KR20100122450A - A method and an apparatus for processing a video signal - Google Patents

Abstract

If the macro block type of the current macro block is intra mixed, the block type identification information of the partition of the current macro block is obtained by specifying block type identification information of the partition of the current macro block, and the block type of the partition of the current macro block is specified. The current macroblock may be predicted by obtaining an intra prediction mode of the current macroblock according to the block type. In the case of macro blocks larger than 16 × 16, the number of bits required for coding the macro block type can be reduced by adaptively using various block types, thereby improving coding efficiency.

Description

Method and apparatus for processing video signal {A METHOD AND AN APPARATUS FOR PROCESSING A VIDEO SIGNAL}

The present invention relates to a method and apparatus for intra or inter prediction and block unit coding.

In order to increase the compression rate of an image signal, a method of removing temporal redundancy and spatial redundancy, that is, intra or inter prediction is performed, and techniques such as orthogonal transform and vector quantization are used.

An object of the present invention is to efficiently process a video signal using intra prediction.

Another object of the present invention is to reduce the number of bits required for coding an area not belonging to an output area.

In order to achieve the above object, the present invention provides a method for defining a macroblock type of an I slice when a macroblock larger than 16 × 16 and obtaining a size and an intra prediction mode of a prediction block of the macroblock accordingly.

The present invention defines a macroblock type of the boundary macroblock according to the slice type, and proposes a method for determining the size of the prediction block and the intra prediction mode for the boundary macroblock of the I slice, and the boundary macro of the P slice or the B slice. For the block, an inter prediction method using a skip mode is presented.

According to the present invention, as the size of a video image increases, the number of bits required for coding a macroblock type can be reduced by adaptively using a macroblock larger than 16 × 16, thereby improving coding efficiency. Even in the case of macroblocks of I slices, the prediction accuracy can be improved by coding in units of sub-macroblocks, and at the same time, the size of the prediction block of the macroblocks is determined based on the transform size information. By reducing the number of bits, coding efficiency can be improved. By deciding or determining the macro block type, intra prediction mode, motion information, etc. for the boundary macro block without separately transmitting, the number of bits required for coding the macro block type, intra prediction mode, motion information, etc. can be reduced. have.

1 is a schematic block diagram of a video signal decoding apparatus.
2 illustrates an apparatus for decoding a macro block using macro block type and transform size information.
3 is a block diagram illustrating a process of decoding a macro block using macro block type and transform size information.
4 is a table showing sub macroblock types of P macroblock and B macroblock when the macroblock is 32x32.
5 illustrates an embodiment of an intra macro block type when the macro block is 32x32.
6 is a block diagram illustrating a method of decoding a sub macroblock using macroblock type information and block type indication information.
7 is a block diagram illustrating an embodiment of a method of determining a sub macroblock type using block type indication information.
8 is a block diagram illustrating a method of obtaining an intra prediction mode of a macro block when the macro block is coded in the intra mode.
9 illustrates a boundary macro block and a boundary sub macro block.
FIG. 10 is a block diagram illustrating a method of decoding a sub macro block in a boundary macro block in the case of an I slice.
11 illustrates one embodiment of a method for setting an intra prediction mode of a non-boundary sub macro block.
12 is a block diagram illustrating a method of decoding a sub macro block within a boundary macro block in the case of a P slice or a B slice.
FIG. 13 illustrates a method of determining a macroblock type according to a position of a boundary macroblock in a P slice or a B slice.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. Prior to this, terms or words used in the specification and claims should not be construed as having a conventional or dictionary meaning, and the inventors should properly explain the concept of terms in order to best explain their own invention. Based on the principle that can be defined, it should be interpreted as meaning and concept corresponding to the technical idea of the present invention. Therefore, the embodiments described in the specification and the drawings shown in the drawings are only the most preferred embodiment of the present invention and do not represent all of the technical idea of the present invention, various modifications that can be replaced at the time of the present application It should be understood that there may be equivalents and variations.

First, the macro block will be described. The macro block is a basic unit for processing a video frame and may be defined as an area composed of 16 × 16 pixels. However, as the size of the video image increases, it may be defined as an area larger than 16 × 16. For example, the macro block may be defined as an area consisting of 32x32 or 64x64 pixels. In this case, a macroblock larger than 16 × 16 may be a basic unit for processing a video frame having a larger resolution than that of the prior art.

Macro blocks larger than 16 × 16 may be adaptively used depending on the resolution of the video image. For example, when the resolution of the video image is less than or equal to VGA, the size of the macro block may be set to 16x16. Accordingly, blocks of size 16x8, 8x16, 8x8, 4x8, 8x4, and 4x4 may be used as partitions of the macroblock. If the resolution is VGA or higher than 1080P or less, the size of the macro block can be set to 32x32. Thus, 32x16, 16x32, 16x16, 16x8, 8x16, and 8x8 sized blocks can be used, and if the resolution is 1080P or higher and 4Kx2K or lower, The size of can be set to 64x64, and 64x32, 32x64, 32x32, 32x16, 16x32, and 16x16 blocks can be used. However, the encoder may determine an optimal macroblock size among 16x16 macroblocks or macroblocks larger than 16x16 according to the resolution of the video image. The determined size information of the macroblock may be transmitted to the decoder at the slice level or the sequence level, and thus the size of the macroblock may be adaptively used.

