WO2011040796A2 - Procédé et appareil de codage/décodage d'image utilisant des macroblocs de taille variable - Google Patents

Procédé et appareil de codage/décodage d'image utilisant des macroblocs de taille variable Download PDF

Info

Publication number
WO2011040796A2
WO2011040796A2 PCT/KR2010/006740 KR2010006740W WO2011040796A2 WO 2011040796 A2 WO2011040796 A2 WO 2011040796A2 KR 2010006740 W KR2010006740 W KR 2010006740W WO 2011040796 A2 WO2011040796 A2 WO 2011040796A2
Authority
WO
WIPO (PCT)
Prior art keywords
prediction
macroblock
subblock
type
information
Prior art date
Application number
PCT/KR2010/006740
Other languages
English (en)
Korean (ko)
Other versions
WO2011040796A3 (fr
Inventor
김수년
전병우
한종기
박형미
박민철
김동원
김기백
이주옥
Original Assignee
에스케이텔레콤 주식회사
임정연
문주희
이영렬
김해광
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 에스케이텔레콤 주식회사, 임정연, 문주희, 이영렬, 김해광 filed Critical 에스케이텔레콤 주식회사
Priority to US13/499,406 priority Critical patent/US9549190B2/en
Priority claimed from KR20100096035A external-priority patent/KR101479129B1/ko
Publication of WO2011040796A2 publication Critical patent/WO2011040796A2/fr
Publication of WO2011040796A3 publication Critical patent/WO2011040796A3/fr
Priority to US14/745,181 priority patent/US9609335B2/en
Priority to US14/745,093 priority patent/US9565444B2/en
Priority to US14/745,152 priority patent/US9609334B2/en

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/134—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or criterion affecting or controlling the adaptive coding
    • H04N19/146—Data rate or code amount at the encoder output
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
    • H04N19/103—Selection of coding mode or of prediction mode
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
    • H04N19/103—Selection of coding mode or of prediction mode
    • H04N19/105—Selection of the reference unit for prediction within a chosen coding or prediction mode, e.g. adaptive choice of position and number of pixels used for prediction
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
    • H04N19/17—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object
    • H04N19/176—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object the region being a block, e.g. a macroblock

