6 having the direction
`of 45°, nofitter may beapplied to a prediction unit having
`a size equal to or smaller than a second size, the first
`filter may be apolied to a prediction unit having a size
`jarger than the second size and equal to or smaller than
`a third size, and the second filter may be applied to a
`prediction unit having a size larger than the third size.
`The second size and the third size may vary according
`to diractional orediction modas.
`{6478} The first fiter may be a 3-tap filter [71, 2, T]ora
`5-tap filter [1, 2, 4, 2, 1]. The second fier has a greater
`smoothing effect than the first filter.
`{6120] The prediction block generator 250 generates
`a prediction block according to the intra prediction mode
`of the current orediction unit received from the prediction
`mode decoder 230. A method of generating the predic-
`tion block is the same as that of the prediction block gen-
`erator 142 in the intra predictor 140 of FIG. 2.
`{6124} Thatis, in the directional intra preciction made,
`the corresponding reference pixels vary according to the
`intra prediction modes. For example, in the vertical mode,
`L upper reference pixels positioned at (x =0,--, and L-1,
`=-14) ara used, and in the horizontal mode, L left refer-
`ence pixels positioned at (& =-1, y= 0,
`--, and L-1) are
`used,
`
`In non-directional intra orediction modes, the
`{81221
`corner pixel, L upper reference pixels positioned at (x =
`0,--,andL-1,y=-1) and | leftreference pixels positioned
`atQ=-1,y =0,--, and L-1) are used. The non-directionat
`intra prediction modes are the DC mode and the planar
`mode.
`
`20
`
`Nhon
`
`40
`
`In the planar mode, a reference pixel of a pre-
`{6123}
`diction block is generated using the cornerreference pix-
`al, a left reference pixel and an upper reference pixel.
`When a reference pixel to be generated is positioned at
`(a, 6}, the reference pixel of the prediction block X{a, b)
`
`oor
`
`is generated using the corner reference pixel C(x =-1, y
`=~ 1}, an upper referencepixel T(x = a, y =-1) and a left
`reference pixel L(x =-1, y = b). Specifically, X(a, b) may
`be L(x =-1,y =b)+Ti = a, y = -4)-O(x = -4, y = -4}.
`{0124}
`in the intra prediction mode existing to the right
`side of the vertical mode (mode number 0) of FIG. 3,
`there is a possibility that differences betweenpixels of a
`lower left region of the generated prediction block and
`pixels of the original prediction unit will increase if the
`prediction block is generated using only upper reference
`pixels. However, when a prediction block is generated
`using upper reference pixels and left reference pixels for
`several modes among the modes, the difference can be
`reduced. This effect is greatest in the intra prediction
`mode 6. Also in intra prediction modes existing underthe
`horizontal mode (mode nurnber 1) of FIG. 3, same meth-
`od may be applied, and the effect is greatest in the intra
`prediction mode 9.
`{0125]
`Thus, in the prediction mode 6 or 9, a prediction
`pixel may be generated using corresponding (for exam-
`ple, positioned at 45° of the prediction pixel} one upper
`interpolation reference pixel and oneleftinterpolation ret-
`erence oixel. The prediction pixel may be generated by
`linear interpolation of the one upperinterpolation refer-
`ence pixel and the oneleft interpolation reference pixel
`
`or using a rounded off average. Likewise, ina eee
`
`mined number of intra prediction modes adjacent to
`mode 6 or 9, a prediction block may be generated using
`the left reference pixels and the upper reference pixels.
`in this case, to reduce complexity, the aforernentioned
`method may not be applied in intra prediction modes hav-
`ing mode numbers greater than a predetermined mode
`number (for example, 9 or 17). Also, the method may be
`applied to only a current prediction unit having a size
`egual to or larger than a predetermined size. The prede-
`termined size is 8X8 or 16x16
`
`[0126] The prediction block post-processor 260 adap-
`tivelyfilters the prediction block generated by the predic-
`tion block generator 250 according to the intra prediction
`rnode of the current orediction unit received frorn the pre-
`diction mode dacoder 230. The prediction block
`post-processor may be integrated into the prediction
`block generator 250. A prediction block filtering method
`is the same as that of the prediction block post-processor
`144 of the intra predictor 140 of FIG. 2.
`{0127] Thatis, to reduce differencesin pixel values be-
`tween a reference pixel and pixels in the prediction block
`adjacent to the referencepixel, the prediction blockfilter
`260 adaptively filters some or ail pixels in the prediction
`block adjacent to the reference pixel according to the
`intra prediction mode. The pixels adjacent to the refer-
`ence pixel exist in ihe prediction block.
`{0128]
`inthe planar mode, pixels in the prediction block
`adjacent to a reference pixel are generated using the
`reference pixel, and thus no filler is applied.
`f6129]
`inthe DC mode, an average of reference pixels
`is used to generate prediction pixel, and thus a filter is
`applied. Different type of filler can be used according to
`
`IPR2021-00827
`Unified EX1002 Page 926
`
`IPR2021-00827
`Unified EX1002 Page 926
`
`

