Method for inducing a merge candidate block and device usingsame

(54) Name: Method for inducing a merge candidate block and device using same

SUMMARY

The present invention relates to a method for determining a candidate block for rate fusion. The method comprises determining whether a boundary between a current block borders a boundary of a major coding unit; and determining a candidate block for temporal fusion attributed to the current block according to a result of said determination, wherein the candidate block for temporal fusion belongs to a decoded image and the decoding image has a different temporal order than a current image comprising the current block.

PROCEDURE TO APPOINT A CANDIDATE BLOCK FOR MERGER AND A DEVICE FOR APPLICATION OF THIS PROCEDURE

Description

Technical area

The present invention relates to a method of encoding and decoding video, more particularly to a method of selecting a candidate block for fusion, and a device for applying this method.

The prior art

More recently, demand in various areas has grown to high-resolution and high-quality video, so-called HD video and UHD video (ultra-high resolution). As the video quality and resolution grow, the amount of video becomes larger compared to existing video, which means that if video is transmitted on a medium such as an existing cable or wireless broadband network, or if it is stored on an existing storage medium, the cost of transmission / storage increases. In order to solve these problems, which are associated with increased resolution and higher quality, more efficient video compression methods are needed.

The various methods of video compression include inter (image) prediction, when the value of a pixel Included in a current image is predicted based on the previous or subsequent image, intra (image) prediction, when the value of a pixel included in a current image is predicted based on pixel information within this image, as well as an entropy coding technique where a shorter code is assigned to frequency values ​​that occur frequently and longer code is assigned to frequency values ​​that occur less frequently; and these video compression methods allow video data to be compressed efficiently for transmission or storage.

Disclosure of the Invention

Technical problem

The first object of the present invention is to provide a method for selecting a candidate for fusion by parallel processing.

The second object of the present invention is to provide a device for applying a method for selecting a candidate for fusion by parallel processing.

Technical solution

According to one aspect of the present invention, in accordance with the first of the aforementioned objects, a method for obtaining a candidate for fusion is offered. The method may include decoding information associated with a motion estimation region (MER, Motion Estimation Region); determining whether a target block for prediction and a candidate block for spatial fusion are included in the same MER; and to determine that the spatial fusion candidate block is not available if there is a fusion candidate block that does not use the spatial fusion candidate block when the prediction target and the spatial fusion candidate block are included in the same MER. The method may further include adaptive determination of a candidate block for spatial fusion based on the size of the MER and the size of the target block for prediction, if the target block for prediction and the candidate block for spatial fusion are included in the same MER. If MER has the size of 8x8 and the size of the target block for prediction is 8x4 or 4x8, then at least one of the candidate blocks for spatial fusion in the target block for prediction can be replaced with a block containing a point outside MER. The procedure may further include determining whether the candidate block for spatial fusion is included in an ERA that has not yet been decoded. The procedure may further include replacing the candidate block for spatial fusion with a block included in any other MER if the target block for prediction and the candidate block for spatial fusion are included in the same MER. The replaced candidate block for spatial fusion can be selected adaptively so that it is included in another MER than the target block for prediction, starting from the location of the replaced candidate block for spatial fusion in the same MER. The information associated with MER can be related to the size of MER and transmitted in an image unit. The determination of whether the target block for prediction and the candidate block for spatial fusion are included in the same MER may include deciding whether the target block for prediction and the candidate block for spatial fusion are included in the same MER based on a control equation based on information on where the target block for prediction is located. the candidate block for spatial fusion is located as well as how large the MER is.

According to another aspect of the present invention with respect to the second object of the above, a device for decoding images is provided. The device may comprise an entropy decoding unit which decodes information associated with a motion calculation region (MER) and a prediction unit that determines whether a target block for prediction and a candidate block for spatial fusion is included in the same MER, and determines that the candidate block for spatial fusion is not available if the target block is available. for prediction and the candidate block for spatial fusion are included in the same MER. The prediction unit may be a prediction unit that adaptively determines a candidate block for spatial fusion based on the size of the MER and the size of the target block for prediction, if the target block for prediction and candidate3 block for spatial fusion is included in the same MER. If MER has the size of 8x8 and the size of the target block for prediction is 8x4 or 4x8, then the prediction unit can replace at least one of the candidate blocks for spatial fusion in the target block for prediction with a block containing a point outside MER. The Prediction Unit can determine if the candidate block for spatial fusion is included in a MER that has not yet been decoded. The prediction unit can be a prediction unit that replaces the candidate block for spatial fusion with a block that is part of any other MER, if the target block for prediction and the candidate block for spatial fusion are included in the same MER. The replaced candidate block for spatial fusion may be a candidate block for spatial fusion that is adaptively replaced so that it is included in a MORE than the target block for prediction, starting from the location of the replaced candidate block for spatial fusion in the same MER. The information on MER can be related to the size of MER and included with an imaging unit. The prediction unit can be a prediction unit that determines whether the target block for prediction and the candidate block for spatial fusion are included in the same MER based on a control equation based on information on where the target block for prediction is located, where the candidate block for spatial fusion is located, and how large the MER is.

Advantages of the invention

With a method for selecting a candidate block for fusion and a device to which the method is applied, according to the embodiments of the present invention, parallel processing can be achieved by applying the method for selecting the candidate block for fusion, which means that the computational volume and application complexity can be reduced.

Brief drawing description

Fig. 1 is a block diagram illustrating a video encoder according to an embodiment of the present invention.

Figure 2 is a block diagram illustrating a video decoder according to another embodiment of the present invention.

Figure 3 is a schematic illustration of candidate blocks for applying a fusion mode and a skipping mode according to an embodiment of the present invention.

Figure 4 is a schematic illustration of a method of determining candidate blocks for fusion according to an embodiment of the present invention.

Fig. 5 is a schematic illustration of a method of determining a candidate block for fusion based on the size of an MER according to an embodiment of the present invention.