Definitions

  • the present invention relates to an image encoding / decoding method using a variable size macroblock. More specifically, the present invention relates to a method and apparatus for efficiently encoding and decoding a high resolution image by encoding and decoding a subblock or a prediction reference unit in a macroblock having a variable size using various prediction modes.
  • Video data compression techniques include H.261, H.263, MPEG-2, and MPEG-4.
  • each image is divided and encoded into macroblocks having a fixed size including 16 ⁇ 16 pixels of luminance component and rectangular areas of 8 ⁇ 8 pixels of each color difference component. All luminance and chrominance components of each macroblock are predicted spatially and temporally, and the prediction residual is transmitted by performing transform, quantization, and entropy coding.
  • the encoding apparatus uses a 16x16 pixel block as a fixed macroblock size and determines whether each macroblock is an intra macroblock or an inter macroblock, and then adds a macroblock. The intra prediction or inter prediction is performed by dividing into small blocks.
  • each macroblock can be divided into 16x16, 8x8, and 4x4 sizes, with four prediction modes for 16x16 blocks and nine for 8x8 and 4x4 blocks. Intra predict using one of the prediction modes.
  • the macroblock may be divided into smaller blocks having sizes of 16x16, 16x8, 8x16, 8x8, 8x4, 4x8, and 4x4 and used for inter prediction through motion compensation.
  • the transform is performed in blocks of 8x8 or 4x4 size, and scalar quantization is used for quantization of transform coefficients.
  • macroblocks may be referred to as a size suitable for small QCIF, CIF-like small images of the past, and is inappropriate as a size of a macroblock for high resolution image compression such as 4Kx2K image.
  • macroblocks are encoded by dividing them into smaller block units, but in the case of intra macroblocks, all subblocks in the macroblocks may use only intra mode, and inter macroblocks. In this case, since all subblocks in a macroblock can use only an inter mode, efficient encoding is difficult when encoding a high resolution image including various regions in an image.
  • the present invention provides a method for efficiently predicting macroblock prediction information using a method of using a macroblock of variable size and using various prediction modes for subblocks in the macroblock and a reference prediction unit.
  • the main purpose is to provide an encoding and decoding method.
  • An image encoding / decoding apparatus for achieving the above object encodes a macroblock for each prediction reference unit candidate, determines a prediction reference unit according to the encoding cost for each prediction reference unit candidate, and determines the determined prediction.
  • An image encoder for generating a bitstream including image data encoded in a reference unit, information about a prediction type for each determined prediction reference unit, and prediction data according to the prediction type; And extracting information about a prediction type for each prediction reference unit, prediction data according to the prediction type, and encoded image data from the bitstream, and decoding the image data encoded according to the prediction type for each prediction reference unit to generate a reconstructed image. It characterized in that it comprises a decoder.
  • an apparatus for encoding an image including: a macroblock divider for dividing a variable size macroblock into a plurality of subblocks; A predictor for selectively determining and predicting inter prediction or intra prediction in each subblock unit; And an encoder for encoding split information indicating the size and shape of subblocks in the macroblock, prediction type information for each subblock indicating whether each subblock is an intra mode or an inter mode, and prediction data according to the prediction type of the subblock. Characterized in that.
  • a video encoding apparatus including: a macroblock divider for dividing a variable size macroblock into a plurality of subblocks; A predictor for selectively encoding and predicting inter prediction or intra prediction in units of macroblocks or subblocks within macroblocks according to a prediction type of a macroblock; And an encoder for encoding the prediction type information and the prediction data of each subblock or the prediction data of each subblock according to the segmentation information indicating the size and shape of the subblocks in the macroblock and the macroblock prediction type.
  • an image decoding apparatus provides partition information indicating a size of subblocks in a macroblock of variable size, whether each subblock is intra predicted or inter predicted, from an input bitstream.
  • a decoder which decodes prediction type information indicating a and prediction mode information required for intra or inter prediction; And a predictor for predicting a current block to be decoded in a macroblock by using prediction type information and prediction mode information for each decoded subblock.
  • a video decoding apparatus including macroblock type information indicating a type of a variable macroblock from an input bitstream, split information indicating a size of macroblock subblocks, and a macroblock.
  • a decoder for extracting and decoding prediction type information and prediction mode data of each subblock according to the type information or prediction mode data of each subblock; Predict the current block to be decoded in the macroblock by using the prediction type information and the prediction mode information of each subblock in the macroblock according to the decoded macroblock type, or the macroblock type and the prediction mode information of each subblock in the macroblock. It characterized in that it comprises a predictor.
  • a video decoding apparatus for achieving the above object, a prediction reference unit setter for setting the prediction reference unit in a macroblock of a variable size by extracting the prediction reference unit information from the input bitstream ; And a decoder for extracting and decoding prediction type information for each subblock in the prediction reference unit according to the prediction type information and the prediction type information from the bitstream in the set prediction reference unit.
  • an image encoding / decoding method encodes a macroblock for each prediction reference unit candidate, determines a prediction reference unit according to the encoding cost for each prediction reference unit candidate, and determines the determined prediction. Generating prediction data according to the prediction type and information about the image data encoded in the reference unit, the prediction type for each determined prediction reference unit; And extracting the information about the prediction type for each prediction reference unit, the prediction data according to the prediction type, and the encoded image data from the bitstream, and decoding the encoded image data according to the prediction type for each prediction reference unit. It is done.
  • a video encoding method comprising: dividing a variable size macroblock into a plurality of subblocks; A prediction encoding step of selectively determining inter prediction or intra prediction in units of each subblock; And encoding split information indicating the size of subblocks in the macroblock, prediction type information for each subblock indicating whether each subblock is an intra mode or an inter mode, and prediction data according to the prediction type of the subblock. It features.
  • intra prediction mode information for an intra block and motion information for an inter block can also be encoded.
  • a video encoding method comprising: dividing a variable size macroblock into a plurality of subblocks; Selectively determining inter prediction or intra prediction in units of macroblocks or subblocks within macroblocks according to the prediction type of the macroblocks; And encoding the prediction type information and the prediction data of each subblock or the prediction data of each subblock according to the partition information indicating the size of the subblocks in the macroblock and the macroblock prediction type.
  • a method of decoding an image comprising: partition information indicating a size of each subblock in a macroblock having a variable size from an input bitstream, and inter prediction whether each subblock is intra predicted.
  • Decoding prediction type information indicating whether the information has been received and prediction mode information required for intra or inter prediction; Predicting a current block to be decoded in the macroblock by using prediction type information and prediction mode information for each decoded subblock.
  • the macroblock type information indicating the type of the variable macroblock from the input bitstream, the partition information indicating the size of each subblock in the macroblock, Extracting and decoding prediction type information and prediction mode data of each subblock according to macroblock type information or prediction mode data of each subblock; Predicting the current block to be decoded in the macroblock using the prediction type information and the prediction mode information of each subblock in the macroblock according to the decoded macroblock type, or the macroblock type and the prediction mode information of each subblock in the macroblock. Characterized in that it comprises a step.
  • a method of decoding an image comprising: setting prediction reference units in a macroblock having a variable size by extracting prediction reference unit information from an input bitstream; And extracting and decoding prediction type information for each subblock in the prediction reference unit according to the prediction type information and the prediction type information from the bitstream in the set prediction reference unit.
  • a high resolution image is efficiently encoded using various combinations of prediction modes such as intra prediction or inter prediction for subblocks in a macroblock. Can be decrypted
  • the split information and the prediction information of the macroblock may be efficiently encoded and decoded using the reference prediction unit.
  • 1 to 3 are exemplary diagrams for describing a macroblock in MxN pixel units according to an embodiment of the present invention.
  • 4 and 5 are exemplary diagrams illustrating various subblock modes according to an embodiment of the present invention.
  • FIG. 6 is a block diagram schematically illustrating a video encoding apparatus according to an embodiment of the present invention.
  • FIG. 7 is a block diagram schematically illustrating an image decoding apparatus according to an embodiment of the present invention.
  • FIG. 8 is a block diagram schematically illustrating a video encoding apparatus according to another embodiment of the present invention.
  • FIG. 9 is a flowchart illustrating a video encoding method according to another embodiment of the present invention.
  • FIG. 10 is a block diagram schematically illustrating an image decoding apparatus according to another embodiment of the present invention.
  • FIG. 11 is an exemplary diagram illustrating a subblock obtained by dividing a macroblock into layers according to an embodiment of the present invention.
  • 12 and 13 are exemplary diagrams illustrating a division type according to an embodiment of the present invention.
  • FIG. 14 is an exemplary diagram illustrating a macroblock divided into subblocks of various block sizes according to an embodiment of the present invention.
  • 15 is an exemplary diagram sequentially illustrating a process of dividing a macroblock into divided layers.
  • FIG. 16 is an exemplary diagram for describing a process of sequentially encoding segmentation type indication information for each layer of a macroblock according to the order of subblocks.
  • a video encoding apparatus (Video Encoding Apparatus), a video decoding apparatus (Video Decoding Apparatus) to be described below is a personal computer (PC), notebook computer, personal digital assistant (PDA), portable multimedia player (PMP) : Portable Multimedia Player (PSP), PlayStation Portable (PSP: PlayStation Portable), Mobile Communication Terminal (Mobile Communication Terminal), and the like, and may be used to encode a video or a communication device such as a communication modem for communicating with various devices or a wired or wireless communication network. It refers to various devices having various programs for decoding and a memory for storing data, a microprocessor for executing and controlling a program.
  • PC personal computer
  • PDA personal digital assistant
  • PMP portable multimedia player
  • PSP Portable Multimedia Player
  • PSP PlayStation Portable
  • Mobile Communication Terminal Mobile Communication Terminal
  • the image encoded in the bitstream by the video encoding apparatus is real-time or non-real-time through the wired or wireless communication network, such as the Internet, local area wireless communication network, wireless LAN network, WiBro network, mobile communication network, or the like, or a cable, universal serial bus (USB: Universal)
  • the image decoding apparatus may be transmitted to a video decoding apparatus through a communication interface such as a serial bus, decoded by the video decoding apparatus, reconstructed, and played back.
  • subblocks of a fixed size macroblock of an inter picture are predictively encoded and predictively decoded using the same prediction method. That is, all of the subblocks in one macroblock are inter prediction coded or all intra prediction coded.
  • an image is encoded using a macroblock having a variable size, and may be encoded or decoded by using inter prediction or intra prediction selectively for each subblock in the macroblock.
  • subblocks in a macroblock may be encoded / decoded using only an intra prediction mode, and may be encoded or decoded using intra prediction or inter prediction selectively for subblocks in a macroblock only in an inter picture. have .
  • the macroblock may be classified into three types, an intra macroblock, an inter macroblock, and an intra / inter macroblock, and may be applied only when the macroblock type is an intra / inter macroblock type.
  • each subblock in the intra macroblock is intra prediction encoded
  • each subblock in the inter macroblock is inter-predictive encoded, and in the case of intra / inter macroblock, inter prediction or intra prediction is selectively performed for each subblock in the macroblock.
  • Information about whether an intra prediction mode or an inter prediction mode is used for each block is encoded only when intra prediction or inter prediction is selectively used for each subblock in the macroblock.
  • 1 to 3 are exemplary diagrams for describing a macroblock in MxN pixel units according to an embodiment of the present invention.
  • FIG. 1 exemplarily illustrates macroblocks (hereinafter, referred to as MxN size macroblocks) in MxN pixel units expressed in a part of an input image having an arbitrary size, and in FIG. 2, 396 16x16 size macroblocks are illustrated.
  • a CIF image composed of blocks is exemplarily illustrated, and
  • FIG. 3 exemplarily shows a CIF image composed of 54 64x32 macroblocks.
  • the image is divided into macroblocks having a fixed size of 16x16, encoded, and decoded.
  • an image may be encoded and decoded using a macroblock having a size of 64x32 macroblocks as well as MxN size (but 16x16 size or more) such as 64x64 size and 32x64 size.
  • 4 and 5 are exemplary diagrams illustrating various subblock modes according to an embodiment of the present invention.
  • FIG. 4 shows a subblock mode that can be used for a macroblock of size 32x32
  • FIG. 5 shows a subblock mode that can be used for a macroblock of size of 32x16.
  • the macroblock of MxN size may be divided into smaller blocks, that is, subblocks as shown in FIGS. 4 and 5, and the macroblock of the image is intra-united in such a subblock unit. It may be predictively coded or inter prediction coded.