`

`21
`
`EP 2 600 613 A2
`
`22
`
`the size of the prediction unit (the size of the prediction
`block}. In a prediction unit of a large size, a fitter that is
`the same as used in a prediction unit of a small size or
`a strong filter having a great smoothing effect can be
`used.
`
`in intra prediction modes existing
`{8130] Meanwhile,
`ic the right side of the vertical rrode (mode number 0} of
`FIG. 3,
`if the prediction block is generated using only
`upper reference pixels, there exists a possibility that a
`difference between a pixel in the generated prediction
`block and the corresponding pixel of the original predic-
`tion unit will increase as the position of the pixel in the
`generated prediction block is going down to the undereft
`region. in particular, the difference remarkably increases
`in the intra prediction mode 6.
`{0134}
`Likewise, in intra prediction modes existing un-
`der the horizontal mode (mode number 1) of FIG. 3, ifa
`prediction block is generated using only left reference
`pixels, there exists a possibility that a difference beiween
`a pixel in the generated prediction block and the corre-
`sponding pixel of the original prediction unit will increase
`as the position of the pixel in the generated prediction
`block is going to the upper-right region. The difference
`remarkably increasesin the intra prediction mode 9.
`{6132]
`In the vertical mode (mode number0), a differ-
`ence between a pixel in the prediction block and the cor-
`responding pixel of the original prediction unit increases
`as the positionof the pixel is going down. In the horizontal
`mode (mode number 1), a difference between a pixel in
`the prediction block and the corresponding pixel of the
`original prediction unit increases as the position of the
`pixel is going right.
`{8133} Therefore, to reduce the difference, sorne pre-
`diction pixels in the prediction block can be adaptively
`fiftered according to the directional intra prediction mode.
`in this case, the some prediction pixels in the prediction
`block arefiltered using referencepixelsin the prediction
`unit not used for generating the prediction block.
`{6134] Aregion to befiltered may be set differently ac-
`cording to a directional intra orediction mode. As a direc-
`tion of intra prediction mode gets closer to made 6 ar
`mode 9, the ragion to be fitered may become larger or
`remain the same.
`
`For one example, when a size of a predict
`{0135}
`unit is 2NX2N in the intra prediction mode6, a filter may
`be applied to only a predetermined number (1~3) ofintra
`prediction modes adjacent to the intra prediction mode
`6 among intra prediction modes existing to the right side
`of the vertical mode (mode number Q). In this case, to
`reduce complexity, no fitter may be applied to theintra
`prediction mode having a mode number greater than a
`predetermined mode number(for axampie, 9 or 17). Also,
`some pixels in the prediction block may be adaptively
`filtered according te a size of the prediction unit. As the
`size of the prediction unit increases, the number of pixels
`to be filtered mayincrease or remain the same.
`{G136]
`For another examole,
`in the intra prediction
`mode 6, the prediction block may not be filtered if the
`
`20
`
`Nhon
`
`40
`
`oor
`
`size of the prediction unit is 44. If the size of the pre-
`diction unitis @x8 or 16X16, only four lowerleft boundary
`pixels of the prediction block among eight pixels posi-
`tioned at (x = 0, y = 0,
`--, and 7) may befiltered. if the
`size of the prediction unit is 32X32 or larger, all eight of
`the boundary pixels may be filtered.
`{137}
`Strength offiter to be applied to the pixels in
`the prediction block may also vary according fo the size
`of the orediction unit. As the size of the prediction unit
`increases, the filter strength may increase or rernain the
`same.
`
`f0138] Theimage reconstructor 270 receives a predic-
`tion block from the prediction block generator 259 or the
`prediction block filter 260 in units of prediction units ac-
`cording to the intra prediction made. The image recon-
`structor 270 receives a residual block reconstructed by
`the residual signal decoder 220in units of transformunits.
`The image reconstructor 270 generates a reconstructed
`image by adding the received prediction block and resid-
`ual block. The image may be reconstructed in units of
`coding units.
`[6139] While the invention has been shown and de-
`scribed with reference to certain exemplary embodi-
`ments thereof, it will be understood by those skilled in
`the art that various changes in form and details may be
`made therein without departing from the spirit and scope
`of the invention as defined by the appendedclaims.
`
`Claims
`
`4. Anintra prediction decoding apparatus, comprising:
`
`an entropy decoder configured ta restore quan-
`tized residual coefficients, intra prediction infor-
`mation and size information on a prediction unit
`fram a received bit stream:
`a prediction mode decoderconfigured to restore
`an intra prediction mode of the current prediction
`unit on the basis of the intra prediction informa-
`tion and the size information on the current ore-
`diction unit received from the entropy decoder;
`a residual signal decoder configured to restore
`aresidual signal accardingtathe intra prediction
`mode received from the prediction mode decod-
`er,
`
`a reference pixel generator configured fo gen-
`erate unavailable referencepixels of the current
`
`prediction unit, and to adaptively filter the refer-
`ence pixels on the basis of the intra prediction
`modeof the current prediction unit receivedfrom
`the prediction mode decoder;
`a prediction block generator configured to gen-
`erate a prediction block using reference pixels
`corresponding to the intra prediction mode re-
`ceived from the prediction mode decoder;
`a prediction block filter configured to adaptively
`filter ihe prediction block generated fromthe pre-
`
`IPR2021-00827
`Unified EX1002 Page 927
`
`IPR2021-00827
`Unified EX1002 Page 927
`
`