Figure 6 is a schematic illustration of a way to determine whether a candidate block for spatial fusion in a current block is available.

Figure 7 is a flowchart illustrating a way of selecting a candidate block for spatial fusion in fusion mode, according to an embodiment of the present invention.

Figure 8 is a flowchart illustrating a method of performing Interprediction in fusion mode, according to an embodiment of the present invention.

Disclosure of the Invention

Several modifications and embodiments are possible, but below only certain embodiments will be described in more detail with reference to the accompanying drawings. However, the present invention is not to be construed as limited to the embodiments described below, but includes all modifications, counterparts or alternatives which are within the scope and technical conditions of the invention. In all drawings, the same reference numerals refer to similar elements.

It is recognized that even if the terms first, second, etc. used to describe different elements, the elements are not limited by these terms. The terms are used only to distinguish different elements from each other. The terms are used only to distinguish different elements from each other. For example, a first element may be named a second element without departing from the content of the invention, as may a second element may be referred to as a first element The term and / or means one of the specified elements or a combination of several of the specified elements.

It is understood that when a property or element is said to be associated with or linked to another property or element, this means that the connection or connection may be direct or have intermediate elements. Where, instead, a property or element is said to be directly connected to or directly linked to another element, this implies that there are no intermediate elements.

The terminology used herein is intended for the more detailed description of the individual embodiments and is not intended to limit the scope of the invention. Singular5 forms as one, one, the, it should be perceived as covering even the plural forms, unless otherwise clearly stated in the context. While it is understood that the terms include or encompass in this specification, the stated properties, numbers, steps, measures, elements, components or combinations thereof are present, but the terms do not preclude one or more other properties, numbers, steps, measures, elements, components or combinations thereof may be added or present.

The invention will be described in more detail below with reference to the accompanying drawings. In the following, the same reference numerals are used in all the drawings to denote the same components, and the same component will not be described more than once.

Fig. 1 is a block diagram illustrating a video encoder according to an embodiment of the present invention.

Figure 1 shows that a video encoder 100 may comprise a module 110 for image splitting, a module 120 for interprediction, a module 125 for intraprediction, a module 130 for transformation, a module 135 for quantization, a module 160 for restructuring, a module 165 for entropy coding, a module 140 for quantization, a module 145 for inverse transformation, a module 150 for filtering, and a memory 155.

Each of the modules shown in Figure 1 is depicted independently to clarify various functions of the video encoder, which should not be interpreted as each module being a separate hardware unit or software component. The modules are thus listed to illustrate the functions, and at least two of the modules can be combined into one, as well as a module can be divided into several elements which together provide the function, and an embodiment in which the respective modules are combined or divided is within the scope of the claims. to the present invention, and is within the technical scope of the invention.

Some of the elements may also be redundant for the essential functions of the invention, but have been added to improve performance. The present invention can be realized with only those elements essential to achieving the basic purpose of the invention, without the involvement of elements added only to improve performance, and a configuration which includes only the essential elements and excludes the added elements which only improve performance. also within the scope of the claims of the present invention.

The image splitting module 110 can divide an incoming image into at least one processing unit. Such a processing unit may be a prediction unit (PU), a transformation unit (TU) or a coding unit (CU). The image splitting module 110 can divide an image into a combination of coding units, prediction units, and transform units, and can encode the image by selecting any combination of coding unit, prediction unit (s) and transformation unit (s), based on some predetermined criterion (e.g. a cost function).

For example, an image can be divided into multiple coding units. To divide a coding unit, one can apply a recursive tree structure, such as a quad tree, wherein a coding unit divided into other coding units, based on an image or a largest coding unit as root, can be divided into as many offspring nodes as the number of divided coding units . A coding unit that is not further subdivided due to some restriction becomes a leaf node. Thus, assuming that only a quadratic division is available for one coding unit, one can divide one coding unit into four different coding units.

In the following embodiments of the present invention, the term coding unit may refer to not only a coding unit but also a decoding unit. The prediction unit can be divided into squares or rectangles of the same size within a coding unit.

When the prediction unit for an intraprediction is created from the coding unit, the intraprediction can be performed without splitting into multiple prediction units as an NxN unit, if the coding unit is not the smallest coding unit.

The prediction module may contain the interprediction module 120 for performing an interprediction and the intraprediction module 125 for performing an intraprediction. With regard to the prediction unit, the prediction module can decide whether an interprediction or an intra-prediction should be performed, based on specific information for each prediction method (eg intra-prediction mode, motion vector, reference image, etc.). Here, the processing unit that performs the prediction and the processing unit that determines the selection of the prediction method and the specific information may be different. For example, the prediction method and the prediction mode can be determined in the prediction unit, while the prediction is performed in the transformation unit. A residual value (residual block) between a generated prediction block and an origin block can constitute input data for the transform module 130. Furthermore, information about the prediction mode, motion vector, etc. can be used. used in the prediction is encoded in the entropy coding module 135, along with the residual value to be transmitted to the decoder. When a particular coding mode is used, it is possible that the prediction block is not generated in the prediction module 120,125, but instead, the origin block is encoded because it is to be transmitted to a decoder.

The interprediction module can perform prediction for the prediction unit based on information on at least one image, sometimes images that are before or after the current image. The interprediction module may include a module for interpolating reference images, a module for motion prediction and a module for motion compensation.

The reference image interpolation module can be provided with reference image information from memory 155 and can generate pixel information from the reference image which is less than one pixel. For a luma pixel, a DCT-based 8-coefficient interpolation filter can be used, where a filter coefficient is varied to generate pixel information that is 1/4 pixel less than a full pixel. For a chrome pixel, a DCT-based 4-coefficient interpolation filter can be used, where a filter coefficient is varied to generate pixel information that is 1/8 pixel less than a full pixel.