  • FIG. 6 is a block diagram schematically illustrating a video encoding apparatus according to an embodiment of the present invention.
  • An image encoding apparatus is an apparatus for encoding an image using a macroblock of an arbitrary size or more, and includes a macroblock splitter (605), a predictor (610), and an encoder (620). ), A reproducer 630, a filter 640, and a frame memory 650.
  • the decompressor 630, the filter 640, and the frame memory 650 may be selectively omitted or included in other components according to an implementation method.
  • the macroblock divider 605 divides the macroblock into one or more subblocks using the division type shown in FIGS. 4 and 5 and generates partition information indicating the size and shape of the subblocks in the macroblock. Each divided subblock is input to the predictor 610, and split information is input to the encoder.
  • the predictor 610 may include a motion estimator 612, a motion compensator 614, and an intra predictor 616 to predict the input block.
  • the block refers to a macroblock of size MxN (where M and N may be an integer greater than or equal to 16) or a subblock or subblock of size OxP (where O and P are integers less than or equal to M or N).
  • the motion estimator 612 compares the block to be predicted with the reference picture stored in the frame memory 650 and estimates the motion of the block to generate a motion vector.
  • the motion compensator 614 may refer to the motion vector generated by the motion estimator 612 to obtain blocks corresponding to the size of the block to be predicted from the reference picture stored in the frame memory 650.
  • the block obtained by the motion compensator 614 becomes a macroblock having the predicted value of the block to be predicted.
  • the intra predictor 616 intra predicts a block to be predicted. To this end, the intra predictor 616 generates a reference block using neighboring pixel information that is already encoded, decoded, and reconstructed, compares the reference block with the encoding target block, and determines an intra prediction mode, and determines a block according to the determined intra prediction mode. Intra prediction. The block predicted by the intra predictor 616 becomes a predicted block having the predicted value of the target block.
  • the encoder 620 encodes partition information indicating the size and shape of subblocks in the macroblock generated by the macroblock divider 605, and may entropy encode the partition information. Also, the encoder 620 encodes prediction information of each subblock in the macroblock divided by the split information. When inter prediction of a target block to be encoded, a motion such as a motion vector generated by the motion estimator 612 may be used. When intra prediction and information are performed, prediction mode information such as an intra prediction mode may be encoded. In addition, the residual signal, which is a difference between the pixel values of each subblock and the prediction block, is encoded. The residual signal may be transformed, quantized, entropy encoded, and encoded.
  • the reconstructor 630 inversely quantizes and inverse transforms the transformed and quantized residual signal and adds the prediction block output from the predictor 610 to reconstruct the target block.
  • the filter 640 filters the restored target macroblock using a filter such as a deblocking filter.
  • the filtered reconstructed macroblock is stored in the frame memory 650 and used by the predictor 610 to predict the next macroblock or the macroblock of the next picture.
  • the size of the macroblock is fixed to 16x16, only subblocks having a small number of shapes, such as 8x8 and 4x4 sizes, may be used accordingly.
  • the size of the macroblock may be variously determined to be 16 ⁇ 16 or more, the size and shape of the subblock may be variously determined, and accordingly, the macroblock may be divided into various types of subblocks.
  • information on how the macroblock is divided into subblocks of what size and in what form is transmitted to the image decoding apparatus in the same manner as the macroblock is divided in the image encoding apparatus. It should be partitioned so that it can be predictively decoded.
  • a macroblock may be divided into subblocks of various sizes for each layer, and prediction encoding and prediction decoding may be performed for each divided subblock.
  • FIG. 11 is an exemplary diagram illustrating a subblock obtained by dividing a macroblock into layers according to an embodiment of the present invention.
  • the size of the macroblock shown in FIG. 11 is N ⁇ N (N is an integer of 16 or more).
  • the minimum subblock size is assumed to be 4 x 4. However, this assumption is only for explaining an embodiment of the present invention, the horizontal and vertical size of the macroblock may not be the same and the minimum subblock size may be other size other than 4x4.
  • a macroblock may be divided into subblocks of various sizes for each layer.
  • the macroblock may be divided into subblocks having four shapes for each layer from layer 0 to layer log2 (N / 4).
  • the subblocks of the K + 1 layer may be used only when the subblocks of the corresponding layer are divided into four subblocks in the layer K (where 0 ⁇ K ⁇ log2 (N / 4)).
  • the macroblock when the macroblock is a 64x64 block, the macroblock may be divided into four layers from layers 0 to 3, and each layer may include subblocks having four different block sizes. Therefore, a 64x64 block sized subblock, a 64x32 block sized subblock, a 32x64 block sized subblock, a 32x32 block sized subblock belong to layer 0, a 32x32 block sized subblock, a 32x16 block sized subblock, Subblocks of 16x32 block size, subblocks of 16x16 block size belong to Layer 1, subblocks of 16x16 block size, subblocks of 16x8 block size, subblocks of 8x16 block size, subblocks of 8x8 block size Subblocks of 8x8 block size, subblocks of 8x4 block size, subblocks of 4x8 block size, and subblocks of 4x4 block size belong to layer 3.
  • subblocks belonging to layer 1 may be used only when a macroblock having a 64x64 block size is divided into subblocks having a 32x32 block size in layer 0, and a subblock having a size of 32x32 block of layer 1 is a subblock having a 16x16 block size
  • Subblocks belonging to layer 2 may be used only when the block is divided into blocks
  • subblocks belonging to layer 3 may be used only when the subblock of layer 2 is divided into subblocks of 8 ⁇ 8 block size.
  • four NxN blocks with layer number K If divided into subblocks, The subblock may belong to layer K or may belong to layer K + 1. That is, when the macroblock is 64x64 size, the 32x32 sized subblock may be determined as a subblock type included in layer 0 or may be determined as a subblock type belonging to layer 1.
  • the method of assigning a layer number to the divided subblocks depends on whether each layer is available. If layer K + 1 is available, layer number K + 1 is assigned to the subblock. If layer K + 1 is not available, layer number K is assigned to the layer number of the subblock.
  • the 32x32 subblock belongs to layer 1. If one 32x32 subblock in a macroblock is divided into four 16x16 subblocks, each 16x16 subblock belongs to Layer 2, and in the same way, if a 16x16 subblock is divided into four 8x8 subblocks, the 8x8 subblock belongs to Layer 3. do. If the 8x8 subblock is divided into four 4x4 subblocks, the 4x4 subblocks belong to layer 3 because layer 4 is not available.
  • K may be allocated to the layer number of the subblock. In this case, If the subblock is divided into smaller subblocks, the layer number K + 1 is allocated.
  • the 32x32 subblock belongs to layer 0 when the macroblock is 64x64 divided into four 32x32 subblocks. If one 32x32 subblock in a macroblock is divided into four 16x16 subblocks, each 16x16 subblock belongs to Layer 1, and in the same way, if a 16x16 subblock is divided into four 8x8 subblocks, the 8x8 subblock belongs to Layer 2. do. When the 8x8 subblock is divided into four 4x4 subblocks, the 4x4 subblocks belong to layer 3.
  • the macroblock may be divided using various types of division as shown in FIGS. 12 and 13. 12 and 13 are exemplary diagrams illustrating a division type according to an embodiment of the present invention.
  • FIG. 12 is an exemplary diagram illustrating a partition type according to an embodiment of the present invention.
  • partition type indication information (Partition Type Number) for identifying a block size of a subblock divided by layer.
  • partition type indication information is given as 0, and the subblock is Going two When divided into subblocks, partition type indication information is given as 1, and the subblocks Going two When divided into subblocks, partition type indication information is given as 2, and the subblocks There are four When divided into subblocks, a partition type number may be assigned to 3.
  • the partition number indicates a number assigned to identify each subblock divided according to the partition type. For example, a subblock of layer K If is not split, the undivided subblock The partition number of 0 is given.
  • subblocks of layer K There are four When divided into subblocks, each The subblocks may be assigned partition numbers 0, 1, 2, and 3 in the raster scan direction from the upper left subblocks. 13 is an exemplary diagram illustrating a partition type according to an embodiment of the present invention.
  • FIG. 13 exemplarily shows partition type indication information (Partition Type Number) for identifying a block size of a subblock divided by layer.
  • partition type indication information partition type indication information is given as 0 and the subblock is There are four When divided into subblocks, a partition type number may be assigned to 1. The partition number indicates a number assigned to identify each subblock divided according to the partition type. For example, a subblock of layer K If is not split, the undivided subblock The partition number of 0 is given. Also, subblocks of layer K There are four When divided into subblocks, each The subblocks may be assigned partition numbers 0, 1, 2, and 3 in the raster scan direction from the upper left subblocks.
  • various division types may be combined for each layer.
  • the division type shown in FIG. 13 may be used, and the lower layers of the layer 1 may use the division type shown in FIG. 12.
  • the encoder 620 encodes the partition information received from the macroblock splitter 605 and a prediction type indicating whether each subblock is an intra mode or an inter mode and prediction data according to the prediction type.
  • the encoder 620 first encodes the partition information to transmit the size and shape of the subblocks in the macroblock to the decoder, and encodes the prediction type and the prediction data of each subblock.
  • the encoder 620 encodes partition information indicating the size and shape of subblocks in the macroblock received from the macroblock divider 605.
  • partition information of a block indicating a form in which a macroblock is divided into subblocks having various sizes may be represented using split type indication information for each partition layer of the macroblock. Accordingly, the plurality of subblocks constituting the macroblock may be identified by partition type indication information for each partition layer.
  • the encoder 620 may encode the partition information of the block by using the partition type indication information of each layer of the macroblock. As described below, the encoder 620 may encode the partition information of the block in various ways.
  • the encoder 620 may encode split information of the current block by sequentially encoding split type indication information for each partition layer of the macroblock according to the split type indication information encoding order.
  • FIG. 14 is an exemplary diagram illustrating a macroblock divided into subblocks of various block sizes according to an embodiment of the present invention.
  • FIG. 14 a macroblock having a block size of 64 ⁇ 64, a maximum partition layer of 4, and a subblock having various block sizes is illustrated using a subblock type shown in FIG. 10.
  • the partition type indication information for each partition layer may be sequentially encoded according to the partition type indication information encoding order to encode the partition information of the macroblock.
  • a process of dividing the macroblocks shown in FIG. 14 for each of the divided layers may be represented as shown in FIG. 15.
  • a 64x64 block sized subblock is divided into four 32x32 block sized subblocks, and in layer 1 (L1), L1-P0 (partition number 0 of layer 1)
  • the subblock of L1-P3 is further divided into four 16x16 block size subblocks, and the subblocks of L1-P1 (partition number 1 of layer 1) and L1-P2 (
  • the subblocks of partition number 2) of layer 1 are divided into subblocks of 16x32 block size and subblocks of 32x16 block size, respectively.
  • the partition number for each layer is not represented.
  • layer 2 the subblocks of L2-P0 (partition number 0 of layer 2) are further divided into four 8x8 block size subblocks and L2-P3 (partition number 3 of layer 2) is divided into two 16x2 blocks. It is divided into subblocks of size.
  • L3 the subblocks of L3-P0 (partition number 0 of layer 3) and the subblocks of L3-P1 (partition number 1 of layer 3) are each divided into four 4x4 block size subblocks.
  • the order of encoding division type indication information is as follows.
  • split type indication information indicating a partition type of a macroblock when the macroblock is divided into four subblocks, split type indication information of each divided subblock is continuously encoded. For example, when an NxN block is divided into four subblocks, the partition type indication information of the first (N / 2) x (N / 2) subblock in the NxN block is encoded and the first (N / 2) x (N / 2). When the subblock is divided into four subblocks again, the partition type indication information of the divided (N / 4) x (N / 4) blocks is encoded.
  • FIG. 16 is an exemplary diagram for describing a process of sequentially encoding split type indication information for each layer of a macroblock. to be.
  • the encoding may be performed as shown in FIG. 16.
  • numerals written in ' ⁇ ' indicate an order of encoding division type indication information of each subblock.
  • the layer type partition type indication information is sequentially encoded in the order shown in FIG. 14.
  • the partition type indication information 3 is encoded. Since the first 32x32 block size subblocks L1-P0 among the four 32x32 block size subblocks in the 64x64 block size subblock are also divided into four 16x16 block size subblocks, the partition type indication information 3 is encoded. The first 16x16 block size subblock (L2-P0) of the four 16x16 block size subblocks in the first 32x32 block size subblock (L1-P0) of Layer 1 is also divided into four 8x8 block size subblocks.
  • split type indication information ⁇ 3, 3, 0, 0 ⁇ is encoded. Since subblocks of the layer 3 cannot be divided into smaller subblocks, the partition type indication information of the subblocks belonging to the layer 3 is not encoded.
  • the partition type indication information of the second 16x16 block size subblock L2-P1 and the third 16x16 block size subblock L2-P2 is encoded. However, since all of them are no longer divided into small blocks, division type indication information 0 is encoded.
  • the fourth 16x16 block size subblock L2-P3 is divided into 16x8 block size subblocks, but since the partition type indication information is not 3, only the partition type indication information 1 is encoded.
  • the split type indication information of the four subblocks in the layer 2 is encoded, the split type indication information of the subblocks L1 to P1 of the second 32x32 block size of the layer 1 is encoded, and the split type indication information of the second 32x32 block size of the layer 1 is encoded.