The macro block type according to the size of the macro block will be described. The macroblock type may be defined according to the prediction mode of the macroblock, that is, whether it is intra mode or inter mode. The macro block is predicted in units of a predetermined size, and the size information of the predicted block may be used as the same meaning as the unit of the predetermined size. The macro block type may include size information of the prediction block, and the macro block type for performing inter prediction may also include information about a reference direction.

Specifically, the macro block coded in the intra mode may be coded in block units of 16x16 pixels, 8x8 pixels, and 4x4 pixels. The macro block coded in inter mode is used as the macro block type for 16x16 pixels, 16x8 pixels, 8x16 pixels and 8x8 pixels for motion compensation, and 8x4 pixels, 4x8 pixels, and 4x4 pixels are used as sub macro block types when 8x8 pixels are used. Can be.

Hereinafter, a macro block type according to a macro block larger than 16 × 16 will be defined. As the size of the video image increases, the relatively small size of the macroblock type will reduce the frequency of use. Therefore, information about the macroblock type can be removed, thereby reducing the number of bits allocated to the macroblock type. Can be. For example, if the resolution of the video image is less than or equal to VGA, the macro block type described above may be used. However, if the resolution is VGA or more and 1080P or less, a macroblock type smaller than 8x8 pixels may not be used. That is, the macro block coded in the intra mode may be coded in block units of 16x16 pixels and 8x8 pixels, and in the macro block coded in the inter mode, 16x16 pixels, 16x8 pixels, and 8x16 pixels 8x8 pixels may be used as the macro block type. have. In this case, 8x4 pixels, 4x8 pixels, and 4x4 pixels may not be used as the sub macroblock type. If the resolution is 1080P or more and 4Kx2K or less, a macroblock type smaller than 16x16 pixels may not be used. That is, 16x16 pixels may be used as a macroblock type for blocks coded in an intra mode and blocks coded in an inter mode.

Similarly, the encoder can determine the optimal macroblock type according to the resolution of the video image, and the determined macroblock type information can be transmitted to the decoder at the slice level or the sequence level, thereby adaptively using the macroblock type. have.

1 is a schematic block diagram of a video signal decoding apparatus.

Referring to FIG. 1, the video signal decoding apparatus of the present invention includes an entropy decoding unit 110, an inverse quantization unit 120, an inverse transform unit 125, a deblocking filtering unit 130, a decoded picture storage unit 140, The inter prediction unit 150 and the intra prediction unit 160 are included.

The entropy coding unit 110 entropy decodes the video signal bitstream to extract transform coefficients, motion vectors, reference picture indexes, and the like of each macro block. The inverse quantization unit 120 inverse quantizes the entropy decoded transform coefficient, and the inverse transform unit 125 restores the original pixel value by using the inverse quantized transform coefficient. The deblocking filtering unit 130 is applied to each coded macro block to reduce block distortion. The filtered picture is stored in the decoded picture storage unit 140 for output or use as a reference picture. The inter prediction unit 150 may predict the current picture by using the reference picture stored in the decoded picture storage unit 140 and the inter prediction information (reference picture index, motion vector, etc.) received from the entropy decoding unit 110. In detail, a motion vector of a block adjacent to the current block (hereinafter, referred to as a neighboring block) may be extracted from the video signal, and a motion vector prediction value of the current block may be obtained from the neighboring block. The neighboring block may include a block on a left side, an upper side, and an upper right side of the current block. For example, the median of each of the horizontal and vertical components of the motion vector of the neighboring blocks may be set to obtain a motion vector prediction value of the current block. Alternatively, when the left block of the current block has a prediction block coded in at least one inter mode, the prediction block may be set to obtain a prediction value of the motion vector of the current block by using the motion vector of the prediction block located at the top thereof. When the upper block of the current block has a prediction block coded in at least one inter mode, the prediction block may be set to obtain a prediction value of the motion vector of the current block by using the motion vector of the prediction block located at the leftmost of them. If the upper or upper right block of the neighboring block is located outside the boundary of the picture or slice, the motion vector of the left block may be set as a predicted value of the motion vector of the current block. If there is only one block having the same reference picture index as the current block among neighboring blocks, the motion vector of the block may be set to be used for prediction.

A prediction value of the motion vector may be obtained based on a partition of the macro block according to the macro block type. For example, when the partition of the current macroblock according to the current macroblock type is 8x16, if the index of the partition of the current macroblock is 0, the left block of the current macroblock, if the index of the current block is 1, It can be set to use the upper right block of the current block for prediction. If the partition of the macro block according to the current macro block type is 16x8, if the index of the partition of the macro block is 0, the upper block of the current block, if the index of the partition of the macro block is 1, the left block of the current block Can be set to use for prediction.

In obtaining the motion vector prediction value of the current block, various motion vectors extractable from the neighboring block may be defined as candidates for the motion vector prediction value. As described above, the candidate of the motion vector prediction value may include a motion vector of one of the neighboring blocks, a median or an average value of the motion vectors of the neighboring blocks, and the motion in consideration of the frequency of the motion vector of the neighboring block. The candidate of the predicted value of the vector may be determined, but is not limited thereto. Accordingly, the encoder may select the most efficient motion vector as the motion vector prediction value of the current block among candidates of the motion vector prediction value, and transmit the selected motion vector information to the decoder. The selected motion vector information may be obtained at a slice level or a macro block level. Candidates of the motion vector prediction value may be defined using an index, and the selected motion vector information may include the index.