`

`23
`
`EP 2 600 613 A2
`
`24
`
`diction block generator according to the intra
`prediction mode received from the prediction
`mode decoder and
`an image reconstructor configured to receive the
`prediction block from the prediction block gen-
`erator of a prediction block fitter in units of pre-
`diction unils according to the intra prediction
`mode received from the prediction mode decod-
`er, and generate a reconstructed image using a
`restored residual block received fromtheresic-
`
`ual signal decoder.
`
`The intra prediction decoding apparatus of claim 14,
`wherein the reference pixel generator adaptivelyfit
`ters the reference pixels according to a size of a pre-
`diction unitin intra prediction modes having direction
`between intra prediction rode 3, 6 or 9 having a
`direction of 45° and the horizontal mode or the ver-
`fical mode.
`
`The intra prediction decoding apparatus of claim 14,
`wherein the reference pixel generator applies no fi-
`ter to reference pixels of a prediction unit smaller
`than a predetermined size.
`
`The intra prediction decoding apparatus of claim 14,
`wherein, when, amonga first directional mode and
`a second directional mode present between a hori-
`zontal mode or a vertical mode and an intra predic-
`tion mode 3, 6 or 9 having direction of 45° with re-
`spect to the horizontal mode or the vertical mode,
`the first directional mode has closer directivity to the
`intra prediction mode having the direction of 45° than
`the second directional mode, if the reference pixel
`generator applies a filter to referance pixels of the
`second directional mode, the reference pixel gener-
`alor also applies the fitter to reference pixels of the
`first directional mode.
`
`The intra prediction decoding apparatus of claim 14,
`wherein, when the intra orediction modeis a olanar
`mode, the prediction block generator generates pre-
`dicted pixels of the prediction block using a corner
`referencepixel, a left reference pixel and an upper
`reference pixel.
`
`The intra prediction decoding apparatus of claim 14,
`wherein, when theinire prediction modeis a vertical
`mode, the reference pixel generator does notfilter
`the reference pixels, and
`the prediction blockfiller uses reference pixels that
`have not been used for generating the prediction
`block to filter some pixels in the prediction block.
`
`The intra prediction decoding apparatus of claim 14,
`wherein, when the intra prediction mode is an intra
`prediction mode having direction of 45° with respect
`to a vertical mode (mode number6) or among a4 pre-
`
`20
`
`Nhon
`
`40
`
`oor
`
`determined number of intra prediction modes having
`directivity close fo the intra prediction made, the pre-
`diction block generaior generates the prediction
`block using upperreference pixels and left reference
`pixels.
`
`IPR2021-00827
`Unified EX1002 Page 928
`
`IPR2021-00827
`Unified EX1002 Page 928
`
`