  • the subblocks L1-P1 are divided into subblocks having a size of 16 ⁇ 32 blocks, and each of the divided subblocks is no longer divided into small subblocks, thereby encoding partition type indication information 2.
  • the third 32x32 block size subblock L1-P2 of layer 1 the fourth 32x32 block size subblock L1-P3 of layer 1, and four 16x16 block size subblocks below it (
  • the split type indication information of L2-P0, L2-P1, L2-P2, and L2-P3) is sequentially encoded, ⁇ 1, 3, 0, 0, 0, 0 ⁇ is encoded.
  • the division type indication information ⁇ 3, 3, 3, 3, 3, 0, 0, 0, as shown in FIG. , 0, 1, 2, 1, 3, 0, 0, 0 ⁇ is encoded.
  • split type indication information encoding may also be performed in the following order.
  • 0, 0 ⁇ may be encoded
  • partition type indication information ⁇ 3, 3, 0, 0 ⁇ of four subblocks of the layer 3 (four subblocks belonging to L2-P0 in L1-P0) may be encoded.
  • split type indication information ⁇ 3, 3, 2, 1, 3, 3, 0, 0, 1, 0, 0, 0, 0, 3, 3, 0, 0 ⁇ is encoded.
  • the method of encoding the split type indication information may be encoded into a binary bit string using lossless compression encoding such as binary arithmetic coding or Huffman coding.
  • each type of partition type indication information may use different binary values according to the layer number of the type of partition type indication information to be currently encoded.
  • the dividing type indication information is one bit long indicating whether or not the current block is divided into four subblocks. It may be a flag.
  • the encoder After encoding the split information of the macroblock, the encoder encodes the prediction data according to the prediction type and the prediction type of each subblock.
  • the syntax and the encoding / decoding method representing the prediction type and the prediction data according to each picture type may be as follows.
  • intra prediction or inter prediction is selected and encoded, and then encoding prediction type information indicating whether intra prediction or inter prediction has been performed for each subblock, and in case of an intra block, intra prediction mode information and inter block In this case, motion information is encoded.
  • an available prediction type may vary according to the picture type.
  • the types of available prediction types that each subblock can use may be SKIP mode, intra prediction, or inter prediction encoding motion information.
  • the types of available prediction types that can be used by each subblock may be a SKIP mode, intra prediction, or an inter prediction or direct mode for encoding motion information.
  • the types of prediction types usable may vary according to the size and shape of the subblock.
  • Tables 1 to 4 show examples of usable prediction types and types according to the size and shape of blocks.
  • the types of prediction types according to the blocks of Tables 1 to 4 may be different only by way of example. This is because when both intra mode and inter mode are available, it is recommended to use inter mode in areas without texture, such as a background, in which case the block size is generally large.
  • the complex region is intended to select one of the intra mode and the inter mode, since the complex region is mainly small in size of the subblock. This is because macroblocks larger than 16 ⁇ 16 may exist in various regions within a macroblock because encoding efficiency may be deteriorated when the prediction encoding is selectively performed using inter prediction or intra prediction in units of macroblocks.
  • prediction type information is not encoded, and prediction type information is encoded only when two or more types are used.
  • intra prediction or inter prediction it may indicate whether intra prediction or inter prediction is performed using a syntax of 1 bit length.
  • the prediction type may be indicated by using a 1-bit or 2-bit length syntax. For example, if there are three types of prediction types that can be used, SKIP, inter prediction, and intra prediction, first, one bit indicating whether or not SKIP is encoded, and then, if not in SKIP mode, one more bit is encoded to inter prediction. It may indicate whether the subblock is an intra predicted subblock. Alternatively, even when the available prediction type is Direct, inter prediction, or intra prediction, one bit indicating whether a direct mode is encoded may be encoded, and if not, the first bit indicating whether intra prediction or inter prediction is further encoded. have.
  • SKIP SKIP
  • Direct a type of prediction type that can be used
  • inter prediction first, one bit indicating whether SKIP or not is encoded, and if it is not in SKIP mode, one more bit is further encoded to determine whether it is a direct predicted subblock. It may indicate whether an inter predicted subblock.
  • the prediction type may be encoded by using a syntax of 1 bit to 3 bits in length. For example, if the types of prediction types available are SKIP, Direct, Inter Prediction, and Intra Prediction, first, one bit indicating whether SKIP is SKIP or not is encoded. Indicates whether or not. If it is not the SKIP mode or the direct mode, one bit may be encoded to indicate whether it is an inter predicted subblock or an intra predicted subblock.
  • the prediction type may be encoded by using a 2-bit fixed length syntax.
  • the prediction type may be encoded by using a promised table.
  • variable length encoding method of the syntax indicating the prediction type may be various binary codes such as unary code, truncated unary code, exponential golem code, etc. Coding) can be used to encode.
  • arithmetic encoding may be performed using the prediction type information of the neighboring block.
  • the prediction type may be encoded using a plurality of syntaxes.
  • three syntaxes may be used: SKIP_flag to indicate whether it is in SKIP mode or Direct_flag to indicate whether it is in direct mode or pred_type to indicate whether it is an intra block or an inter block.
  • SKIP_flag to indicate whether it is in SKIP mode
  • Direct_flag to indicate whether it is in direct mode
  • pred_type to indicate whether it is an intra block or an inter block.
  • Each syntax can also be encoded using a different method.
  • the encoded data of the next subblock is encoded.
  • Pred_dir which is information indicating a prediction direction
  • Pred_dir is encoded in the prediction indicating whether the L0 prediction, the L1 prediction, or the bidirectional prediction is performed.
  • the picture type is a P picture and the prediction type is a subblock that encodes motion information
  • a difference motion vector (MVD) and a reference picture index (ref_idx) indicating a reference picture are encoded.
  • Pred_dir is encoded and one or two pieces of motion vector information and a reference picture index according to Pred_dir are coded.
  • the reference picture index is not encoded.
  • Intra prediction mode information is encoded for a subblock in which the prediction type identified by the subblock in the intra picture or the prediction type information in the inter picture is intra predicted. For example, when the size of the subblock is a 4x4 subblock, a prediction mode indicating a selected prediction mode among nine intra prediction modes is encoded.
  • FIG. 7 is a block diagram schematically illustrating an image decoding apparatus according to an embodiment of the present invention.
  • An image decoding apparatus may include a decoder 710, a predictor 720, a decompressor 730, a filter 740, and a frame memory 750.
  • the decoder 710 extracts three types of information required for macroblock decoding from the input bitstream.
  • the decoder 710 firstly extracts partition information indicating the size and shape of each subblock in the macroblock to be decoded from the bitstream and entropy decodes it. The decoder 710 may then identify which type of subblock the macroblock is divided by using the reconstructed partition information. Second, the decoder 710 entropy decodes and extracts prediction type information indicating whether each subblock is intra predicted or inter predicted and prediction mode information required for intra or inter prediction. The type of prediction data to be decoded and the method of decoding the prediction data vary depending on whether each block is an intra block or an inter block.
  • reference picture information necessary for motion compensation of each subblock from the bitstream Extracts and decodes information about a motion such as a motion vector and a motion vector.
  • information about an intra prediction mode of a luminance component and a chrominance component is extracted and decoded from a bitstream.
  • the decoder 710 decodes information necessary for decoding the residual signal.
  • information indicating whether each subblock has a nonzero transform coefficient for example, CBP
  • transform information indicating a type of transform and a quantized transform coefficient are decoded.
  • the predictor 720 predicts a current block to be currently decoded, and may include a motion compensator 722 and an intra predictor 724.
  • the motion compensator 722 obtains a pixel equal to the size of the current block from the reference picture stored in the frame memory using the motion vector decoded and reconstructed by the decoder to generate a prediction block.
  • the intra predictor 724 predicts the current block according to the intra prediction mode decoded and reconstructed by the decoder 710 to generate a predicted block.
  • the decompressor 730 inversely quantizes the quantized transform coefficients decoded by the decoder 710 and inversely transforms the dequantized transform coefficients using the transform type extracted and reconstructed by the decoder to generate a residual signal.
  • the reconstructed block is generated by adding the generated residual signal to the prediction block generated by the predictor.
  • the generated reconstruction block is filtered by the filter 740 and stored in the frame memory 750 and used to reconstruct the next macroblock or the next picture.
  • the macroblock is divided into one or more subblocks using the reconstructed split information. Thereafter, the prediction type and the prediction mode information of each subblock are extracted from the bitstream using the size and shape information of the subblocks in the macroblock.
  • partition information which is information indicating the size and shape of subblocks used for prediction or transformation in a macroblock.
  • the split type indication information is decoded in the order promised with the image encoding apparatus by using a subblock type available for each layer promised by the encoder.
  • the usable subblock type for each layer may be the subblock type shown in FIGS. 15 and 16, and the order of decoding the partition type indication information may be sequentially decoded according to the order shown in FIG. 14 or 16. have.
  • the decoder 710 extracts and decodes the first partition type indication information from the bitstream to restore the partition type indication information of the macroblock layer 0. If the reconstructed partition type indication information value is 0, it means that the macroblock is not divided into subblocks, and thus the decoding of the partition type indication information of the current macroblock ends. Thereafter, prediction or inverse transformation is performed with macroblock N ⁇ N size.
  • the macroblock When the partition type indication information of the reconstructed layer 0 is 1, the macroblock is divided into two Nx (N / 2) subblocks, and the decoding of the partition type indication information of the current macroblock is completed. Thereafter, the macroblock is predicted or inversely transformed in units of Nx (N / 2).
  • the macroblock is divided into two (N / 2) xN subblocks, and the decoding of the partition type indication information of the current macroblock is completed. Thereafter, the macroblock is predicted or inversely transformed in units of (N / 2) ⁇ N.
  • the subblock decodes partition type indication information of the first subblock (the subblock having partition number 0 of layer 1).
  • the layer number of the subblocks is 1, which is increased by 1 from the upper layer number.
  • the partition type indication information of the subblock having partition number 0 of Layer 1 extracted and decoded from the bitstream is not 3, the second in the macroblock The division type indication information of the subblock (block having partition number 1 of layer 1) is decoded.
  • the partition type indication information of the subblock having partition number 0 of layer 1 extracted from the bitstream and decoded is 3, the current subblock is divided into four subblocks, and the layer number is 2 at this time. Thereafter, the partition type indication information of the subblock corresponding to partition number 0 of the rare 2 is extracted from the bitstream and decoded.
  • the layer number K of the current subblock having the partition number Y is the maximum value that the layer number can have, if the split type indication information of the decoded current subblock (ie, layer K-partition number Y) is 3, the current subblock After dividing the block into four subblocks, the partition type indication information of the next subblock (the subblock having the layer K-partition number Y + 1) is decoded in the raster scan order.
  • the partition type indication information of the unblocked subblocks of the higher layer is decoded.
  • the value encoded by the split type indication information in the encoder is ⁇ 3, 3, 3, 3, 3, 0, 0, 0, 0, 1, 2, 1, 3, 0, 0, 0, 0 ⁇ .
  • split type indication information of layer 0 is decoded.
  • the 64x64 macroblock is divided into four 32x32 blocks (L1-P0, L1-P1, L1-P2, and L1-P3).
  • each 32x32 subblock can be divided into smaller subblocks, the partition type indication information of the first 32x32 subblocks L1-P0 in the 64x64 macroblock is decoded.
  • the split type indication information decoded is 3
  • the L1-P0 subblock is divided into four 16x16 subblocks (L2-P0, L2-P1, L2-P2, and L2-P3), and the split type indication information of L2-P0 is obtained. Is extracted from the bitstream and decoded.
  • the L2-P0 subblock having a size of 16x16 is divided into four 8x8 subblocks (L3-P0, L3-P1, L3-P2, and L3-P3), and the partition type of L3-P0 is divided.
  • the indication information is extracted from the bitstream and decoded.
  • the 8x8 size L3-P0 subblock is divided into four 4x4 subblocks. Since the maximum partition layer of the 4x4 subblock is 4, it can no longer be divided into small subblocks, and thus the partition type indication information of L3-P1 is extracted from the bitstream and decoded.
  • the 8x8 size L3-P1 subblock is divided into four 4x4 subblocks, and the split type indication information of L3-P2 is extracted from the bitstream and decoded.
  • the 8x8 size L3-P2 subblock is not divided, and the split type indication information of the next subblock L3-P3 is extracted from the bitstream and decoded.
  • the seventh decoded split indication information is 0, no 8x8 size L3-P3 subblock is split. Since the partition number of the current subblock is the maximum value of the partition number belonging to the current layer, the partition type indication information of the upper layer L2-P1 is extracted from the bitstream and decoded.
  • the block size of L2-P1 is 16x16.
  • each subblock type is determined by extracting and decoding the split type indication information of L2-P2 and L2-P3 from the bitstream.
  • the block size of L2-P2 is 16x16 and the 10th decoded split indication information is 1, so L2-P3 is split into two 16x8 subblocks.
  • the split type indication information of the second 32x32 sized subblock L1-P1 of the upper layer 1 is decoded.
  • the split indication information decoded by the eleventh is 2, the 32x32 block corresponding to L1-P1 is divided into two 16x32 subblocks, and the split type indication information of L1-P2 is decoded.