By encoding only the difference vector between the motion vector of the current block and the motion vector prediction value, the amount of bits to be encoded can be reduced. The motion of the current block is compensated by using the obtained motion vector prediction value and the difference vector extracted from the video signal. In addition, such motion compensation may be performed using one reference picture or may be performed using a plurality of pictures.

For example, when the macroblock is 16x16, a total of seven partitions up to 16x16, 16x8, 8x16, 8x8, 8x4, 4x8, and 4x4 may be used for inter prediction. These may be displayed hierarchically by macro block type or sub macro block type. In more detail, the size of the prediction block may be represented by a macro block type, and one of 16x16, 16x8, 8x16, and 8x8 may be selected. If 8x8 is selected, one of 8x8, 8x4, 4x8, and 4x4 partitions may be selected as the sub macroblock type.

A reference picture index and a motion vector may be coded for each prediction block. Alternatively, the reference picture index may be coded for each macro block and the motion vector for each prediction block. In the sub macro block, the coding scheme may be determined in the same manner as the macro block. It should be understood that the above-described content related to inter prediction may be equally applied to the following inter prediction.

The skip macro block may mean a macro block coded in a skip mode, and when reconstructing the current macro block, the skip macro block does not have information about the current macro block, that is, motion vector, reference picture index, and residual data, The pixel value of the macro block in the previously coded reference picture may be defined as it is. In skip macroblock coding, motion compensation using a motion vector may be involved, and the motion vector may be derived using a neighboring block. The method of deriving the motion vector from the neighboring block may apply the same method of the motion vector derivation described above. Hereinafter, a method of deriving a motion vector and a reference picture index of a B skip macro block will be described. A picture having the smallest reference index among List1 reference pictures may be defined as an anchor picture. The block of the anchor picture at the same spatial location as the current block may be defined as an anchor block.

For example, the motion information of the current block may be predicted using the motion information of the anchor block. The motion vector in the List0 direction of the anchor block may be defined as mvCol. At this time, if there is no motion vector in the List0 direction of the anchor block and there is a motion vector in the List1 direction, the motion vector in the List1 direction may be mvCol. Here, in the case of the B picture, any two sheets may be used as reference pictures, and the prediction used at this time is called List0 prediction or List1 prediction. For example, List0 prediction may mean prediction for the forward direction, and List1 prediction may mean prediction for the backward direction.

The List0 reference picture of the current block becomes the picture referenced by the mvCol, and the List1 reference picture becomes the anchor picture. If the anchor block does not have motion information (eg, coded in an intra mode), the size of the motion vector is 0, and the reference picture index of the List0 reference picture is -1.

Further, the motion vector mvL0 of List0 and the motion vector mvL1 of List1 can be obtained from the motion vector mvCol. Assuming that the velocity of motion is constant between the reference picture, the current picture, and the anchor picture, the sizes of mvL0 and mvL1 are proportional to the time intervals of the reference picture, the current picture, and the anchor picture. Therefore, mvL0 and mvL1 can be obtained using the distance between the reference picture and the current picture and the distance between the reference picture and the anchor picture. In detail, when the List0 reference index of the anchor block refers to the long-term reference picture, it may be obtained as in Equation 1 below.

When the List0 reference index of the anchor block does not refer to the long-term reference picture, it can be obtained as in Equation 2 below.

Meanwhile, the motion information of the current block may be predicted using the motion information correlation in the spatial direction. In this case, the anchor picture and the anchor block may be used to derive motion information.

First, the prediction mode of the current slice is identified. For example, when the slice type of the current slice is a B slice, the B skip mode may be used. In addition, motion information of neighboring blocks of the current block may be obtained. For example, if the block on the left of the current block is the neighbor block A, the block on the upper side is the neighbor block B, and the block on the upper right is the neighbor block C, motion information of the neighbor blocks A, B, and C is obtained. can do. The reference picture index of the List0 or List1 direction of the current block may be derived using the motion information of the neighboring blocks. The first variable may be derived based on the reference picture index of the current block. The first variable may mean a variable used to predict the motion vector of the current block with an arbitrary value. For example, the reference picture index for the List0 or List1 direction of the current block may be derived to the smallest value among the reference picture indexes of the neighboring blocks. In this case, Equation 3 below may be used.

As a concrete example, MinPositive (0,1) = 0. That is, if two valid indexes exist, a small value can be obtained. Or MinPositive (-1,0) = 0. That is, when one valid index exists, a large value that is a valid index value can be obtained. Or MinPositive (-1, -1) = -1. That is, when both are invalid indexes, for example, when two neighboring blocks are both coded in an intra mode or two neighboring blocks are not available, a large value of -1 is obtained. As a result, there must be no valid value for the result to be invalid.

First, the first variable may be set to 0 as an initial value of the first variable. In addition, when the derived reference picture indexes for the List0 or List1 direction are all less than zero, the reference picture index for the List0 or List1 direction of the current block may be set to zero. The first variable may be set to a value indicating that a reference picture of the current block does not exist. In this case, when the derived reference picture indices for the List0 or List1 directions are all less than 0, for example, when the neighboring block is a block coded in an intra mode or the neighboring block becomes unavailable. It may mean a case. In this case, by setting the first variable to 1, the motion vector of the current block can be set to 0.