`

`EP 2 600 613 A2
`
`Gobo
`“Gb
`
`SSYSANI
`
`HAZULNYAD
`
`S5Y3ANI
`
`daaWeOaSNVel
`
`-~1S0d
`
`HOSSAIOdd
`
`
`BUNLOd|nani|43009HANNSHACIAIGSciiAdOHLNGsae
`
`
`
`
`wevewewd——wweernennannnnanpennnnnnnnnnnnnnditnnnnptannennnnnnnnnnnnnd nnnnnnnnnnnnnnnnnnnnpenntteamedrNt
`
`
`
`|Sis
`
`YHLNi
`
`OLodsad
`
`YaLNi
`
`HOLOMSYd
`
`TWNEIS
`
`IPR2021-00827
`Unified EX1002 Page 929
`
`IPR2021-00827
`Unified EX1002 Page 929
`
`
`
`
`
`
`
`

`

`EP 2 600 613 A2
`
`
`
`
` PREDICTION
`
`
`MODE CODER
`
`
`|
`
`
`i PREDICTION MODE
`DETERMINER
`
`ttttt
`
`
`PREDICTION
`
`| POST-PROCESSOR
`1erentrorsnpinianna
`
`
`
`GENERATOR
`
`
`IPR2021-00827
`Unified EX1002 Page 930
`
`|
`
`
`
`
`PREDICTION UNIT
`
`RECEIVER
`
`IPR2021-00827
`Unified EX1002 Page 930
`
`

`

`EP 2 600 613 A2
`
`FIG. 3
`
`ry
`
`
` a Lf]|HI\IW,
`
`ONhs9AonowoeA)TTSHe1Peeem
`OshOMLEEUAYY
`
`IPR2021-00827
`Unified EX1002 Page 931
`
`IPR2021-00827
`Unified EX1002 Page 931
`
`

`

`EP 2 600 613 A2
`
`FIG. 4
`
`
`78120
`
`wo DETERMINE
`
`7WHETHEROR NOT TO
`< CHANGE INTRA PREDICT!ON >
`me
`‘MODE OF
`an
`
`
`
`
`
`
`
`
`BUILD CANDIDATE LIST
`
`
`
`SEARCH CANDIDATES
`
`
`
`
`
`
`
`_- DETERMINES 7 7198
`LWHETHER OR NOT INTRA,
`PREDICTION MODE OF CURRENT,
`reper(ON UNIT IS
`Zoro OBTAIN MODE CHANGE VALUE
`
`_YES
`
`\
`
`Os
`
`INCLUDEDIN LIST?
`
`NO]
`
`DETERMINE CANDIDATE INDEX]
`
`IPR2021-00827
`Unified EX1002 Page 932
`
`IPR2021-00827
`Unified EX1002 Page 932
`
`

`

`EP 2 600 613 A2
`
`096-—
`NOILOMSdd
`
`HasMOO
`
`‘\
`
`!=3Ode
`
`
`
`qdCGOQOACWNDIS
`
`WACHS3e
`
`OMLOGsde
`
`NOLOSed
`
`HEN
`
`HOLVHSNADMOOT]
`
`
`Haqd0030JGOW
`
`1Y3GOOIGAdO
`
`_
`
`_|
`
`
`
`FONSYHSASH
`
`YOLVHANSSTAXI
`
`SQOWNOLLODIGdddVen!
`
`
`
`premeweresirwmnneeiesaenatten+
`
`©—
`NS
`
`IPR2021-00827
`Unified EX1002 Page 933
`
`IPR2021-00827
`Unified EX1002 Page 933
`
`
`
`
`
`
`
`