The motion prediction module can perform motion prediction based on a reference image that has been interpolated by the reference image interpolation module. There are various methods to obtain the motion vector, such as FBMA (full search-based block matching algorithm), TSS (three step search) and NTS (new three-step search). The motion vector may have a motion vector value of 1/2 or 1/4 pixel starting from the interpolated pixel. The motion prediction module can predict a current prediction unit by switching the method of motion prediction. Various methods for performing motion prediction can be used, for example methods such as a skipping mode, a fusion mode, or an advanced motion vector prediction (AMVP) mode can be used.

According to embodiments of the present invention, the motion calculation region (MER) can be determined at the interprediction so that the interprediction can be performed in parallel. For example, when fusion mode or skipping mode is used in interprediction, it is possible to determine whether a target block for prediction and a candidate block for spatial fusion are included in the same MER, and if the target block for prediction and candidate block for spatial fusion is not included in the same MER, then the candidate block for spatial fusion can be determined. as unavailable, or a candidate block for fusion can be determined by deciding whether the candidate block for spatial fusion is included in a MER that has not yet been decoded. The following describes how the prediction unit works when an interprediction is performed, according to exemplary embodiments of the present invention.

The interprediction unit can generate the prediction unit based on information about reference pixels adjacent to a current block, the reference pixels being included in the current image. If an adjacent block in the current prediction unit is a block for which interprediction is performed, with reference pixels being the pixels for which interprediction is performed, then the reference pixels contained in the interprediction block can be replaced by the reference pixels in adjacent blocks for which intraprediction is performed. Therefore, if a reference pixel is missing, missing reference pixels can be replaced by at least one reference pixel that is available.

In intra-prediction, different direction-dependent prediction modes can be used, using information from the reference pixels, starting from a prediction direction, while applying direction-independent modes that do not utilize direction information in the prediction. A prediction mode for information about luma samples and a prediction mode for information about chrome samples need not be the same. Furthermore, information about the selected intraprediction mode for lumbar samples or information about predicted lumar signals can be used to predict information about chromium samples.

In a case where the size of the prediction unit and the size of the transformation unit are equal in intraprediction, the intraprediction can be performed on the prediction unit based on left-to-left pixels at the top left and pixels at the top of the prediction unit. In a case where the size of the prediction unit and the size of the transform unit are different in intraprediction, instead the intraprediction can be performed based on the reference pixels based on the transform unit. Furthermore, intraprediction that utilizes division into NxN can only be used for the smallest coding unit.

In the intra-prediction method, depending on the prediction mode, a mode-dependent equalization filter (MDIS) can be used for the reference pixels to generate the prediction block. The type of MDIS filters for the reference pixels may be different. In the intra-prediction mode, the intra-prediction mode of the current prediction unit can be predicted from the intra-prediction mode of the prediction unit adjacent to the current prediction unit. When predicting the prediction mode of the current prediction unit by using mod information for an adjacent prediction unit, if the intra-prediction modes of the current prediction unit and the adjacent prediction unit are the same, then information about the prediction modes of the current prediction unit and the adjacent prediction unit may be the same. using a predetermined flag information, and if the intra-prediction modes of the current prediction unit and the adjacent prediction unit are different, the prediction mode of the current block can be decoded by entropy coding.

Further, a residual block with residual value information is obtained, which is the difference between the prediction unit on which the prediction is performed, the prediction unit generated in the prediction module 120 or 125, and the origin block in the prediction unit. The generated residual block can be fed into the transform module 130. The transform module 130 can transform the residual block with the residual value from the initial block and the prediction unit generated in the prediction module 120 or 125, using a transformation method such as discrete cosine transform (DCT) or discrete sinus transform (DST). Whether DCT or DST should be used to transform the residual block can be determined based on information about the intraprediction mode of the prediction unit used to generate the residual block.

The quantization module 135 can quantize values ​​transformed into a frequency domain by the transform module 130. Depending on the block or the meaning of the image, the quantization parameter may be varied. A value output from the quantization module 135 may be input to the quantization module 140 and the restructuring module 160.

The restructuring module 160 can restructure the quantized coefficient value in relation to the residual value.

The restructuring module 160 can modify a coefficient in a two-dimensional block-shaped matrix to become a one-dimensional vector, using a coefficient calculation method. For example, by diagonal scanning of coefficients that lie between direct current and the high frequency range, the restructuring module 160 can restructure them into a one-dimensional vector. Depending on the size of the transformer unit and the intraprediction mode, vertical scanning of two-dimensional block shape in the column direction, or a horizontal scan of two-dimensional block shape in the row direction, can be used instead of diagonal scanning. In other words, one can decide which scanning mode - diagonal scanning, vertical scanning or horizontal scanning - to use, based on the size of the transformer unit and the intraprediction mode used.

Entropy coding module 165 performs entropy coding based on values ​​derived from restructuring module 160. Entropy coding can be performed using various coding methods, such as Exponential Golomb or Context-Adaptive Binary Arithmetic Coding (CABAC).

Entropy coding unit 165 can encode various types of information, such as residual coefficient for the coding unit, block type, prediction mode, partition unit, prediction unit, transfer unit, motion vector, reference image, block interpolation, filter, MORE, etc., derived from the restructuring module 160 and predlecton modules 120 and 125.

Entropy coding unit 165 can perform the entropy coding of the coefficient value in the coding unit derived from the restructuring module 160, using an entropy coding method, for example CABAC.

The quantization module 140 and the inverse transformation module 145 quantize values ​​quantized by the quantization module 135, and inversely transform the values ​​transformed by the transform module 130. The residual value generated by the quantization module 140 and the inverse transform module 145 can be added to the prediction unit predicted by the motion calculation module, the motion compensation module and the intraprediction module of the prediction module 120 and 125, thereby obtaining a reconstructed block.

The filtration module 150 may contain at least one of a deblock filter, an offset correction module, and an adaptive loop filter (ALF).