Since three neighboring blocks are used in the present embodiment, when all three neighboring blocks are blocks coded or unavailable in intra mode, the reference picture index of the current block may have an invalid value of -1. . Therefore, when all of the reference picture indices of the current block are invalid, the first variable may be set to one. That is, the first variable may be set to 1 when there is no reference picture to be used by the current block.

The second variable may be derived by using motion information of the anchor block in the anchor picture. The second variable may mean a variable used to predict the motion vector of the current block with an arbitrary value. For example, when the motion information of the anchor block satisfies a predetermined condition, the second variable may be set to one. When the second variable is set to 1, the motion vector for the List0 or List1 direction of the current block can be set to 0. The constant condition may be as follows. First, the picture having the smallest reference index among the reference pictures for the List1 direction should be a short-term reference picture. Second, the reference index of the reference picture of the anchor block should be zero. Third, the horizontal component and vertical component size of the motion vector of the anchor block should be ± 1 pixel or less. As such, when all of the above predetermined conditions are satisfied, the motion vector of the current block is set to 0 by determining that the image is close to the image having little motion.

The motion information of the current block may be predicted based on the derived first variable and the second variable. For example, when the first variable is set to 1, the motion vector for List0 or List1 direction of the current block may be set to 0. In addition, when the second variable is set to 1, the motion vector for the List0 or List1 direction of the current block can be set to 0. Here, setting to 0 or 1 is only an example, and the first variable or the second variable may be set to another constant value.

If the predetermined condition is not satisfied, motion information of the current block may be derived from motion information of neighboring blocks in the current picture. For example, in the case of a motion vector, the motion vector derivation method of the skip macro block may be equally applied.

The reference picture index may calculate a minimum value among the reference picture indexes of the motion vectors located at the left, the top, and the upper right, and may use this as the reference picture index of the current block. It should be understood that the contents related to the skip mode described above may be equally applied to the case where the current block is coded in the skip mode.

The intra predictor 160 may perform intra prediction by using the reconstructed pixel in the current picture. The reconstructed pixel in the current picture may be a pixel to which deblocking filtering is not applied. The predicted current picture and residuals from the inverse transform unit are added to reconstruct the original picture. Looking at the intra mode in detail, 4x4 pixels, 8x8 pixels, and 16x16 pixels may be used as a block unit for performing intra prediction in H.264 / AVC. Intra 4x4 prediction is based on the current intra 4x4 block, using the right 4 pixels in the left block, the bottom 4 pixels in the top block, the bottom right 1 pixel in the top left block, and the bottom left 1 pixel in the top right block. Obtain the predicted value of. In this case, the intra prediction mode selects one of the most suitable prediction directions out of nine prediction directions in H.264 / AVC for each 4x4 pixel block, and performs prediction in units of 4x4 pixels based on the selected prediction direction. Intra 8x8 prediction may perform prediction in the same manner as the intra 4x4. Intra 16x16 prediction performs prediction by selecting one of four methods: vertical prediction, horizontal prediction, average prediction, and plane prediction from the lower 16 pixels of the upper macro block and the right 16 pixels of the left macro block based on the current macro block. .

Hereinafter, a detailed method of determining the macro block type described above will be described with reference to FIGS. 2, 3, and 4.

2 illustrates an apparatus for decoding a macro block using macro block type and transform size information.

Referring to FIG. 2, the macro block type determiner 200 may obtain a macro block type of a macro block currently decoded from a bitstream transmitted from an encoder. Hereinafter, a case in which a macroblock is 32x32 will be described with reference to FIG. 3.

3 is a block diagram illustrating a process of decoding a macro block using macro block type and transform size information.

Referring to FIG. 3, skip macro block indication information on a macro block may be obtained (S300). The skip mode indication information may be information indicating whether a macro block is coded in a skip mode. For example, when the skip mode indication information is 1, it may mean that the macro block is coded in a skip mode. When the skip mode indication information is 0, it may mean that the macro block is not coded in a skip mode. have. When the macroblock is not coded in the skip mode according to the skip mode indication information, a macroblock type may be obtained (S310). In the case of an I slice, the macro block may be intra predicted in blocks of 32x32 pixels, 16x16 pixels, or 8x8 pixels according to the macroblock type.

Alternatively, like the P macroblock and the B macroblock, a sub macroblock type may be defined, and the submacroblock type may be obtained based on the macroblock type (S320).

For example, when the macroblock type is 16 × 16, a sub macroblock type may be obtained. The sub macroblock type may include prediction mode information and prediction block size information. Therefore, according to the sub macroblock type, the sub macroblock may be intra predicted in units of blocks of 16x16 pixels, 8x8 pixels, or 4x4 pixels. The sub macroblock type may include only prediction mode information. That is, the sub macroblock type may indicate only whether the sub macroblock is predicted in the intra mode. Hereinafter, when the sub macroblock type includes only prediction mode information, a method of obtaining size information of the prediction block of the sub macroblock and the intra prediction mode will be described.