`

`i des brevets
`
`; Européisches
`Patentamt
`European
`Patent Office
`Office européen
`
`(4 9)
`
`5>
`
`“ =
`
`Nae
`
`
`(14)
`EP 2600 614 A2
`
`EUROPEAN PATENT APPLICATION
`published in accordance with Art. 153(4) EPC
`
`(43) Date of publication:
`05.06.2013 Bulletin 2043/23
`
`(B41)
`
`Int Ch:
`HOAN 7/34 (2008.04)
`
`(21) Application number: 14814798.2
`
`(86) International application number:
`PCTIAR2011/005592
`
`(22) Date offiling: 29.07.2041
`
`(87) International publication number:
`WO 2072/018198 (08.02.2012 Gazette 2012/06}
`
`| {72} Inventor: Oh, Seo-mi
`Seoul 138-908 (KR)
`
`(84) Designated Contracting States:
`AL AT BE BG CH CY CZ DE DK EE ES FIFR GB
`GRHRHUTEISITLILT LULVY MC MK MT NL NO
`PL PT RO RS SE SISK SMTR
`
`(74) Representative: Griinecker, Kinkeldey,
`Stockmalr & Schwanhausser
`
`(30) Priority: 28.08.2017 KR 20110062603
`31.07.2070 KR 20100074462
`
`Lecpoldstrasse 4
`80802 Miinchen (DE}
`
`(71) Applicant: Gh, Soo-rni
`Seoul 138-908 (KR)
`
`(54)
`
`PREDICTION BLOCK GENERATING DEVICE
`
`A prediction block generating device of the
`(57)
`present invention determines additional information for
`generating a@ prediction block included in an additional
`information container received, and an intra prediction
`mode of a current prediction unit using available intra
`prediction mode candidate information of the current pre-
`diction unit, generates referencepixels that are not in a
`unavailable position for generating an intra prediction
`block by using available reference pixels, adaptively fil-
`
`ters the reference pixels close to the current prediction
`init based on the determined intra prediction mode of the
`current prediction unit or size information of the current
`prediction unit, and generates a prediction block of the
`current prediction unit using the reference pixels corre-
`sponding to the determined intra prediction mode of the
`current prediction unit. Therefore, ihe present invention
`can improve image compression ratio by generating the
`prediction block close to an original image.
`
`FIG.
`
`1
`
`
`
`IPR2021-00827
`Unified EX1002 Page 934
`
`P2600614AZ
`
`IPR2021-00827
`Unified EX1002 Page 934
`
`

`

`_s
`
`EP 2 600 614 A2
`
`2
`
`Description
`
`{Technical Field]
`
`{G01} The present invention relates io a prediction
`biock generating device, and more particularty, to an ap-
`paratus of generating a prediction block capable of min-
`imize the arnount of coding bits of a residual block.
`
`{Background Art
`
`In image compression methods such as Motion
`{6002]
`Picture Experts Group (MPEG)-1, MPEG-2, MPEG-4
`and H.264/MPEG-4 Advanced Video Coding (AVC}, one
`picture is divided into macroblocks (MBs) to encode an
`image. Then, the respective MBs are encoded using inter
`prediction or intra prediction.
`{GG03]
`In intra prediction, a current block of a current
`picture is encoded notusing areferencepicture, butusing
`values of pixels spatially adjacent to the current block.
`An intra prediction mode with litte distortion is selected
`by comparing a prediction block generated using the ad-
`jacent pixel values with an original MB. Then, using the
`selectedintra prediction mode andthe adjacent pixel val-
`ues, prediction values of the current block are calculated.
`Differences between the prediction values and pixels val-
`ues of the original current block are calculated and then
`encoded through transform coding, quantization and en-
`tropy coding. The intra prediction mode is also encoded.
`{0604}
`Intra prediction modes are generally classified
`into 4x4 intra prediction mode, 8X8 intra prediction
`mode and 1616 intra prediction mode for luminance
`components and intra prediction mode for chrominance
`components.
`{6905}
`In 16X16 intra prediction mode according to a
`prior art, there are four modesof a vertical mode, a hor-
`izontal mode, a direct current (DC) mode and a plane
`mode.
`
`In 4x4intra prediction according tc the priorart,
`{6006]
`there are nine modes of a vertical mode, a horizontal
`mode, a DC mode, a diagonal down-left mode, a diagonal
`down-right mode, a vertical right mode, a vertical left
`mode, a horzontal-up mode and a horizontal-down
`mode.
`
`40
`
`{e607] Each prediction mode has indexed according
`fo the frequency of use of the respective modes. The
`vertical mode of which mode number is 0 shows the high-
`est possibility of being used most frequently for perform-
`ing intra prediction on a target block, and the horizon-
`tal-up mode of which made number is 8 showsthe highest
`possibility of being used mostinfrequently.
`{e608] According to H.264 standards, a current block
`is encoded using a total of 13 modes, that is, 4 modes
`of the 4x4 intra prediction and 9 modes of the 16x16
`intra prediction. A bit stream of the current block is gen-
`erated according to an optimal mode among these
`modes.
`
`oor
`
`jacent to current block do not exist or are not already
`encoded, itis irnpossible to apply some or all of the intra
`prediction modes to the current Block. Also, when intra
`prediction is performed by selecting prediction mode
`among applicable intra mode, a residue signal between
`@ prediction block and the current block becomeslarge.
`Therefore, the coding efficiency is degraded.
`
`[Disclosure]
`
`[Technical Problem]
`
`{@010] The present inventionis directed to an appara-
`tus of generating a prediction block similar to an original
`block.
`
`[Technical Solution}
`
`20
`
`Nhon
`
`{0044] One aspect of the present invention provides
`an apparatus inchuding: an intra prediction mode decod-
`ing unit configured to restore anintra prediction mode of
`a current prediction unit using additional information in-
`cluded in a received additional information container and
`
`available intra prediction mode candidate information of
`the current prediction unit; a reference pixel generating
`unit configured to generate reference pixels using avail
`able reference pixels; a reference pixel filtering unit con-
`figured to adaptively filter the reference pixels based on
`the restored intra prediction mode of the current predic-
`tion unit and a@ size of the current prediction unit; and a
`prediction block generating unit configured to generate
`a prediction block of the current prediction unit using the
`reference pixels corresponding to the restored intra pre-
`diction mode.
`
`(Advantageous Effects]
`
`{00412] An apparatus according to the presentinvention
`generates reference pixels and adaptively fitters the ref-
`erence pixels in order to generate 4 prediction block min-
`mizing the difference betweenthe prediction block and
`an original block. Also, by adaptively filtering the predic-
`tion block according to intra prediction mode, residual
`signals become smalier and thus an image compression
`can be improved.
`
`[Description of Drawings]
`
`(6073)
`
`FIG. Tis a block diagramilusirating a moving picture
`coding apparatus according to the present invention.
`FiG. 2 is a flow chart illustrating an operation of a
`scanning unit according to the present invention.
`FIG. 3is a block diagram Hlustrating a moving picture
`decoding apparatus according to the present inven-
`tion.
`
`{[OG08] However, when some or all values of pixels ad-
`
`PG. 4 is a block diagramillustrating an intra predic-
`
`IPR2021-00827
`Unified EX1002 Page 935
`
`IPR2021-00827
`Unified EX1002 Page 935
`
`