The unblocking filter can remove a block distortion caused by a block boundary in a reconstructed image. To determine if unblocking filtering is needed, it is possible to determine whether the unblocking filter should be applied to the current block based on the pixels contained in certain columns or rows included in the block. When applying the unblocking filter to the block, a strong filter or a weak filter can be applied, depending on the filter strength required. In unblocking filtration in the vertical and horizontal directions, the filtering can take place in parallel.

The offset correction module can correct an offset from an original image with a pixel in relation to the image for which unblocking filtering is performed. To perform offset correction in relation to a particular image, one can use a method that includes classifying pixels in the image to a preselected number of regions, determining which region the offset should apply to, and applying the offset to the corresponding region, or a method comprising apply offset based on edge information for each pixel.

The adaptive loop filter (ALF) can perform filtration based on a comparison between the filtered and reconstructed image and the original image, respectively. When pixels included in car11 it has been classified into a preselected group and a filter has been established to be used for the corresponding group, the filtering can be performed on the respective group with differentiation for different filters. Information on whether to apply ALF can be transmitted by the coding unit (CU) and the size and coefficient of ALF may be different for each block. ALF can have different shape, which means that different coefficients in the filter can be different for different filters. ALF information relevant to the filtering (filter coefficient, ALF On / Off, filter form, etc.) can be included and transmitted within a preselected parameter set in a bit stream.

The memory 155 may store a reconstructed block or image coming from the filter module 150, and the stored reconstructed block or image may proceed to the prediction module 120,125 when interprediction is to be performed.

Figure 2 is a block diagram illustrating an image decoder according to another embodiment of the present invention.

As can be seen in Figure 2, a video decoder may comprise a module 210 for entropy decoding, a module 215 for restructuring, a module 220 for quantization, a module 225 for inversion transformation, a module 230, 235 for prediction, a module 240 for filtering, and a memory 245. .

When a bit stream of video data arrives from the video encoder, this bit stream can be decoded in the opposite order to the processing in the video encoder.

Entropy decoding module 210 may perform entropy decoding in the opposite order to the entropy coding in the video encoder's entropy coding module. Information for generating the prediction block in the information decoded by the entropy decoding module 210 can pass to the prediction module 230, 235, and the residual values ​​that have been entropy decoded in the entropy decoding module can pass to the restructuring module 215.

The entropy decoding module 210 can decode information associated with the intraprediction and interprediction performed by the encoder. As described above, if a predetermined restriction exists in the video encoder for intraprediction and interprediction, the information associated with intraprediction and interprediction for the block in question can be obtained by entropy decoding based on the restriction.

The restructuring module 215 can perform restructuring of the bit flow entropy decoded by the entropy decoding module 210, starting from the restructuring method in encoder12. Coefficients that exist as a one-dimensional vector can be reconstructed and restructured into two-dimensional block shape.

The quantization module 220 can perform quantization based on the quantization parameter coming from the encoder and the restructured coefficient block.

The inverse transform module 225 can perform an inverse DCT and an inverse DST on the result of the quantization done by the video encoder, starting from the DCT and DST performed by the transform module. The inverse transformation can be performed from the transmission unit determined by the video encoder. In the video encoder transform module, DCT and DST can be selectively executed based on various information, such as the prediction method, the size of the current block and the prediction direction, while the video decoder module 225 for inverse transformation can perform inverse transformation based on information about the transformation in the video encoder's transform module.

The prediction module 230, 235 may generate the prediction block based on information associated with the generation of the prediction block and originate from the entropy decoding module 210, as well as information on the nearest preceding decoded block or image coming from memory 245.

The prediction module 230, 235 may include a prediction unit module, an interprediction module, and an intraprediction module. The prediction unit module can receive various information, such as prediction unit, intra-prediction mode for the inter-prediction method, and motion prediction for the inter-prediction method from the entropy decoder, and can distinguish the prediction unit in the current coding unit from the received information and decide whether to predict the interprediction on the prediction.

The interpreter can perform interprediction for the current prediction unit, based on information contained in at least one image between the preceding images and the images that follow the current image, including the current prediction unit, based on information required for interprediction of the current prediction unit and which is obtained from the video encoder.

To perform the interprediction, it can be determined, from the coding unit, whether the motion prediction method of the prediction unit contained in a corresponding coding unit is in skipping mode, fusion mode or AMVP mode.

According to an embodiment of the present invention, the motion calculation region (MER) in interprediction can be delimited for parallel prediction. For example, when fusion mode or skip mode is used in interprediction, it can be determined whether a target block for prediction and a candidate block for spatial fusion are included in the same MER. When the target block for prediction and the candidate block for spatial fusion are not included in the same MER, the candidate block for spatial fusion can be set as unavailable or it can be set as a candidate block for fusion, if it is part of a MER that has not yet been decoded. The following describes in more detail an embodiment of the present invention for how the prediction module works.

The intra-prediction module can generate a prediction block based on pixel information in the current image. When the prediction unit is an intra-prediction prediction unit, intra-prediction can be performed based on intra-prediction mode information for the prediction unit obtained from the video encoder. The intraprediction module may include the MDIS filter, a module for interpolating reference pixels and a DC filter. The MDIS filter is a module for filtering reference pixels in the current block, and the prediction mode of the current prediction unit may decide whether to use the filter. The filtering can be performed on reference pixels in the current block with the prediction mode used for the prediction unit, using the MDIS filter information obtained from the video encoder. When the prediction mode for the current block is such that no filtering is to be performed, the MDIS filter is not needed.

The reference pixel interpolation module can generate a reference pixel in a pixel unit smaller than an integer, by interpolating the reference pixel when the prediction unit prediction mode is used for intraprediction based on a pixel value of the interpolated reference pixel. When the prediction mode of the current prediction unit generates is a prediction mode that generates the prediction block without interpolation of the reference pixel, no such interpolation of the reference pixel is needed. The DC filter can generate the prediction block by filtering if the prediction mode of the current block is DCmod.