  • the twelfth decoded split indication information is 1, the 32x32 block corresponding to L1-P2 is divided into two 32x16 subblocks, and the split type indication information of L1-P3 is decoded.
  • the thirteenth decoded split indication information is 3, so a 32x32 block corresponding to L1-P3 is divided into four 16x16 subblocks (L2-P0, L2-P1, L2-P2, and L2-P3) and each sub in the same manner. Decode the partition type indication information of the blocks.
  • the subblock type of L2-P0 is 16x16, and since it is no longer divided, the next subblock L2-P1 decodes the split type indication information.
  • the subblock type of L2-P1 is 16x16, and since it is no longer divided, the next subblock L2-P2 decodes the split type indication information.
  • the subblock type of L2-P2 is 16x16, and since the split block indication information is no longer divided, the next subblock, L2-P3, decodes the split type indication information.
  • the subblock type of L2-P3 is 16x16, and since the type of the mode subblocks in the macroblock is determined, decoding of the split indication information for decoding the current macroblock is finished.
  • the value encoded by the split type indication information is ⁇ 3, 3, 2, 1, 3, 3, 0, 0, 1, 0, 0, 0, 0, 3, 3, 0, 0 ⁇ .
  • split type indication information of layer 0 is decoded.
  • the 64x64 macroblock is divided into four 32x32 blocks (L1-P0, L1-P1, L1-P2, and L1-P3).
  • L1-P0 and L1-P3 have four 16x16 bits.
  • L1-P1 is divided into two 16x32 blocks
  • L1-P2 is divided into two 32x16 blocks.
  • the partition type indication information of eight 8x8 subblocks of layer 2 belonging to L1-P0 and L1-P3 is extracted and decoded from the bitstream.
  • the partition type indication information of four subblocks (L2-P0, L2-P1, L2-P2, L2-P3) belonging to the restored L1-P0 is ⁇ 3, 0, 0, 1 ⁇ , and 4 belonging to L1-P3 Since the partition type indication information of the two subblocks (L2-P0, L2-P1, L2-P2, L2-P3) is ⁇ 0, 0, 0, 0 ⁇ , L2-P0 belonging to L1-P0 has four 4x4 subblocks. L2-P3 is divided into two 8x4 subblocks.
  • the four subblocks belonging to L2-P1, L2-P2, and L1-P3 belonging to L1-P0 are not divided because the partition type indication information is all zero.
  • the subblocks L2-P0 belonging to L1-P0 are divided into four subblocks, but cannot be divided into smaller subblocks, so that the decoding of the split type indication information for decoding the current macroblock is completed.
  • the entropy decoding method of the split type indication information is entropy decoded using a method promised by an encoder among lossless compression coding / decoding methods such as binary arithmetic coding or Huffman coding.
  • the decoder when the encoder uses binary arithmetic coding as a method of encoding the partition type indication information and uses a binary table according to the layer number of the partition type indication information, the decoder also uses a binary table according to the layer number promised by the encoder. Perform entropy decoding.
  • the encoder when the encoder is promised to use the partition type according to another embodiment of the present invention shown in FIG. 16, it is determined whether the current subblock is divided into four subblocks by entropy decoding one bit for decoding the partition type indication information. You can also determine whether or not.
  • the decoder decodes prediction type information and prediction data according to the prediction type of each subblock in the macroblock.
  • the prediction type information decoding method may not decode from the bitstream according to the type of the picture.
  • the prediction type information decoding method may not decode from the bitstream according to the type of the picture.
  • all the macroblocks in the intra picture and subblocks in the macroblocks perform intra prediction without extracting and decoding prediction type information from the bitstream.
  • the prediction type information is extracted and decoded from the bitstream, and then the prediction type indicating whether each block is intra predicted or inter predicted is restored.
  • intra mode information or motion information according to the prediction type of each block determined by the type of the picture or the prediction type information received through the bitstream is extracted and decoded from the bitstream.
  • the prediction type information decoding process is performed.
  • the prediction type information decoding process will be described.
  • the types of usable prediction types vary according to the types of pictures and the size and shape of the subblocks, and the method of decoding prediction type information varies according to the types of usable prediction types.
  • the types of available prediction types that can be used by each subblock may be SKIP mode, intra prediction, or inter prediction encoding motion information.
  • the types of usable prediction types that each subblock can use may be SKIP mode, intra prediction, or inter prediction or direct mode for encoding motion information.
  • prediction types usable according to the size and shape of the subblock may vary as described in Tables 1 to 4 above.
  • prediction type information is not decoded, and prediction type information extraction and decoding are performed only on two or more subblocks. .
  • one bit is extracted from the bitstream and entropy decoded to set a selected prediction type of intra prediction and inter prediction.
  • one or two bits are extracted and decoded from the bitstream. For example, if there are three types of prediction types available, SKIP, inter prediction, and intra prediction, one bit indicating whether or not SKIP is first decoded, and then, if not in SKIP mode, one bit is further extracted from the bitstream. Then, decoding is performed to determine whether an inter predicted subblock or an intra predicted subblock. Or 1 bit indicating whether it is intra prediction or inter prediction if not in direct mode, after extracting and decoding one bit indicating whether the mode is Direct mode from the bitstream even if the available prediction type is Direct, inter prediction, or intra prediction. You can also extract and decode more.
  • SKIP SKIP
  • Direct a type of prediction type that can be used
  • inter prediction first, one bit indicating whether or not SKIP is decoded, and then, if not in SKIP mode, further decoding one bit to determine whether it is a direct predicted subblock. It may be determined whether it is an inter predicted subblock.
  • one to three bits may be extracted from the bitstream and decoded to decode the prediction type. For example, if the type of prediction type available is SKIP, Direct, Inter Prediction, or Intra Prediction, first decode one bit indicating whether it is SKIP or not, and if it is not in SKIP mode, decode one more bit to determine whether it is Direct mode. Indicates whether or not. If neither the SKIP mode nor the direct mode, 1 bit may be decoded to determine whether an inter predicted subblock or an intra predicted subblock.
  • the prediction type may be decoded by extracting and decoding 2 bits from the bitstream.
  • the prediction type may be decoded using the table promised by the encoder.
  • variable length decoding of the syntax indicating the prediction type may be performed using various methods such as unary code, truncated unary code, and exponential golem code. Can be decrypted
  • arithmetic decoding may be performed using the prediction type information of the neighboring block.
  • the prediction type may be decoded using a plurality of syntaxes. For example, three syntaxes may be used: SKIP_flag to indicate whether it is in SKIP mode or Direct_flag to indicate whether it is in direct mode or pred_type to indicate whether it is an intra block or an inter block. Each syntax can also be decoded using a different method.
  • Pred_dir which is information indicating a prediction direction among predictions indicating whether L0 prediction, L1 prediction, or bidirectional prediction, is decoded.
  • the difference motion vector MVD and a reference picture index ref_idx representing the reference picture are decoded. If the type of picture is a B picture, Pred_dir is decoded, and one or two pieces of motion vector information and a reference picture index according to Pred_dir are decoded. However, if there is only one reference picture that can be referred to, the reference picture index is not decoded.
  • the intra prediction mode information is decoded for the subblock in which the prediction type identified by the subblock in the intra picture or the prediction type information in the inter picture is intra predicted. For example, when the size of the subblock is 4x4, the prediction mode indicating the selected prediction mode among the nine intra prediction modes is decoded.
  • the image encoding apparatus and the image decoding apparatus can encode and decode an image using blocks of arbitrary sizes, and SKIP, DIRECT, and intra prediction for each subblock. It can be encoded / decoded using various prediction types and modes such as and inter prediction.
  • an apparatus and method for selectively encoding and decoding using intra prediction or inter prediction for a subblock in a macroblock of variable size while saving bits required for prediction type information Let's explain.
  • the macroblock is compared to determine the type of the macroblock by comparing the encoding cost, and the type of the selected macroblock. And predictively encode each subblock according to the available prediction type and the prediction mode according to the macroblock type to generate a bitstream.
  • a macroblock type is compared by comparing an encoding cost when using only intra prediction, an encoding cost when using only inter prediction, and an encoding cost when using both intra prediction and inter prediction.
  • each of the subblocks is predictively encoded according to the type of the selected macroblock and the selected prediction method of each subblock, thereby generating a bitstream.
  • the prediction type information and the prediction mode information are encoded for subblocks in the macroblock only when the determined macroblock type is a macroblock type using both intra prediction and inter prediction. Only the information is encoded to generate a bitstream.
  • the prediction block may be predictively encoded using only one of inter prediction and intra prediction according to the size of the macroblock, or may be selectively encoded using inter prediction or intra prediction. That is, the type of macroblocks that can be used may vary depending on the size of the macroblocks. For example, when the size of the macroblock is 32x32 or more, the prediction block may be predicted and encoded using a macroblock type that may select and use intra prediction or inter prediction for each subblock unit in the macroblock. However, the size of the macroblock may be 16x16 or 8x8. In the case of a macroblock, intra prediction or inter prediction is selectively used on a macroblock basis. In this case, when intra prediction or inter prediction may be selected and encoded in units of subblocks in the macroblock, prediction type information is encoded in units of subblocks only in the case of a macroblock type.
  • the size of the macroblock when the size of the macroblock is large, there may be various regions within the macroblock. For example, in the case of a 64x64 macroblock, a background area within the macroblock and a new area not included in the previous picture may be included in the macroblock. If intra prediction or inter prediction is selected in macroblock units, encoding efficiency may be reduced. Therefore, in the case of large macroblocks, various prediction types and modes in the macroblocks can be used to increase coding efficiency, and in the case of small macroblocks, intra prediction or inter prediction can be selected and used in units of macroblocks to minimize the types of prediction types. This is to reduce the number of bits required for the prediction type information.
  • the image encoding apparatus may be configured identically or similarly to the image encoding apparatus according to the embodiment of the present invention described above with reference to FIG. 6.
  • the image encoding apparatus encodes a macroblock using only a prediction type according to a macroblock type, and compares the encoding cost to determine a macroblock type and a prediction type of subblocks in the macroblock.
  • a bitstream is generated by predictively encoding each subblock according to the selected prediction scheme. For example, a macroblock type is determined by comparing the encoding cost using only intra prediction, the encoding cost using only inter prediction, and the encoding cost using both intra prediction and inter prediction. Accordingly, each subblock is predictively encoded to generate a bitstream.
  • the subblock in the macroblock is a macroblock type capable of selectively using intra prediction or inter prediction
  • whether an image is encoded by using inter prediction on each subblock or Information about a prediction type indicating whether to encode using intra prediction may be encoded and included in the bitstream.
  • all subblocks in one macroblock may be predictively encoded using inter prediction, all subblocks may be predictively encoded using intra prediction, and some subblocks may be May be inter predictively coded and the remaining subblocks may be intra predictively coded and may save bits required for prediction type information.
  • the macroblock type may be a SKIP macroblock, an intra macroblock, an inter macroblock, or a mixed inter / intra macroblock.
  • the SKIP macroblock encodes all subblocks in the macroblock using only the SKIP mode
  • the intra macroblock encodes all subblocks in the macroblock using only intra prediction.
  • the inter macroblock is encoded using inter prediction in which all subblocks in the macroblock transmit motion information or residual signal information.
  • the mixed-inter / intra macroblock is encoded using inter prediction that transmits intra prediction or motion information or residual signal information in units of subblocks within the macroblock.
  • intra macroblock there may be only an intra macroblock and an inter macroblock as a macroblock type.
  • all subblocks in the intra macroblock may be encoded using only intra prediction, and all subblocks in the inter macroblock may be selected and encoded by intra prediction or intra prediction that performs motion estimation and compensation in subblock units.
  • the encoder After encoding the segmentation information of the macroblock, the encoder encodes the macroblock type.
  • the type of the macroblock may be fixed according to the type of the picture or may be selected by comparing encoding costs.
  • all macroblock types in the picture are fixed to intra macroblocks, and only the prediction mode data indicating the intra prediction modes of the respective subblocks are encoded without encoding the macroblock type and the prediction type.
  • the macroblock type is determined according to an encoding cost, and after encoding the selected macroblock type, if there are two or more available prediction types according to the macroblock type, the prediction type of each subblock is encoded.
  • intra prediction mode information is encoded
  • motion information is encoded.
  • the method of encoding prediction mode data is encoded using the prediction mode information encoding method of the above-described embodiment.
  • a usable macroblock type according to a picture type may vary as shown in Tables 6-9.
  • the types of macroblocks of Tables 6 to 8 and the types of usable prediction types of subblocks according to the types of the macroblocks may be merely an example.
  • the macroblock type may be represented by encoding 1 bit or 2 bits. For example, after encoding a 1-bit long syntax indicating whether or not it is a SKIP macroblock, if it is not a SKIP macroblock, an additional 1-bit long syntax indicating whether it is a mixed inter-intra macroblock or an inter macroblock is encoded. .