`

`3
`
`EP 2 600 614 A2
`
`4
`
`tion unit according to the present invention.
`FIG. Sis a conceptual diagrarn showing positions of
`reference pixels usedfor intra prediction according
`to the cresent invention.
`FIG. 6 a flow chartillustrating a process of generating
`reference pixels accarding to the present invention.
`FIG. 7 is a block diagrarn Hustrating an intra predic-
`tion unit of a moving picture decoding apparatus ac-
`cording to the present invention.
`
`{Mode for Invention]
`
`the
`embodiments of
`{6014} Hereinafter, various
`present invention will be described in detail with refer-
`ence to the accompanying drawings. However,
`the
`present invention is not limited to the exemplary embod-
`iments disclosed below, but can be implemented in var-
`ious types. Therefore, many other modifications and var-
`iations of the present invention are possible, and itis to
`be understood that within ihe scopeof the disclosed con-
`cept, the present invention may be practiced ctherwise
`than as has been specifically described.
`{0645}
`For image coding, each picture consists of a
`plurality of slices, and each slice consists of a plurality of
`coding units. Since an image of a high-definition (HD)
`grade or higher has many smooth regions, an image com-
`pression can be improved by encoding the image with
`coding units of larger than an MB of whichsize is 16x16.
`{8616] Asizeofthe coding unitaccording ic the present
`invention may be 16x16, 32x32 or 64x64. A size of the
`coding unit may also be 8x8 or less. A coding unit of the
`largest size is referred to as a super macroblock (SMB).
`Asize of SMB is indicated by a smallest size of the coding
`unit and depth information. The depth information indi-
`cates a difference value between the size of SMB and
`
`the smallest size of the coding unit.
`{0647] Therefore, coding units to be used for coding
`pictures may be SMB or sub-block of SMB. The coding
`units are set to defaults value or are indicated in a se-
`
`quence header.
`{6078] A SMB consists of one or more coding units.
`The SMB has a form of a recursive coding tree so as to
`include the coding units and a division structure of the
`coding units. When the SMB is not divided into four
`sub-coding units, the coding tree may consist of informa-
`tion indicating that the SMB is not divided and one coding
`unit. When the SMB is divided into four sub-coding units,
`the coding tree may consist of informationindicating that
`the SMB is divided and four sub-coding trees. Likewise,
`each sub-coding tree has the same structure as the SMB.
`However, a coding unit of ihe smallest coding unit {SCU)
`size