The reconstructed block or image may proceed to the filter module 240. The filter module 240 may contain a deblock filter, an offset correction module, and an adaptive loop filter (ALF).

Information on whether unblocking filtering is performed on a particular block or image, and if a strong or weak filter is used in this case, can be obtained from the video encoder. The unblocking filter in the video decoder can receive information about the unblocking filter in the video encoder, and can perform unblocking filtering of the block in the video decoder. In the same way as in the video encoder, first a vertical and a horizontal unblocking filtering is performed, and at least one vertical and horizontal unblocking can be performed in an overlapping area. In the overlapping area for vertical and horizontal unblocking filtration, the vertical or horizontal unblocking filtration can be performed that has not been performed previously. By this technique for unblocking filtration, parallel unblocking filtration can be carried out.

The offset correction module can perform offset correction on the reconstructed image, based on the offset correction performed on the image and the offset value in question.

ALF can perform filtration based on a value obtained when comparing the filtered and reconstructed image to the original image. ALF can be used for the coding unit based on information from the decoder about whether to use ALF and ALF coefficient. The ALF information can be included in a specific set of parameters that is communicated.

The memory 245 may contain the reconstructed image or blocks to be used as a reference image or reference block, and the reconstructed image may proceed to the output module.

In the above description, a coding unit has been designated a coding unit in one embodiment, but it may refer to a unit for both coding and decoding. Below, a prediction method illustrated by Figures 3-11 will be described, according to an embodiment of the invention, which can be used by, for example, the prediction module depicted in Figures 1 and 2.

Figure 3 is a schematic illustration of candidate blocks for application of fusion mode and skipping mode according to an embodiment of the present invention.

Below, for illustrative purposes, fusion mode will be described in an embodiment of the invention; however, the same method can be used for skipping mode, and such an embodiment is within the scope of the claims of the present invention.

Figure 3 shows that interprediction in fusion mode can be accomplished using the candidate blocks for spatial fusion 300, 305, 310, 315 and 320, and the candidate blocks for temporal fusion 350 and 355.

If a dot (xP, yP) is located on the upper left of the prediction block, the prediction block having the width nPSW and the height sPSH, each of the candidate blocks for spatial fusion 300, 305, 310, 315 and 320 can be any of a first block 300 with a point (xP-1, yP + nPSH-MinPuSize), a second block 305 with a point (xP + nPSW-MinPuSize, yP-1), a third block 310 with a point (xP + nPSW, yP-1) , a fourth block 315 with a dot (xP-1, yP + nPSH), and a fifth block 320 with a dot (xP-MinPuSize, yP-1).

The temporal fusion candidate block may use multiple candidate blocks, and a first Col block (colocalized block) 350 may be a block with a dot (xP + nPSW, yP + nPSH) located on a Col image (colocalized image). If the first Col block 350 is missing or inaccessible (eg if the first Col block does not perform any interprediction), a second Col block 355 with a point (xP + (nPSW »1), yP +) may instead be used. (nPSH »1)) in the Col image.

According to an embodiment of the present invention, it is possible to determine whether a candidate block for fusion should be used in relation to a particular area, to perform parallel interprediction in fusion mode in motion prediction. For example, in order to determine candidate blocks for fusion in fusion mode, in relation to a managed area of ​​a certain size, it is possible to determine whether the candidate block for fusion is within the default area together with the target block for prediction and based on this it can decide whether the candidate block for fusion should be used or replaced with another candidate block for fusion, and thereby carry out the movement prediction in parallel with the chosen area. In the following, for an embodiment of the present invention, a method of parallel motion prediction in fusion mode is described.

Figure 4 is a schematic illustration of a method of determining candidate blocks for fusion according to an embodiment of the present invention.

Figure 4 shows an assumed case where a largest coding unit (LCU) is divided into four regions for motion calculation (MER).

For a first prediction block PUO, included in a first MER (MER0), as shown in Figure 4, there may be five candidate blocks for spatial fusion 400, 405, 410, 415 and 420 when interpredicting in fusion mode for the first prediction block PUO. for fusion 400,405,410,415 and 420 may exist at sites not included in the first MER (MER0), and may be blocks for which encoding / decoding has already occurred.

The second prediction block (PU1) is a prediction block included in a second MER (MER1) and four candidate blocks for fusion 430, 435, 445 and 450 of the spatial fusion candidate blocks 430, 435, 440,445 and 450 for interprediction in fusion mode can be blocks that is located in the second MER (MER1) and blocks belonging to the MER where prediction is currently performed. The remaining candidate bio for fusion 440 may be a block to the right of the current MER and a block in a LOU or MER where coding / decoding has not yet been performed.

According to one embodiment of the present invention, if the candidate fusion merge in the current block and the current block is contained in the same MER, then the merger candidate merger in the current block is excluded, and movement information for at least one block elsewhere can be added as a candidate for merger, based on the size of the current block and MER.

A block containing a point located in another MER in the vertical or horizontal direction can be added as a candidate block for fusion. Alternatively, a block included in another MER at a location closest to the candidate bio may be added as a candidate block for fusion. Alternatively, a block at a preselected location based on the shape and size of the current block can be added as a candidate block for fusion.

For example, for the candidate bio for fusion 435, which is at the top of the second prediction unit (PU1), and for the candidate bio for fusion 450, located at the top left of the second prediction unit, instead of blocks 455 and 460, which contain points outside the other MORE in vertical direction, used as replacement candidate block for fusion. For the candidate bio for fusion 430, located to the left of the second prediction unit, and for the candidate bio for fusion 445, located on the lower left of the second prediction unit, blocks 465 and 470 containing points outside MER in the horizontal direction may be used instead. as a replacement candidate block for merger. When a block is part of the same MER as the current prediction unit and therefore cannot be used as a candidate block for fusion, the candidate candidate bio for fusion can be replaced with another block containing a point in another MER, based on where the candidate candidate bio for fusion is located.