  • the macroblock type may be represented by encoding 1 to 3 bits. For example, in Table 7, a 1-bit long syntax indicating whether or not a SKIP macroblock is encoded, for example, skip_flag, and then, when not a SKIP macroblock, a syntax indicating a macroblock type using a binary string of Table 10 is used. Can also be encoded.
  • the macroblock type of the current encoding target macroblock may be encoded using neighboring macroblocks of the current encoding target macroblock, for example, macroblock types of the upper and left macroblocks.
  • the coder may code using a code word table promised by the encoder and the decoder.
  • the macroblock type may be represented by encoding 1 bit or 3 bits.
  • a 1-bit length syntax such as skip_flag, indicating whether or not it is a SKIP macroblock, and if it is not a SKIP macroblock
  • the syntax indicating a macroblock type using a binary string of Table 11 is used.
  • the macroblock type of the current encoding target macroblock may be encoded by using macroblock types of neighboring macroblocks of the current encoding target macroblock, for example, upper and left macroblocks.
  • the coder may code using a code word table promised by the encoder and the decoder.
  • variable length encoding method of a syntax representing a macroblock type includes various binary encodings such as unary code, truncated unary code, and exponential golem code. It can be coded using Binary Coding).
  • arithmetic coding may be performed using prediction type information of a block.
  • macroblock types there are two syntaxes (skip_flag) to indicate whether or not a SKIP macroblock is used to indicate a macroblock type, and a syntax (mb_type) to indicate a macroblock type except a SKIP macroblock type when a non-SKIP macroblock is used.
  • skip_flag a syntax to indicate whether or not a SKIP macroblock is used to indicate a macroblock type
  • mb_type a syntax
  • macroblock types may be encoded using one or three or more syntaxes.
  • the prediction type information of each subblock in the macroblock is encoded.
  • a 1-bit length syntax indicating a prediction type of each subblock is encoded.
  • the prediction type may be indicated by using a 1-bit or 2-bit length syntax. For example, if there are three types of prediction types that can be used, SKIP, inter prediction, and intra prediction, first, one bit indicating whether or not SKIP is encoded, and then, if not in SKIP mode, one more bit is encoded to inter prediction. It may indicate whether the subblock is an intra predicted subblock. Alternatively, even when the available prediction type is Direct, inter prediction, or intra prediction, one bit indicating whether a direct mode is encoded may be encoded, and if not, the first bit indicating whether intra prediction or inter prediction is further encoded. have.
  • SKIP SKIP
  • Direct a type of prediction type that can be used
  • inter prediction first, one bit indicating whether SKIP or not is encoded, and if it is not in SKIP mode, one more bit is further encoded to determine whether it is a direct predicted subblock. It may indicate whether an inter predicted subblock.
  • the prediction type may be encoded by using a syntax of 1 bit to 3 bits in length. For example, if the types of prediction types available are SKIP, Direct, Inter Prediction, and Intra Prediction, first, one bit indicating whether SKIP is SKIP or not is encoded. Indicates whether or not. If it is not the SKIP mode or the direct mode, one bit may be encoded to indicate whether it is an inter predicted subblock or an intra predicted subblock.
  • the prediction type may be encoded by using a 2-bit fixed length syntax.
  • the prediction type may be encoded by using a promised table.
  • variable length encoding method of the syntax indicating the prediction type may be various binary codes such as unary code, truncated unary code, exponential golem code, etc. Coding) can be used to encode.
  • arithmetic encoding may be performed using the prediction type information of the neighboring block.
  • the prediction type may be encoded using a plurality of syntaxes.
  • three syntaxes may be used: SKIP_flag to indicate whether it is in SKIP mode or Direct_flag to indicate whether it is in direct mode or pred_type to indicate whether it is an intra block or an inter block.
  • SKIP_flag to indicate whether it is in SKIP mode
  • Direct_flag to indicate whether it is in direct mode
  • pred_type to indicate whether it is an intra block or an inter block.
  • Each syntax can also be encoded using a different method.
  • An image decoding apparatus may be configured similarly or similarly to the image decoding apparatus according to an embodiment of the present invention described above with reference to FIG. 7.
  • an image decoding apparatus decodes and reconstructs a type of a macroblock of a macroblock of an arbitrary size from a bitstream, and usable prediction type of subblocks in a macroblock according to a macroblock type. Only in two or more cases, prediction type information is extracted and decoded from the bitstream. After reconstructing the prediction type of each subblock, if the prediction type of the subblock is intra prediction, inter prediction to extract and decode intra prediction mode information from the bitstream and to include motion information or residual signal information in the bitstream. If, and extracts the motion information or the residual signal information from the bitstream to recover each subblock.
  • the decoder After decoding the segmentation information of the macroblock, the decoder extracts and extracts the macroblock type information from the bitstream using the picture type information. In this case, when the usable macroblock type according to the picture type is fixed, the macroblock type information decoding process is not performed from the bitstream.
  • macroblocks in an intra picture can use only intra prediction
  • extraction and decoding processes are not performed from a bitstream of macroblock type information and prediction type information, and the intra prediction mode information of each block is extracted and decoded into a bitstream.
  • the prediction type information is extracted from the bitstream only when there are two or more types of available prediction types determined by the macroblock type. And decrypt.
  • the prediction mode information according to the prediction type is extracted from the bitstream and decoded to recover the block.
  • a method of decoding prediction mode data is decoded using the prediction mode information decoding method of the above-described embodiment.
  • usable macroblock types according to picture types may vary as described in Tables 6 to 8 above.
  • the macroblock type may be decoded by decoding 1 bit or 2 bits. For example, after decoding a 1-bit length syntax indicating whether or not a SKIP macroblock is, and if not a SKIP macroblock, one more bit is extracted from the bitstream and decoded so that the type of the macroblock to be decoded is Mixed Inter-Intra Macro. Determine whether a block or an inter macroblock.
  • the macroblock type may be represented by decoding 1 to 3 bits. For example, in Table 7, after decoding a 1-bit length syntax, such as skip_flag, indicating whether or not it is a SKIP macroblock, and if it is not a SKIP macroblock, an additional 1 or 2 bits are additionally extracted and decoded to decode the macro. Restore the block type.
  • a macroblock type using Table 10 for the type of the macroblock other than the SKIP macroblock if the codeword reconstructed by extracting and decoding 1 bit from the bitstream is 1, it is determined as an inter macroblock. In this case, one bit from the bitstream is further extracted and decoded to determine whether it is an intra macroblock or a mixed inter-intra macroblock.
  • the macroblock type of the current decoding target macroblock may be decoded using neighboring macroblocks of the current decoding target macroblock, for example, macroblock types of the upper and left macroblocks.
  • it may be decoded using a code word table promised by an encoder and a decoder.
  • the macroblock type may be represented by decoding one or three bits.
  • a syntax indicating a macroblock type using a binary string of Table 11 is used. Can also be decoded.
  • the macroblock type of the current decoding target macroblock may be decoded using neighboring macroblocks of the current encoding target macroblock, for example, macroblock types of upper and left macroblocks.
  • it may be decoded using a code word table promised by an encoder and a decoder.
  • variable length decoding method of a syntax representing a macroblock type includes various binary codes such as unary code, truncated unary code, and exponential golem code. It can be decoded using Binary Coding).
  • arithmetic decoding may be performed using prediction type information of a block.
  • syntaxes to indicate whether or not a SKIP macroblock is used to indicate a macroblock type
  • mb_type a syntax that indicates a macroblock type except a SKIP macroblock type when it is not a SKIP macroblock type.
  • syntax is used, one or three or more syntaxes may be used to decode the macroblock type.
  • the prediction type information of each subblock in the macroblock is decoded.
  • a 1-bit length syntax indicating a prediction type of each subblock is decoded.
  • the prediction type may be decoded using syntax of 1 bit or 2 bits in length. For example, if there are three types of prediction types that can be used, SKIP, inter prediction, and intra prediction, first decode one bit indicating whether or not it is SKIP, and if not in SKIP mode, further decode one bit to inter prediction. It may be determined whether the subblock is an intra predicted subblock. Alternatively, even when the available prediction type is Direct, inter prediction, or intra prediction, one bit indicating whether a direct mode is decoded, and if it is not in direct mode, one bit indicating whether intra prediction or inter prediction is further decoded. have.
  • SKIP SKIP
  • Direct a type of prediction type that can be used
  • inter prediction first, one bit indicating whether or not SKIP is decoded, and then, if not in SKIP mode, further decoding one bit to determine whether it is a direct predicted subblock. It may be determined whether it is an inter predicted subblock.
  • the prediction type may be decoded using syntax of 1 bit to 3 bits in length. For example, if the type of prediction type available is SKIP, Direct, Inter Prediction, or Intra Prediction, first decode one bit indicating whether it is SKIP or not, and if it is not in SKIP mode, decode one more bit to determine whether it is Direct mode. Determine whether or not. If neither the SKIP mode nor the direct mode, 1 bit may be decoded to determine whether an inter predicted subblock or an intra predicted subblock.
  • the prediction type may be decoded using a 2-bit fixed length syntax.
  • the prediction type may be decoded using the promised table.
  • variable length decoding method of the syntax indicating the prediction type may include various binary codings such as unary code, truncated unary code, and exponential golem code. Can be decoded using the
  • arithmetic decoding may be performed using the prediction type information of the neighboring block.
  • the prediction type may be decoded using a plurality of syntaxes. For example, three syntaxes may be used: SKIP_flag to indicate whether it is in SKIP mode or Direct_flag to indicate whether it is in direct mode or pred_type to indicate whether it is an intra block or an inter block. Each syntax can also be decoded using a different method.
  • the prediction may be performed by using inter prediction or intra prediction selectively for each macroblock unit or each subblock within the macroblock.
  • only one of the inter prediction and intra prediction may be selected according to the prediction reference unit.
  • the prediction encoding may be performed by using, or the prediction encoding may be selectively performed by using inter prediction or intra prediction.
  • the prediction reference unit refers to a size of a block capable of selecting inter prediction or intra prediction within a macroblock. That is, the macroblock is a reference unit of encoding decoding processing, and the prediction reference unit is a unit capable of selecting a prediction method.
  • the prediction reference unit may be 32x32 or 16x16 for a 32x32 macroblock, or may be a block smaller than 16x16.
  • intra prediction or inter prediction may be selected only in units of 16 ⁇ 16 blocks, and subblocks in the reference unit use the same prediction type.
  • FIG. 8 is a block diagram schematically illustrating an example of an image encoding apparatus according to another embodiment of the present invention.
  • An example of an image encoding apparatus 800 includes a candidate prediction basic unit configuration 810, an image encoder 820, and a prediction basic unit determiner. 830 may be configured.
  • the candidate prediction reference unit setter 810 sets candidates of prediction reference units for macroblocks of any size.
  • the candidate of the prediction reference unit may be a predetermined candidate group (for example, 16x16, 32x32, etc.) input by a user or other device or may be determined according to characteristics of an image. For example, when the candidate of the prediction reference unit is determined according to the characteristics of the image, the candidate may be determined by various candidate groups in consideration of the size of the image, the aspect ratio, and the like.
  • the image encoder 820 may be implemented as an image encoding apparatus 600 according to an embodiment of the present invention described above with reference to FIG. 6.
  • the apparatus for encoding an image encodes by selectively using inter prediction or intra prediction in the corresponding prediction reference unit for each candidate of the prediction reference unit set by the candidate prediction reference unit setter 810.
  • a prediction reference unit in a macroblock may be transmitted to a decoder by encoding split information in macroblock units or may be transmitted only once in a sequence header, a picture, and a slice header. Thereafter, the prediction type information is encoded in the prediction reference unit instead of the subblock unit, and the prediction mode information is encoded in the subblock unit within the prediction reference unit.
  • the method of encoding the prediction type information and the prediction mode information according to the prediction type is encoded by using the encoding method of the encoder of FIG. 6 described above.
  • the prediction reference unit determiner 830 calculates an encoding cost (that is, an encoding cost of image data encoded for each candidate of the prediction reference unit) when the image encoder 820 encodes an image for each candidate of the prediction reference unit, and predicts each prediction.
  • the encoding cost of the image data encoded for each candidate of the reference unit is compared with each other to determine an optimal prediction reference unit among the candidates of the prediction reference unit.
  • the optimal prediction reference unit may be a candidate of a prediction reference unit having a minimum encoding cost when the image is encoded with a candidate of the prediction reference unit among candidates of each prediction reference unit, but if the encoding cost is used, It can be determined variously using.
  • the image encoder 820 encodes a macroblock using an available prediction type according to the picture type among the intra prediction mode and the inter prediction modes for each block, and compares the encoding cost to obtain an optimal prediction type and prediction mode for each block. After determining, if all subblocks in the macroblock use the intra prediction mode or the inter prediction mode, the prediction reference unit is determined as the macroblock size. If the prediction type of the subblocks in the macroblock includes intra prediction and inter prediction at the same time, intra prediction sets the maximum size of the selected subblock as a reference unit of prediction.