The intra predictor 210 may include a prediction block size determiner 220 and an intra prediction mode acquirer 230. When the prediction block size determiner 220 indicates that the sub macroblock is predicted in the intra mode according to the sub macroblock type, the prediction block size determiner 220 may determine the size information of the prediction block by using transform size information. The intra prediction mode obtaining unit 230 may obtain the intra prediction mode of the sub macro block from the bitstream transmitted from the encoder (S330).

For example, when the sub macro block is predicted in the intra mode according to the sub macro block type, transform size information about the sub macro block may be obtained. The transform size information may be information indicating a block unit for performing transform coefficient decoding and image generation. When the transform size information indicates to perform transform coefficient decoding and image generation in units of 4x4 pixel blocks, the size information of the prediction block of the sub macroblock may be determined to be 4x4 pixels, and the sub macroblock may be determined to be 4x4 pixels. Intra prediction may be performed on a block basis. The same scheme can be applied to the case where the converted size information is 8x8 pixels or 16x16 pixels.

When the sub macro block is predicted to the intra mode according to the sub macro block type, an intra prediction mode of the sub macro block may be obtained based on intra prediction mode reference information. The intra prediction mode reference information may refer to information indicating whether an intra prediction mode of the prediction block is derived from an intra prediction mode of a neighboring block based on the prediction block.

For example, when the sub macro block is predicted in the intra mode according to the sub macro block type, the intra prediction mode reference information may be obtained. When the intra prediction mode reference information indicates that the intra prediction mode of the prediction block is derived from the intra prediction mode of the neighboring block, the intra prediction mode in the sub-macro block may be used by using the intra prediction modes of the left block and the right block with respect to the prediction block. The intra prediction mode for the prediction block can be derived. The derived intra prediction mode may be derived as a minimum value between the intra prediction mode of the left block and the intra prediction mode of the right block. When the intra prediction mode reference information indicates that the intra prediction mode of the prediction block is not derived from the intra prediction mode of the neighboring block, the intra prediction mode for the prediction block in the sub macro block may be parsed. The above-described method of obtaining the size information of the prediction block and the intra prediction mode may be equally applied to the block coded in the intra mode.

The sub macroblock may be predicted using the size information of the prediction block, the intra prediction mode, and pixel values in the neighboring block of the sub macroblock (S350). The reconstruction unit 250 may reconstruct the macro block by adding the predicted macro block and the residual transmitted from the encoder. The residual may be defined as a difference value between the pixel value of the original macro block and the pixel value of the predicted macro block.

When the sub macroblock is coded in the inter mode according to the sub macro block type information, the inter predictor 240 may acquire motion information of the sub macro block (S340).

For example, the sub macro block type may be defined as shown in FIG. 4. FIG. 4 is a table illustrating an embodiment of a sub macroblock type of a P macroblock and a B macroblock when the macroblock is 32x32. Referring to FIG. 4, when the sub macroblock type is P_L0_16X8, the sub macroblock may have partitions of two submacroblocks and may be inter predicted in units of blocks of 16 × 8 pixels. That is, a motion vector and a reference picture index may be obtained for each partition of the sub macroblock. The sub macroblock may be predicted using the motion vector and the reference picture index (S350). The reconstruction unit 250 may reconstruct the macro block by adding the predicted macro block and the residual transmitted from the encoder.

When the macroblock is coded in the skip mode according to the skip mode indication information, the macroblock may be predicted by deriving motion information from the reference picture and the reference block (S350).

Hereinafter, a macro block type for a macro block larger than 16 × 16 will be described with reference to FIG. 5.

5 illustrates an embodiment of a macro block type when the macro block is 32x32.

Referring to FIG. 5A, when the macroblock is 32x32, the macroblock may be intra predicted in units of 32x32 pixels, 16x16 pixels, or 8x8 pixels. When the macroblock type is intra 32x32, one of four modes of vertical prediction, horizontal prediction, average prediction, and plane prediction is selected from the lower 32 pixels of the upper macro block and the right 32 pixels of the left macro block based on the macro block. Can be obtained and predicted. If the macro block type is intra 16x16, the 16 pixels on the right side in the left block, the bottom 16 pixels in the top block, the bottom right 1 pixel in the top left block, and the bottom left 1 pixel in the top right block are based on the 16x16 blocks in the macro block. Can be used to obtain the prediction value of 16x16 pixels. In this case, the intra prediction mode may acquire one of the most suitable intra prediction modes of every 16 × 16 pixel block among the nine modes in H.264 / AVC. The prediction may be performed in units of 16x16 pixels based on the obtained intra prediction mode. If the macroblock type is intra 8x8, it can be predicted in the same manner as the intra 16x16 described above.

Referring to FIG. 5B, when the macroblock is 32x32, the macroblock may be intra predicted in blocks of 4x4 pixels other than 32x32 pixels, 16x16 pixels, or 8x8 pixels, and in this case, the same as the intra 16x16 described above. It can be predicted by the method.

Referring to FIG. 5C, when the macroblock is 32x32, the macroblock type may include intra mixed in addition to intra 32x32, intra 16x16, or intra 8x8. When the macro block type is intra mixed, block type information about a partition of the macro block may be obtained, and accordingly, intra prediction may be adaptively performed in an intra 8x8 or intra 4x4 block unit. The partition of the macro block means a block in the macro block divided into a predetermined size, which may mean a sub macro block.