For a third prediction block (PU2), a candidate block for fusion 475, located in the same MER as the third prediction block, can be replaced with a block 480 located at the top just above it. Furthermore, according to one embodiment of the present invention, the location of the candidate block for spatial fusion can be replaced by a block included in another MER in a direction other than vertical or horizontal, and this embodiment is also within the scope of the claims of the present invention.

The following steps can be performed to apply a procedure for determining candidate blocks for fusion.

1) To decode information related to a business calculation region (MER). The information related to MER can contain information about the size of MER. Whether the target block for prediction is included in the MER can be determined based on information about the size of the MER and the size of the target block for prediction.

2) To determine whether the target block for prediction and the candidate block for spatial fusion are included in the same MER.

If the target block for prediction and the candidate block for spatial fusion are included in the same MER, then the following steps can be taken to adaptively determine the candidate block for spatial fusion, based on the size of MER and the size of the target block for prediction.

3) To determine that the candidate block for spatial fusion is not available if the target block for prediction and the candidate block for spatial fusion are included in the same MER.

The candidate block for spatial fusion can be determined as unavailable if the target block for prediction and the candidate block for spatial fusion are included in the same MER, and the candidate block for spatial fusion included in the same MER can be replaced by another candidate block for fusion. Furthermore, as described below, it may occur that the candidate block for fusion that is determined to be unavailable cannot be used in fusion prediction. According to another embodiment of the present invention, a method may be used which does not utilize the candidate block for fusion contained in the same MER as the target block for prediction.

For example, among the candidate blocks for fusion are those available for parallel interprediction in fusion mode that is part of an MER where coding / decoding has already been performed, and not in an MER where current prediction is performed. The blocks can be used as candidate blocks for interprediction in fusion mode. However, blocks included in the MER where prediction is performed cannot be used as candidate blocks for interprediction in fusion mode. Also, the block for which coding / decoding has not been performed cannot be used as a candidate block for interprediction. The exemplary embodiment is also within the scope of the claims of the present invention.

Figure 5 is a schematic illustration of a method for determining candidate blocks for fusion according to an embodiment of the present invention, starting from the size of an MER.

Figure 5 shows that the candidate for fusion can be determined adaptively based on the MER size and the size of the current prediction unit. In a case where a candidate for fusion corresponds to one of the places A, B, C, D and E for candidates for fusion and is included in the same MER as the prediction unit in question, the candidate for fusion is determined as inaccessible. Here, motion information for at least one block at another location can be added as a candidate for fusion, based on the size of the current block and MER.

In Figure 5, it is assumed that the size of the MER is 8x8 and the size of the target block for prediction is 4x8. If MER has the size 8x8, a block A that is part of the target block for prediction belongs to both the same MER and the same target block for prediction, while blocks B, C, D and E are included in another MER than the target block for prediction.

This block A can be replaced by a block (eg block A ') that is part of another MER. According to one embodiment of the present invention, therefore, if the candidate block for fusion in the current block and the current block is contained in the same MER, then the candidate block for fusion in the current block can be excluded, so that motion information for at least one block at another location can be added. as candidate for merger, based on the size of the current block and MER.

According to an embodiment of the present invention, the information on the size of the MER can be included in the higher level syntax information to be transmitted.

Table 1 below is a method for transmitting information about the size of MER in the higher-level syntax.

Table 1

<td>pic_parameter_set_rbsp () {</td><td>descriptor</td>

<td>pic_para_meter_set_id</td><td>ue (v)</td>

<td>seq_parameter_set_id</td><td>ue (v)</td>

<td>entropy_coding_mode_flag</td><td>u (1)</td>

<td>num_temporal_layer_switching_point_flags</td><td>ue (v)</td>

<td>for (i = 0; i &lt;num_temporaljayer_switching_point_flags: i-)</td><td></td>

<td>temporal_layer_switching_point_flag [i]</td><td>u (D</td>

<td>num_ref_idxJ0_default_active_minus1</td><td>ue (v)</td>

<td>num_ref_i dxJ1 _defa ult_acti ve_mi nus 1</td><td>ue (v)</td>

<td>pic_init_qp_minus26 / * relative to 26 * /</td><td>see (v)</td>

<td>constrained_intra_pred_flag</td><td>u (l)</td>

<td>shared _pps_info_enabled_flag</td><td>u (l)</td>

<td>if (shared_pps_info_enabled_flag)</td><td></td>

<td>if (adaptive_loop_filter_enabled_flag)</td><td></td>

<td>alf_param ()</td><td></td>

<td>if (cu_qp_delta_enabled_flag)</td><td></td>

<td>max_cu_qp_delta_depth</td><td>u (4)</td>

<td>Iog2_parallel_merge_level_minus2</td><td>ue (v)</td>

<td>rbsp_trailing_bits ()</td><td></td>

<td> }</td><td></td>

Table 1 shows that information on the size of the MER can be obtained from a syntax element Iog2_parallel_merge_level_minus2 that is part of a higher level syntax structure, for example a set of image parameters. A syntax element

Iog2_parallel_merge_level_minus2 may also be included in another syntax structure for higher level than the set of image parameters, and this embodiment is also within the scope of the claims of the present invention.

Table 2 below describes the relationship between a value of io Iog2_parallel_merge_level_minus2 and the size of MER.

Table 2

<td>Iog2_&gt; parallel_merge_level_minus2</td><td>MORE- size</td><td>Comment</td>

<td> 0</td><td>4x4</td><td>Sequential fusion / skipping mode for everyone PU within an LCI, because of the minimum PU size which is allowed according to HEVC is 4x4</td>

<td> 1</td><td>8x8</td><td>Parallel search in fusion / skipping mode</td>

<td></td><td></td><td>allowed for all PU within an 8x8 block</td>

<td> 2</td><td>16x16</td><td>Parallel search in fusion / skipping mode allowed for all PU within a 16x16 block</td>

<td> 3</td><td>32x32</td><td>Parallel search in fusion / skipping mode allowed for all PU within a 32x32 block</td>

<td> 4</td><td>64x64</td><td>Parallel search in fusion / skipping mode allowed for all PU within a 64x64 block</td>

Table 2 shows that the value of Iog2_parallel_merge_level_minus2 may have a value in the closed range 0 to 4, and the size of MER can be specified otherwise based on the value of the syntax element. If MER is 0, this means that the interprediction is performed in fusion mode without using MER.