  • the macroblock type is encoded by setting the macroblock type to a value representing an intra macroblock, and when all subblocks in the macroblock use only inter prediction, the macroblock type is set to a value representing an inter macroblock. To encode it. If there is a block using intra prediction and inter prediction modes in the macroblock, the macroblock type is encoded by setting the macroblock type to a value representing an intra / inter macroblock, and the maximum size of the subblock in which the intra prediction is selected Can also be set.
  • the prediction reference unit determiner 830 may generate image data encoded in the prediction reference unit into a bitstream and transmit the image data to the decoder.
  • the determined prediction reference unit may be encoded only once in a sequence header or a header or slice header of each picture that is not a macroblock unit. In this case, segmentation information from a macroblock of any size to a prediction reference unit may not be encoded.
  • a reference unit of prediction may or may not be transmitted as a flag (Set_predBlockSize_flag) on whether to send the size of the reference unit of prediction to the sequence header or each picture or slice header.
  • an arbitrary size for example, a 16x16 size, may be used as a reference unit of prediction, or the prediction reference unit may be encoded by encoding split information for each macroblock.
  • the reference unit of the prediction is designated, information on the reference unit of the prediction is transmitted.
  • the horizontal size and the vertical size of the reference unit of the prediction may be set separately so that the reference unit of the prediction of any size may be used.
  • the reference unit of the prediction as a specific value, and send a value for how many times to increase or decrease from a predetermined size.
  • a log function may be applied to the size value of the prediction reference unit to represent the size value of the prediction reference unit using fewer bits. For example, log 2 (the reference of the selected prediction). Unit / X) (where X is any positive integer that is a multiple of 2), for example, if the value of X is 16, if the reference unit of the selected prediction is 16x16, encode 0 and 32x32. Code 1
  • the ratio between the width and the height may be separately encoded.
  • it may be an index value of a table promised by the encoder and the decoder.
  • the encoding may be performed using various binary coding methods such as unary code, truncated unary code, and exponential golem code.
  • the following shows an example of syntax according to the first method of the aforementioned prediction reference unit coding.
  • the contents of setting the horizontal and vertical sizes are not described, but the horizontal and vertical sizes may be set.
  • the data is encoded on the sequence header and the picture header, the encoding may be performed on the slice header.
  • the MxN size is set as the default prediction reference unit, and a flag indicating whether to use the default prediction reference unit for each picture header is encoded, and if the default prediction reference unit is not used, the reference unit of the selected prediction is encoded. .
  • a flag indicating whether to set the default prediction reference unit is encoded in the sequence header, and if the default prediction reference unit is not set, a predetermined size, for example, 16x16 size is used as the reference unit of the default prediction and the reference unit of the default prediction is set.
  • the reference unit of the default prediction may be encoded.
  • the method of encoding defalt_predBlockSize which is information indicating the size of the default prediction criterion unit, or predBlockSize, which is information indicating the size of the current prediction criterion unit, may specify the size of the actual prediction criterion unit or increase the number of times from the predetermined size. Or you can send a value to shrink.
  • the macroblock size value instead of encoding the prediction reference unit size value as it is, the macroblock size value may be represented using fewer bits by applying a log function to the prediction reference unit size value.
  • the ratio between the width and the height may be separately encoded.
  • it may be an index value of a table promised by the encoder and the decoder.
  • the encoding may be performed using various binary coding methods such as unary code, truncated unary code, and exponential golem code.
  • the following shows an example of syntax according to the second method of size coding of the aforementioned prediction reference unit.
  • the method of encoding the predBlockSize value which is information indicating the size of the current prediction reference unit, may specify the size of the actual prediction reference unit or may send a value for increasing or reducing the number of times from the predetermined size.
  • the macroblock size value instead of encoding the prediction reference unit size value as it is, may be represented using fewer bits by applying a log function to the prediction reference unit size value.
  • the ratio between the width and the height may be separately encoded.
  • it may be an index value of a table promised by the encoder and the decoder.
  • the encoding may be performed using various binary coding methods such as unary code, truncated unary code, and exponential golem code.
  • the following shows an example of syntax according to the second method of size coding of the aforementioned prediction reference unit.
  • FIG. 9 is a flowchart for explaining an example of an image encoding method according to another embodiment of the present invention.
  • the image encoding apparatus sets a candidate of the prediction reference unit, encodes the input image for each candidate of the prediction reference unit, and determines the prediction reference unit according to the encoding cost of the image data for each candidate of the prediction reference unit, and determines the determined prediction reference unit. Accordingly, a bitstream including coded image data and information about the determined prediction reference unit is generated by selectively using inter prediction or intra prediction for each subblock of the macroblock.
  • the apparatus for encoding a prediction criterion unit setting a candidate of a prediction reference unit, encoding an image for each candidate of the prediction reference unit, and determining the prediction reference unit according to the encoding cost of the image data encoded for each candidate of the prediction reference unit is illustrated in FIG. 8. As described above, detailed description thereof will be omitted.
  • FIG. 10 is a block diagram illustrating an implementation of an image decoding apparatus according to another embodiment of the present invention.
  • An example of an image decoding apparatus according to another embodiment of the present invention may be configured similarly or similarly to the image decoding apparatus 700 according to an embodiment of the present invention described above with reference to FIG. 7.
  • the image decoding apparatus may include a prediction reference unit setter and an image decoder.
  • An example of an image decoding apparatus is a prediction reference unit for selectively using inter prediction or intra prediction for a macroblock having an arbitrary size from a bitstream as well as image data encoded from a bitstream.
  • a macroblock equal to or smaller than the size of the prediction criterion unit identified by the information on the extracted prediction criterion unit (or, if necessary, the prediction criterion unit identified by decoding information on the extracted prediction criterion unit)
  • Reconstruction images are generated by decoding the encoded image data by selectively using inter prediction or intra prediction for each of the subblocks within.
  • the prediction reference unit seter extracts prediction reference unit information from the bitstream to set the prediction reference unit in the macroblock. If the encoder is promised to include the prediction reference unit information only once in the bitstream, the image may be reconstructed using the prediction reference unit extracted to decode the entire image by decoding only once in the bitstream of the entire picture, or for each picture. In the case where P is promised to encode / decode, the prediction reference unit information may be extracted for each picture from the bitstream, and the image may be reconstructed using different prediction reference units for each picture. In addition, according to another embodiment of the present invention, the information may be extracted for each picture, slice, or macroblock layer to reconstruct an image using the selected prediction reference unit.
  • the decoder When the prediction reference unit is set in the prediction reference unit setter, the decoder extracts the prediction type information from the bitstream as the prediction reference unit, and extracts and decodes the prediction mode information for each subblock according to the prediction type from the bitstream to restore the macroblock. do.
  • a flag (Set_predBlockSize _flag) is included to send information about the size of the prediction criterion unit, whether to send information about the size of the prediction criterion unit at the promised position, such as the sequence header or the header or slice header of each picture.
  • the decoded prediction criterion unit size designation flag (Set_predBlockSize _flag) indicates that the size of the prediction criterion unit is specified, Unary code, Truncated Unary-Code, and Exponential Gollum Among the various entropy decoding methods such as Exp-Golomb Code, the size of the prediction reference unit is entropy decoded and extracted by a method promised by the encoder.
  • the horizontal size and the vertical size of the prediction reference unit may be obtained by entropy decoding, or when the square prediction reference unit is used, only information indicating the length of one side may be entropy decoded. It may be.
  • the decoded value may be designated as the size of the actual prediction criterion unit, or the value obtained by increasing or reducing the decoded value several times from a predetermined size agreed with the encoder may be designated as the size of the prediction criterion unit.
  • the size of the prediction criterion unit may be set by applying an exponential function to the entropy-decoded value.
  • the encoder may log 2 (PredBlockSize / If y is encoded as a value of X) (X is any positive integer that is a multiple of 2), the decoder entropy decodes the y value and multiplies the value of X by 2 y to obtain the size of the prediction reference unit selected by the encoder. have.
  • the value of X may be a value promised by the encoder and the decoder or may be a value extracted before decoding the prediction reference unit size from the bitstream. If the value of X is 8, if the value of the decoded y is 0, the size of the prediction reference unit is set to 8x8. If the value of the decoded y is 1, the size of the prediction reference unit is set to 16x16.
  • the size of the prediction reference unit may be obtained by entropy decoding the ratio of the width and the height.
  • the size of the prediction reference unit may be obtained using the decoded value as the index value of the table.
  • the size of the prediction reference unit to be transmitted may be determined using various binary coding methods such as unary code, truncated unary code, and exponential golem code. Can be decrypted
  • the contents of decoding and extracting the horizontal and vertical sizes are not described, but the horizontal and vertical sizes may be extracted.
  • the data is decoded on the sequence header and the picture header.
  • the encoder encodes the size of the prediction reference unit in the slice header or the macroblock header, the size of the prediction reference unit may be decoded in the slice header.
  • the NxN size promised by the encoder is set as the reference prediction reference unit size, and a flag indicating whether to use the reference prediction reference unit size from each picture, slice, or macroblock header is entropy decoded.
  • the prediction reference unit size information is entropy decoded to extract related information of the prediction reference unit and to set the size of the prediction reference unit. If the decoded flag value indicates that the reference prediction unit size is used, the prediction reference unit size information is not included in the bitstream. Therefore, the size of the preset reference prediction unit is set to the size of the prediction reference unit. A series of decoding processes are performed.
  • the size of the reference prediction reference unit When the size of the reference prediction reference unit is included in the sequence header in the encoder and transmitted to the decoder, the size of the reference prediction reference unit may be extracted at a predetermined position such as a sequence header in the bitstream.
  • the size of the reference prediction reference unit or the size of the current prediction reference unit may be assigned an entropy decoded value, or the size of the prediction reference unit may be obtained by scaling (reducing or enlarging) a predetermined size with the entropy decoded value.
  • the size value of the prediction reference unit may be obtained using the exponential function.
  • the size of the prediction reference unit may be obtained by entropy decoding the ratio of the width and the height.
  • the size of the prediction reference unit may be obtained using the decoded value as the index value of the table.
  • the size of the prediction reference unit to be transmitted may be determined using various binary coding methods such as unary code, truncated unary code, and exponential golem code. Can be decrypted
  • the third method entropy-decoding the size information of the prediction reference unit after entropy decoding the size information of the prediction reference unit when not using the flag of the reference prediction reference unit size and the size of the reference prediction reference unit in the first picture
  • the prediction reference unit size required for the first picture decoding is obtained using the information.
  • the size of the prediction reference unit is set equal to the size of the reference prediction reference unit, and then the first picture is decoded.
  • the size of the prediction reference unit of the previous picture is set to the same value as the size of the prediction reference unit of the previous picture to decode the second picture.
  • the method of decoding PredBlockSize which is information indicating the size of the prediction criterion unit of the current picture, may use the entropy-decoded value as the size of the prediction criterion unit or enlarge it from a predetermined size by using the entropy-decoded value as an enlargement ratio or a reduction ratio.
  • the size of the prediction reference unit may be reduced.
  • the size value of the prediction reference unit may be obtained using the exponential function.
  • the reference prediction reference unit size may be obtained by entropy decoding the ratio of the width and the height.
  • the size of the reference prediction reference unit may be obtained using the decoded value as the index value of the table.
  • the reference prediction reference unit size is determined using various binary coding methods such as unary code, truncated unary code, and exponential golem code. Can be decrypted
  • the embodiment of the present invention generates an effect of efficiently encoding / decoding a high resolution image by encoding and decoding a subblock or a prediction reference unit in a macroblock having a variable size using various prediction modes. It is a very useful invention.