Referring to FIG. 5 (d), another embodiment of intra_mixed will be described. If the macroblock is 32x32, intra 32x32 and intra mixed can be defined as the macroblock type of the I slice. When the macro block type is intra mixed, intra prediction may be adaptively performed in blocks of intra 16x16, intra 8x8, or intra 4x4 with respect to a partition of the macroblock. Therefore, unlike in FIG. 5C, intra 16x16 and intra 8x8 may not be used as the macro block type.

Hereinafter, when a macro block type of an I slice is defined as intra NxN (N is an integer greater than 16) and intra mixed, a method of decoding a macro block will be described with reference to FIGS. 6 to 8.

6 is a block diagram illustrating a method of decoding a macro block using macro block type and block type identification information.

A macroblock type for the current macroblock may be acquired in the macroblock layer (S600). If the macro block type is intra mixed, block type identification information may be obtained with respect to a partition of the current macro block (S610). The partition of the macro block means a block in the macro block divided into a predetermined size. For example, when the macroblock is 32x32, the partition of the macroblock may be set to four 16x16 blocks. The block type for the partition of the current macroblock may be determined using the block type identification information (S620), which will be described with reference to FIG. 7.

FIG. 7 is a block diagram illustrating an embodiment of a method of determining a block type for a partition of a macro block using block type identification information.

Referring to FIG. 7, first block type identification information may be obtained for a partition of a macro block (S700). The first block type identification information may mean information indicating whether a block type for a partition of the macroblock is intra 4x4. When the block type of the partition of the macroblock is intra 4x4 according to the first block type identification information, the partition of the macroblock may be intra predicted in units of blocks of 4x4 pixels (S720). If the block type of the partition of the macroblock is not intra 4x4 according to the first block type information, second block type identification information may be obtained (S710). The second block type identification information may refer to information indicating whether a block type of a partition of the macroblock is intra 8x8. When the block type of the partition of the macroblock is intra 8x8 according to the second block type identification information, the partition of the macroblock may be intra predicted in units of blocks of 8x8 pixels (S740). If the block type of the partition of the macroblock is not intra 8x8 according to the second block type identification information, the partition of the macroblock may be intra predicted in units of blocks of 16x16 pixels (S730).

Referring to FIG. 6, an intra prediction mode for a partition of the current macroblock may be obtained (S630). As described above, the intra prediction mode for the partition of the current macroblock may be derived from the intra prediction mode of the neighboring block or parsed from the bitstream based on the intra prediction mode reference information. For example, when the partition of the current macroblock is 16x16 and the block type for the partition of the current macroblock is intra 4x4, the intra prediction mode may be acquired in units of 4x4 blocks in the partition of the current macroblock.

However, when the macroblock type of the current macroblock is not an intra mix, the current macroblock may be predicted based on the macroblock type (S650). In this case, the macro block type may include an intra prediction mode and CBP information. Accordingly, the pixel value of the current macroblock may be predicted using the intra prediction mode according to the macroblock type and the pixels of the neighboring block. The neighboring block may mean a block previously restored to the current macro block. For example, it may mean a block located on the left side and a block located on the upper side based on the current macro block. The pixel of the neighboring block may be a reconstructed pixel to which deblocking filtering is not applied.

However, when the macro block type does not include the intra prediction mode and the CBP information, the intra prediction mode and the block pattern information may be separately obtained, which will be described with reference to FIG. 8.

8 is a block diagram illustrating a method of obtaining an intra prediction mode of a macro block when the macro block is coded in the intra mode.

Referring to FIG. 8, a macroblock type of a current macroblock may be obtained (S800). When the current macroblock is predicted to the intra mode according to the macroblock type, intra prediction mode reference information about the current macroblock may be obtained (S810). For example, when the current macroblock is predicted to be an intra mode, the macroblock type may be intra NxN or intra mixed. The intra prediction mode reference information may be obtained in units of prediction blocks of the current macro block. When the intra prediction mode reference information indicates that the intra prediction mode of the prediction block is derived from the intra prediction mode of the neighboring block, the intra prediction mode of the left block and the right block based on the prediction block is used for the prediction block. An intra prediction mode may be derived (S820). The derived intra prediction mode may be derived as a minimum value between the intra prediction mode of the left block and the intra prediction mode of the right block. When the intra prediction mode reference information indicates that the intra prediction mode of the prediction block is not derived from the intra prediction mode of the neighboring block, the intra prediction mode for the prediction block may be parsed from the bitstream (S840). A pixel of the prediction block may be predicted using the intra prediction mode and the pixels of the neighboring block (S830). The neighboring block may mean a block previously restored to the prediction block. For example, it may mean a block located on the left side and a block located on the upper side based on the prediction block. The pixel of the neighboring block may be a reconstructed pixel to which deblocking filtering is not applied.

When the current macro block is predicted in the inter mode according to the macro block type, motion information on the current macro block may be obtained (S850), and the current macro block may be predicted in the inter mode using the motion information. (S860).

Hereinafter, a method of determining a macroblock type of the macroblock when the macroblock corresponds to a boundary macroblock will be described.

9 illustrates a boundary macro block and a boundary sub macro block.