The syntax element with information on the size of MER can in one embodiment of the present invention be the expression MER size information syntax element and defining the information syntax element for MER size in the manner given in Table 2 is an example. It is possible to specify the MER size in different ways, and such a way of expressing the syntax element is also within the scope of the claims of the present invention.

Figure 6 is a schematic illustration of a way to determine whether a candidate block for spatial fusion in a current block is available.

According to Figure 6, it is possible to determine whether a candidate block for spatial fusion is available based on the locations of a target block for prediction 600 and a candidate block for spatial fusion 650 adjacent to the target block for prediction 600 plus the syntax element with information on the size of MER.

Assuming that (xP, yP) is a point at the top left of the target block for prediction and that (xN, yN) is a point at the top left of the candidate block for fusion, one can determine whether the candidate block for spatial fusion is available using of the following equations Math 1 and Math 2.

Math 1:

(xP »(log2_parallI_inerge_kvel_ininus2 + 2)) == (xN» (log2_panilkl_iiieige_kveI_uiiiius2 + 2))

Math 2:

(yP (log2_par; illel_merge_level_ininus2 + 2)) == (yN »(log2_ |&gt; arallel_nierge_level_niinus2 + 2))

Math 1 and Math 2 above are examples of equations that can be used to determine whether the candidate block for fusion and the target block for prediction are included in the same MER. If one wishes to determine whether the candidate block for fusion and the target block for prediction are included in the same MER, one can also use a different method than the one described above, as long as it does not deviate from the core of the present invention.

Figure 7 is a flowchart illustrating a way of selecting a candidate block for spatial fusion in fusion mode, according to an embodiment of the present invention.

Figure 7 shows how MER-related information is decoded (step S700).

The MER-related information may be syntax element information as described above, and may be included in the higher-level syntax structure. It is possible to determine whether the target block for prediction and the candidate block for spatial fusion are included in the same or different MER based on the decoded MER-related information.

It is determined whether the target block for prediction and the candidate block for spatial fusion are included in the same MER (step S710).

According to an embodiment of the present invention, if the candidate block for fusion in the current block and the current block is contained in the same MER, then the candidate block for fusion in the current block can be excluded, and movement information for at least one block elsewhere can be added as a candidate. for fusion, based on the size of the current block and MER (step S720). According to another embodiment of the present invention, instead of using the candidate block for spatial fusion contained in the same MER as the target block for prediction, a block in another MER can replace the candidate block for fusion for performing the interprediction.

Further, according to another embodiment of the present invention, when the spatial fusion candidate block and the target block for prediction are included in the same MER, the candidate block for spatial fusion that is included in the same MER need not be used as the target block for prediction.

When the candidate block for spatial fusion and the target block for prediction are not included in the same MER, the interprediction is performed based on the corresponding candidate block for spatial fusion (step S730).

Figure 8 is a flowchart illustrating a method of performing interprediction in fusion mode, according to an embodiment of the present invention.

In Figure 8, the motion prediction information is obtained from the spatial fusion candidate (step S800).

The candidate for spatial fusion can be selected from the adjacent prediction unit in the prediction target block. To select the candidate for spatial fusion, you can retrieve information about the width and height of the prediction unit, information about MER, information about the singleMCLFlag, and information about the location of the partitioned. spatial fusion is available, if the reference image (refldxLO, refldxLI), about list usage (predFlagLON, predFlagLIN), and about the motion vector (mvLON, mvL1N), for the location where the candidate for spatial fusion is located. The candidate for spatial fusion may be several blocks adjacent to the target block for prediction.

According to one embodiment of the present invention, the spatial fusion candidate block can be classified into three classes: 1) a spatial fusion candidate block that is not included in the same MER and has already been coded or decoded; 2) a candidate block for spatial fusion included in the same MER; and 3) a spatial fusion candidate block for which coding and decoding has not yet been performed.

According to an exemplary embodiment of the present invention, to perform parallel interprediction in an MER unit, one can select as the spatial fusion candidate block from among the spatial fusion candidate blocks available for interprediction the spatial fusion candidate block not included in the same MER and already coded or decoded. Furthermore, as a candidate block for spatial fusion, you can use the candidate block for spatial fusion that replaces a location for the candidate block for spatial fusion included in the same MER. In other words, according to one embodiment of the present invention, if the candidate block for fusion in the current block and the current block is included in the same MER, then the candidate block for fusion in the current block can be excluded, and movement information for at least one block elsewhere can be added. as candidate for merger, based on the size of the current block and MER. As described above, one can apply a method for determining the candidate block for merger by the step of decoding information associated with the Motion Estimation Region, the step of determining whether the target block of prediction and the candidate block of merger are in the same MER, and the step of determining that the candidate block for fusion is not available for interprediction in fusion mode if the candidate block for fusion is in the same MER as the target block for prediction.

According to another embodiment of the present invention, in order to perform interprediction as candidate blocks for fusion, one can select from among the candidate blocks for spatial fusion available for interprediction the candidate blocks for spatial fusion not included in the same MER and already coded or decoded.

A reference image index value is obtained for the candidate for temporal fusion (step S810).