Landscapes

  • Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Compression Or Coding Systems Of Tv Signals (AREA)

Abstract

La présente invention concerne un appareil et un procédé de codage/décodage d'image utilisant des macroblocs de taille variable. L'appareil de codage/décodage d'image selon un mode de réalisation de l'invention comprend : un codeur d'image qui code des macroblocs par unité de référence de prédiction candidate, détermine une unité de référence de prédiction en fonction du coût de codage pour chaque unité de référence de prédiction candidate, et génère un flux binaire qui contient des données d'image codée dans l'unité de référence de prédiction déterminée, des informations concernant un type de prédiction pour chaque unité de référence de prédiction déterminée et des données de prédiction par type de prédiction, ainsi qu'un décodeur d'image qui extrait, à partir du flux binaire, des informations concernant le type de prédiction pour chaque unité de référence de prédiction, des données de prédiction par type de prédiction et les données d'image codée, et décode les données d'image codée par type de prédiction pour chaque unité de référence de prédiction afin de générer une image récupérée.
PCT/KR2010/006740 2009-10-01 2010-10-01 Procédé et appareil de codage/décodage d'image utilisant des macroblocs de taille variable WO2011040796A2 (fr)

Priority Applications (4)

Application Number Priority Date Filing Date Title
US13/499,406 US9549190B2 (en) 2009-10-01 2010-10-01 Method and apparatus for encoding/decoding image using variable-size macroblocks
US14/745,181 US9609335B2 (en) 2009-10-01 2015-06-19 Method and apparatus for encoding/decoding image using variable-size macroblocks
US14/745,093 US9565444B2 (en) 2009-10-01 2015-06-19 Method and apparatus for encoding/decoding image using variable-size macroblocks
US14/745,152 US9609334B2 (en) 2009-10-01 2015-06-19 Method and apparatus for encoding/decoding image using variable-size macroblocks

Applications Claiming Priority (8)

Application Number Priority Date Filing Date Title
KR20090093987 2009-10-01
KR20090093982 2009-10-01
KR10-2009-0093982 2009-10-01
KR10-2009-0093987 2009-10-01
KR20090100443 2009-10-21
KR10-2009-0100443 2009-10-21
KR10-2010-0096035 2010-10-01
KR20100096035A KR101479129B1 (ko) 2009-10-01 2010-10-01 가변 크기의 매크로블록을 이용한 영상 부호화/복호화 방법 및 장치

Related Child Applications (2)

Application Number Title Priority Date Filing Date
US13/499,406 A-371-Of-International US9549190B2 (en) 2009-10-01 2010-10-01 Method and apparatus for encoding/decoding image using variable-size macroblocks
US14/745,093 Continuation US9565444B2 (en) 2009-10-01 2015-06-19 Method and apparatus for encoding/decoding image using variable-size macroblocks

Publications (2)

Publication Number Publication Date
WO2011040796A2 true WO2011040796A2 (fr) 2011-04-07
WO2011040796A3 WO2011040796A3 (fr) 2011-09-29

Family

ID=43826820

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/KR2010/006740 WO2011040796A2 (fr) 2009-10-01 2010-10-01 Procédé et appareil de codage/décodage d'image utilisant des macroblocs de taille variable

Country Status (1)

Country Link
WO (1) WO2011040796A2 (fr)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20070029450A (ko) * 2005-09-09 2007-03-14 삼성전자주식회사 영상의 부호화 및 복호화 장치와, 그 방법, 및 이를수행하기 위한 프로그램이 기록된 기록 매체
KR20070069073A (ko) * 2005-12-27 2007-07-02 닛본 덴끼 가부시끼가이샤 부호화 데이터 출력 방법 및 장치, 재부호화 데이터 생성방법 및 장치, 부호화 데이터 복원 방법 및 장치, 부호화방법 및 장치, 및 컴퓨터 판독 가능한 기록 매체
KR20070118560A (ko) * 2006-06-12 2007-12-17 경희대학교 산학협력단 Ｆｇｓ코딩에서 우선적인 인코딩/디코딩 방법 및 장치
KR20080034131A (ko) * 2005-07-22 2008-04-18 미쓰비시덴키 가부시키가이샤 화상 부호화 장치, 화상 복호 장치, 화상 부호화 방법, 화상 복호 방법, 화상 부호화 프로그램을 기록한 컴퓨터 판독 가능한 기록 매체, 화상 복호 프로그램을 기록한 컴퓨터 판독 가능한 기록 매체

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20080034131A (ko) * 2005-07-22 2008-04-18 미쓰비시덴키 가부시키가이샤 화상 부호화 장치, 화상 복호 장치, 화상 부호화 방법, 화상 복호 방법, 화상 부호화 프로그램을 기록한 컴퓨터 판독 가능한 기록 매체, 화상 복호 프로그램을 기록한 컴퓨터 판독 가능한 기록 매체
KR20070029450A (ko) * 2005-09-09 2007-03-14 삼성전자주식회사 영상의 부호화 및 복호화 장치와, 그 방법, 및 이를수행하기 위한 프로그램이 기록된 기록 매체
KR20070069073A (ko) * 2005-12-27 2007-07-02 닛본 덴끼 가부시끼가이샤 부호화 데이터 출력 방법 및 장치, 재부호화 데이터 생성방법 및 장치, 부호화 데이터 복원 방법 및 장치, 부호화방법 및 장치, 및 컴퓨터 판독 가능한 기록 매체
KR20070118560A (ko) * 2006-06-12 2007-12-17 경희대학교 산학협력단 Ｆｇｓ코딩에서 우선적인 인코딩/디코딩 방법 및 장치

Also Published As

Publication number Publication date
WO2011040796A3 (fr) 2011-09-29

Similar Documents

Publication Publication Date Title
WO2011040794A2 (fr) Procédé et appareil de codage/décodage d'image utilisant des macroblocs de taille variable
WO2011040795A2 (fr) Procédé et appareil de codage/décodage d'image utilisant une couche divisée
WO2018097693A2 (fr) Procédé et dispositif de codage et de décodage vidéo, et support d'enregistrement à flux binaire mémorisé en son sein
WO2020076143A1 (fr) Procédé et appareil de traitement de signal vidéo utilisant la prédiction à hypothèses multiples
WO2019182385A1 (fr) Dispositif et procédé de codage/décodage d'image, et support d'enregistrement contenant un flux binaire
WO2018030773A1 (fr) Procédé et appareil destinés au codage/décodage d'image
WO2018066867A1 (fr) Procédé et appareil de codage et décodage d'image, et support d'enregistrement pour la mémorisation de flux binaire
WO2017192011A2 (fr) Procédé et appareil de codage/décodage d'image utilisant une prédiction intra-écran
WO2018097692A2 (fr) Procédé et appareil de codage/décodage d'image et support d'enregistrement contenant en mémoire un train de bits
WO2015122549A1 (fr) Procédé et appareil de traitement d'une vidéo
WO2020076116A1 (fr) Procédés et appareils de codage/décodage d'image
WO2011096741A2 (fr) Procédé et appareil permettant de coder de la vidéo sur la base d'un ordre de balayage d'unités de données hiérarchiques, et procédé et appareil permettant de décoder de la vidéo sur la base d'un ordre de balayage d'unités de données hiérarchiques
WO2017026681A1 (fr) Procédé et dispositif d'interprédiction dans un système de codage vidéo
WO2020060158A1 (fr) Procédé de codage et de décodage d'informations de mouvement et appareil de codage et de décodage d'informations de mouvement
WO2011126272A2 (fr) Procédé et appareil destinés à coder une vidéo à l'aide d'une transformation de gamme dynamique et procédé et appareil destinés à décoder une vidéo à l'aide d'une transformation de gamme dynamique
WO2012005520A2 (fr) Procédé et appareil d'encodage vidéo au moyen d'une fusion de blocs, et procédé et appareil de décodage vidéo au moyen d'une fusion de blocs
WO2019027241A1 (fr) Procédé et dispositif de codage/décodage d'images et support de mémorisation de stockage de flux binaire
WO2012096539A2 (fr) Procédés de codage et de décodage vidéo et appareils correspondants utilisant un mode de balayage sélectif
WO2014081226A1 (fr) Procédé de décodage d'image et appareil utilisant celui-ci
WO2018135885A1 (fr) Procédé de décodage et de codage d'image pour fournir un traitement de transformation
WO2013141596A1 (fr) Procédé et dispositif pour le codage vidéo échelonnable sur la base d'unité de codage de structure arborescente, et procédé et dispositif pour décoder une vidéo échelonnable sur la base d'unité de codage de structure arborescente
WO2011129672A2 (fr) Appareil et procédé de codage/décodage vidéo
WO2020111843A1 (fr) Procédé et dispositif de traitement de signal vidéo utilisant un filtrage d'intraprédiction
WO2021015524A1 (fr) Procédé et dispositif de traitement de signal vidéo
WO2020005045A1 (fr) Procédé et appareil d'inter-prédiction basée sur un mode de fusion

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 10820869

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase in:

Ref country code: DE

WWE Wipo information: entry into national phase

Ref document number: 13499406

Country of ref document: US

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS (EPO FORM 1205A DATED 06-07-2012)

122 Ep: pct application non-entry in european phase

Ref document number: 10820869

Country of ref document: EP

Kind code of ref document: A2