Referring to FIG. 9, a coded region may refer to a frame output by a decoding process. The output region may refer to a rectangular region specified within the coding region, and may be narrowly viewed as an area actually displayed in the coding region. The boundary macro block may mean a macro block including an area that does not belong to the output area, and the non-bound macro block may mean a macro block including only an area belonging to the output area. The boundary sub macro block is a sub macro block in the boundary macro block and may mean a sub macro block including an area belonging to the output area. On the contrary, the non-boundary sub macro block may refer to the remaining sub macro block except for the boundary sub macro block in the boundary macro block. That is, the sub macro block in the boundary macro block may mean a sub macro block that does not include an area belonging to the output area.

FIG. 10 is a block diagram illustrating a method of decoding a sub macro block in a boundary macro block in the case of an I slice.

Referring to FIG. 10, frame offset information of an image sequence may be obtained (S1000). The frame offset information is samples in a picture of an image sequence and may mean information for specifying samples belonging to a specific region in a frame. Alternatively, this may mean information for specifying a sample belonging to the output area. The boundary macroblock in the frame may be specified using the frame offset information (S1010). When the current macroblock corresponds to a boundary macroblock according to the frame offset information, the macroblock type of the current macroblock may be set to intra mixed (S1020). In this case, it may be specified whether a partition of the current macroblock is a boundary submacroblock based on the frame offset information. If the partition of the current macro block is a boundary sub macro block, as shown in FIG. 7, the boundary sub macro block type may be determined using block type identification information (S1050). An intra prediction mode of the boundary sub macro block may be obtained (S1060). The pixel value of the boundary sub macroblock may be predicted using the intra prediction mode and pixel values in the neighboring block of the boundary sub macroblock (S1040). However, when the current sub macro block is a non-boundary sub macro block, the current sub macro block may be intra predicted in units of blocks of a predetermined size. For example, when the current macroblock is 32x32, the block type of the current sub macroblock may be set to intra 16x16. In addition, CBP (Coded Block Pattern) information indicating whether the current sub-macroblock has a luminance coefficient may be set to zero.

On the other hand, when the current sub-macroblock is a non-boundary sub-macroblock, the intra prediction mode of the current sub-macroblock is not obtained separately, and among the nine intra prediction modes based on the position information of the current sub-macroblock. An efficient mode may be set as an intra prediction mode of the current sub macro block (S1030). The current sub macro block may be located at the left, right, top, bottom, top left, bottom left, top right, or bottom right in the current frame, and considers a neighboring block of the current sub macro block according to the position. Intra prediction mode may be determined. The neighboring block is a block adjacent to the current sub macro block and may mean a block having a pixel value reconstructed prior to the current sub macro block. A detailed embodiment will be described with reference to FIG. 11. A pixel value of the non-boundary sub macroblock may be predicted using the intra prediction mode and pixel values in a neighboring block of the non-boundary sub macroblock (S1040).

11 illustrates one embodiment of a method for setting an intra prediction mode of a non-boundary sub macro block.

Referring to FIG. 11A, when the non-boundary sub macroblock is located at the right side of the frame, the left sub-macro block of the non-boundary sub macroblock has a reconstructed pixel value, and thus, nine intra prediction modes. It may be efficient to predict the non-boundary sub macroblock using horizontal prediction.

Referring to FIG. 11B, when the non-boundary sub macroblock is located at the bottom of the frame, it may be efficient to predict the non-boundary sub macroblock using vertical prediction among nine intra prediction modes.

Referring to FIG. 11C, when the non-boundary sub macroblock is located at the lower right side of the frame, the non-boundary sub macroblock may be predicted using vertical prediction and horizontal prediction. That is, the non-boundary sub-macroblock having an index of 1 has a boundary submacroblock having an index of 0 on the left side, so that horizontal prediction may be set to an intra prediction mode. A non-boundary sub macroblock with an index of 2 has a boundary sub macroblock with an index of 0 at the top, so that vertical prediction can be set to intra prediction mode, and a non-boundary sub macroblock with an index of 3 has an index of Since a non-boundary sub macroblock of 1 is located at the top, vertical prediction may be set to an intra prediction mode. In other cases, the intra prediction mode of the non-boundary sub macro block may be set as the average prediction.

The pixel value of the non-boundary sub macroblock may be predicted using the set intra prediction mode and the reconstructed pixel value of the neighboring block of the non-boundary sub macroblock (S1030).

12 is a block diagram illustrating a method of decoding a sub macro block in a boundary macro block in case of a P / B slice.

Referring to FIG. 12, frame offset information of an image sequence may be obtained (S1200). As described above, the boundary macroblock in the frame may be specified using the frame offset information (S1205).

If the current macroblock corresponds to a non-boundary macroblock according to the frame offset information, skip mode indication information may be obtained from the slice header (S1210). When the current macroblock is coded in the skip mode according to the skip mode indication information, the motion information and the reference picture index may not be obtained, and the motion information and the reference picture index may be derived from the reference block (S1215). The current macroblock may be predicted using the derived motion information and the reference picture index (S1230). When the current macroblock is not coded in the skip mode according to the skip mode indication information, a macroblock type may be obtained from a macroblock layer (S1220). Motion information may be obtained for a partition of the current macroblock according to the macroblock type (S1225), and the current macroblock may be predicted based on the motion information (S1230).