The value of the reference image index for the candidate for temporal fusion is an index value for the Col image containing the candidate for temporal fusion (Col block) and can be obtained via a special condition as below. For example, if the target block for prediction has a point (xP, yP) at the top left, width nPSW and height nPSH, then the value of the reference image index for the candidate for temporal fusion can be determined at the same value as the reference image index for the adjacent prediction unit (hereinafter referred to as the adjacent prediction unit). derivation of reference image index) if 1) there is an adjacent prediction unit in the prediction target block corresponding to the site (xP-1, yP + nPSH-1); 2) a partition index value for adjacent prediction unit for reference image index derivation is 0; 3) adjacent prediction unit for deriving reference image index is not a block which performs prediction in intraprediction mode; and 4) the target block for prediction and adjacent prediction unit for derivation of the reference image index is not included in the same MER (Motion Estimation Region). If these conditions are not met, the reference image index of the candidate for temporal fusion can be set to 0.

The candidate for temporal fusion is determined, and the information for movement prediction is derived from the candidate for temporal fusion (step S820).

To determine the candidate block for temporal fusion (Col block) and extract the information for motion prediction based on the designated candidate block for temporal fusion (Col block), one can determine a location for the Col block used to extract a motion vector for temporal prediction. , based on such conditions as, for example, whether the Col block is available for the target block for prediction, or where a location for the target block for prediction is relative to the LCU (e.g. whether the location of the prediction target block is at the bottom edge or right edge relative to the LCU). By extracting the information for motion prediction based on the established information for the reference image in the Col block and the information for the motion vector for prediction, one can extract the information for motion prediction from the candidate block for temporal fusion (Col block).

A list of candidates for merger is prepared (step S830).

The list of candidates for fusion can be prepared by including at least one candidate for spatial fusion and the candidate for temporal fusion. The candidate for spatial fusion and the candidate for temporal fusion that are on the list of candidates for fusion can be arranged with a fixed priority.

The list of candidates for merger can be drawn up by including a fixed number of candidates for merger. If there are not enough candidates for fusion to create the fixed number of candidates, a candidate can be generated by combining the motion prediction information for the candidate for fusion, or the list of candidates for fusion can be generated by adding a zero vector that candidate for merger.

As described above, the method can be used to select a candidate for fusion not only for interprediction between image frames in fusion mode, but also for intraprediction within an image frame in skipping mode, and this embodiment also falls within the scope of the claims of the present invention.

Although the present invention has been described with reference to exemplary embodiments, those skilled in the art will recognize that various changes and variations thereof may be made without departing from the present invention and the scope of the claims below.

Embodiments

1st A method for selecting a candidate for fusion, comprising the following steps: to decode information associated with a motion calculating region (MER): determining whether a target block for prediction and a candidate block for spatial fusion is included in the same MER; and to determine that the candidate block for spatial fusion is an inaccessible candidate block for fusion if the target block for prediction and the candidate block for spatial fusion are included in the same MER.

2nd Method according to Embodiment 1, also comprising adaptively determining a candidate block for spatial fusion based on the size of the MER and the size of the target block for prediction, if the target block for prediction and the candidate block for spatial fusion are included in the same MER.

3rd Method according to embodiment 2, comprising replacing at least one of the candidate blocks for spatial fusion in the target block for prediction with a block containing a point outside the MER, if the size of the MER is 8x8 and the size of the target block for prediction is 8x4 or 4x8.

4th Method according to Embodiment 1, also comprising determining whether the candidate block for spatial fusion is included in a MER that has not yet been decoded.

5th Method according to Embodiment 1, also comprising replacing the candidate block for spatial fusion with a block included in any other MER if the target block for prediction and the candidate block for spatial fusion are included in the same MER.

6th Method according to Embodiment 5, comprising the spatial fusion substituted candidate block is a spatial fusion candidate block that has been adaptively replaced so that it will be included in a MORE than the target block for prediction, starting from the location of the spatial fusion candidate block contained in the same MORE.

7th Method according to Embodiment 1, wherein said information related to MER relates to the size of MER and is transmitted in an image unit.

8.

Method according to embodiment 1, wherein the step of determining whether the target block for prediction and the candidate block for spatial fusion is included in the same MER includes determining whether the target block for prediction and the candidate block for spatial fusion is included in the same MER based on a control equation based on information on where the target block for prediction is located, where the candidate block for spatial fusion is located and how large the MER is.

9th Device for decoding video, comprising:

an entropy decoding module that decodes information associated with a motion calculation region (MER); and a prediction module that determines whether a target block for prediction and a candidate block for spatial fusion are included in the same MER; and determines that the candidate block for spatial fusion is an inaccessible candidate block for fusion if the target block for prediction and the candidate block for spatial fusion are included in the same MER.

10th Device for decoding video according to embodiment 9, wherein said prediction module adaptively determines a candidate block for spatial fusion based on the size of the MER and the size of the target block for prediction, if the target block for prediction and the candidate block for spatial fusion are included in the same MER.

11th Device for decoding video according to embodiment 10, wherein said prediction module replaces at least one of the candidate blocks for spatial fusion in the target block for prediction with a block containing a point outside MER, if MER has the size 8x8 and the size of the target block for prediction is 8x4 or 4x8.

12th Device for decoding video according to embodiment 9, wherein said prediction module determines whether the candidate block for spatial fusion is part of a MER that has not yet been decoded.

13th Device for decoding video according to embodiment 9, wherein said prediction module replaces the candidate block for spatial fusion with a block included in another MER, if the target block for prediction and the candidate block for spatial fusion are included in the same MER.

14th Device for decoding video according to embodiment 13, where the replaced spatial fusion candidate block is a spatial fusion candidate block that is adaptively replaced so that it is part of a MORE than the target block for prediction, starting from the location of the spatial fusion candidate block included in the the same MORE.

15th Device for decoding according to embodiment 9, wherein said information associated with MER has to do with the size of MER and is transmitted in an image unit.

16th Device for decoding video according to embodiment 9, wherein said prediction module determines whether the target block for prediction and the candidate block for spatial fusion are included in the same MER based on a control equation based on information on where the target block for prediction is located, and the candidate block for spatial fusion is located and how big MORE is.