When the current macroblock corresponds to a boundary macroblock according to the frame offset information, the macroblock type information of the current macroblock is not transmitted to the decoder and may be set in consideration of the slice type (S1235). The macroblock type information of the current macroblock may include partition information of the current macroblock, and the partition information of the current macroblock may mean a block unit for performing inter prediction.

For example, when the current macroblock is 32x32, the slice type may be set to P_16x16, P_32x16, or P_16x32 when the slice type is P slice, and set to B_16x16, B_32x16, or B_16x32 when the slice type is B slice. Can be. Additionally, the macroblock type of the current macroblock may be determined in consideration of the position of the boundary macroblock, which will be described with reference to FIG. 13.

FIG. 13 illustrates a method of determining a macro block type according to a position of a boundary macro block in a P / B slice.

Referring to FIG. 13A, when the boundary macroblock is located at the left or right side of the frame, the macroblock type of the boundary macroblock may be set to P_16x32 or B_16x32 in consideration of the slice type.

Referring to FIG. 13B, when the boundary macroblock is located at the top or bottom of the frame, the macroblock type of the boundary macroblock may be set to P_32x16 or B_32x16 in consideration of the slice type.

If the current macroblock corresponds to the boundary macroblock, it may be specified whether the partition of the current macroblock is a boundary submacroblock using the frame offset information (S1240). The partition of the current macroblock may be interpreted as the same meaning as the current submacroblock. If the partition of the current macroblock is a boundary sub-macroblock, a block type for the partition of the current macroblock is obtained in the sub-macroblock layer (S1245), and based on the block type, for a partition of the current macroblock. Motion information may be obtained (S1250). The partition of the current macroblock may be predicted using the obtained motion information (S1255).

If the partition of the current macro block is a non-boundary sub macro block, the partition of the current macro block may be set to be coded using a skip mode (S1260). Accordingly, the partition of the current macroblock may derive motion information from a reference block and perform inter prediction based on the derived motion information (S1265).

The present invention can be applied to encode and decode video signals.

Claims (11)

Obtaining a macro block type of a current macro block;
If the current macroblock type is intra mixed, obtaining block type identification information on a partition of the current macroblock;
Specifying a block type of a partition of the current macro block based on the block type identification information;
Acquiring an intra prediction mode of the current macro block according to a block type of a partition of the current macro block; And
Predicting the current macroblock using the intra prediction mode and pixel values of a neighboring block. The video signal processing method according to claim 1, wherein the block type identification information is information specifying a prediction mode of a partition of the current macro block and a size of a prediction block. The method of claim 1, wherein the obtaining of the intra prediction mode
Acquiring intra prediction mode reference information based on a block type of a partition of the current macro block,
The intra prediction mode reference information is information indicating whether an intra prediction mode of the prediction block is derived from an intra prediction mode of a neighboring block based on the prediction block. The intra prediction mode of claim 3, wherein the intra prediction mode of the current macro block is derived from the intra prediction mode of the neighboring block when the intra prediction mode reference information indicates that the intra prediction mode of the prediction block is derived from the intra prediction mode of the neighboring block. And a video signal processing method. The method of claim 4, wherein the neighboring block includes a left block and a right block based on the prediction block, and the intra prediction mode of the current macro block includes the intra prediction mode of the left block and the intra prediction of the right block. The video signal processing method characterized in that it is derived to the minimum value among the modes. 4. The method of claim 3, further comprising parsing the intra prediction mode of the current macro block when the intra prediction mode reference information indicates that the intra prediction mode of the prediction block is not derived from the intra prediction mode of the neighboring block. The video signal processing method characterized in that. Acquiring frame offset information for an image sequence, wherein the frame offset information is samples in a picture of the image sequence and is information for specifying samples belonging to a specific region in a frame;
Specifying whether a current macroblock corresponds to a boundary macroblock by using the frame offset information, wherein the boundary macroblock is a macroblock including an area not belonging to the output area;
If the current macroblock corresponds to a boundary macroblock according to the frame offset information, deriving a macroblock type of the current macroblock in consideration of a slice type;
The output region is a rectangular region specified within the coding region, and the coding region is a frame output by a decoding process. The method of claim 7, wherein when the current macroblock is coded in the intra mode according to the macroblock type, deriving the size information, the CBP information, and the intra prediction mode of the prediction block for the non-boundary sub macroblock. Include more,
The non-boundary sub macro block is a sub macro block of the boundary macro block, which is a sub macro block that does not include an area belonging to the output area, and the intra prediction mode includes horizontal prediction, vertical prediction, and average prediction. A video signal processing method. 10. The method of claim 8, wherein the CBP information is set to 0, and the intra prediction mode is derived in consideration of a position within a current frame of the non-boundary sub macro block. 10. The method of claim 9, wherein when the non-boundary sub macroblock is located at the left or top of the current frame, the intra prediction mode is derived with an average prediction, and the non-boundary sub macroblock is right in the current frame. If located at, the intra prediction mode is derived from horizontal prediction, and if the non-boundary sub macroblock is located at the bottom of the current frame, the intra prediction mode is derived from vertical prediction. Treatment method. 8. The method of claim 7, wherein the non-boundary sub macroblock is coded in a skip mode when the current macroblock is coded in an inter mode according to the macroblock type.

Images