KR102329705B1 - Apparatus for encoding an image - Google Patents

Abstract

An apparatus for encoding a residual signal in intra prediction according to the present invention includes a transform unit that transforms a residual signal using a transform matrix determined according to the size of a transform unit to generate a transform block, and a transform block using a quantization step size Corresponds to an intra prediction mode selected from a quantization unit generating a quantized block by quantization, a scanning unit generating one-dimensional quantization coefficient information by scanning the coefficients of the quantization block, and a preset number of intra prediction modes according to the size of the current block an intra prediction unit generating a prediction block that A scan is performed using a scan pattern determined according to A plurality of subsets are scanned using a pattern. Accordingly, it is possible to generate a prediction block close to the original image, thereby increasing the image compression rate.

Description

Image encoding apparatus {Apparatus for encoding an image}

The present invention relates to an apparatus for encoding an image, and more particularly, to an apparatus for generating a prediction block capable of reducing a coding amount of a residual block.

In video compression methods such as MPEG-1, MPEG-2, MPEG-4 H.264/MPEG-4, and Advanced Video Coding (AVC), one picture is divided into macroblocks in order to encode an image. Then, each macroblock is encoded using inter prediction or intra prediction.

Among these, intra prediction does not refer to a reference picture to encode a block of a current picture, but performs encoding using pixel values spatially adjacent to the current block to be encoded. First, an intra prediction mode with less distortion is selected by comparing a prediction block generated using pixel values of adjacent pixels with an original macroblock. Next, the prediction value of the current block to be encoded is calculated using the selected intra prediction mode and adjacent pixel values, the difference between the predicted value and the pixel value of the original current block is obtained, and then this is encoded through transform encoding, quantization, and entropy encoding. . And the prediction mode is also encoded.

Here, the intra prediction modes are largely divided into a 4×4 intra prediction mode of a luminance component, an 8×8 intra prediction mode, a 16×16 intra prediction mode, and an intra prediction mode of a chrominance component.

In the 16×16 intra prediction mode according to the prior art, there are a total of four modes: a vertical mode, a horizontal mode, a direct current (DC) mode, and a plane mode.

The 4×4 intra prediction mode according to the prior art includes a vertical mode, a horizontal mode, a direct current (DC) mode, a diagonal down-left mode, a diagonal down-right mode, There are a total of nine modes: a vertical right mode, a vertical left mode, a horizontal-up mode, and a horizontal-down mode.

The prediction mode numbers indexed in each mode are numbers determined according to the frequency at which each mode is used. The vertical mode, which is the probabilistic mode 0, is the most used mode when performing intra prediction on the target block, and the horizontal-up mode, which is the 8th mode, is the least used mode.

According to the H.264 standard, when encoding an image, a current block is encoded in a total of 13 modes, the 4×4 intra prediction mode and the 16×16 intra prediction mode, and the bitstream for the current block according to the optimal mode among them. create

However, when some or all of the pixel values adjacent to the pixels included in the current block do not exist or are not encoded, some or all of the intra prediction modes cannot be applied. In addition, if intra prediction is performed by selecting one of the applicable intra modes, the residual between the prediction block and the original block is large, so that the efficiency of image compression is reduced.

SUMMARY OF THE INVENTION It is an object of the present invention to provide an apparatus for generating a prediction block that generates a prediction block similar to an original image.

An apparatus for encoding a residual signal in intra prediction according to the present invention includes: a transform unit configured to transform the residual signal using a transform matrix determined according to a size of a transform unit to generate a transform block; a quantization unit that quantizes the transform block using a quantization step size to generate a quantized block; a scanning unit that scans the coefficients of the quantization block to generate one-dimensional quantization coefficient information; an intra prediction unit generating a prediction block corresponding to an intra prediction mode selected from among a plurality of preset intra prediction modes; and an entropy encoding unit entropy encoding the one-dimensional quantization coefficient information, wherein the scanning unit determines the coefficients of the quantization block according to the selected intra prediction mode when the transform unit is smaller than or equal to a first reference size. scan pattern using the selected scan pattern, and the scanning unit divides the coefficients of the quantization block into a plurality of subsets if the transform unit is greater than the first reference size and not greater than the second reference size, and is applied to the selected intra prediction mode. The plurality of subsets are scanned using a scan pattern determined according to the plurality of subsets.

Advantageously, said quantization step size is determined per coding unit.

Preferably, if the quantization step size of the left coding unit and the quantization step size of the upper coding unit are both invalid, the quantization step size is set to the quantization step size of the previous coding unit.

Preferably, the first reference size is 4x4, and the second reference size is 8x8.

Preferably, if the selected intra prediction mode is one of a vertical mode or a predetermined number of intra prediction modes adjacent to the vertical mode, the scan pattern is a horizontal scan.

Preferably, if the selected intra prediction mode is a non-directional mode, the scan pattern is a predetermined scan pattern.

Advantageously, said plurality of subsets are scanned in a reverse direction.

Preferably, the scan pattern applied to the plurality of subsets and the scan pattern applied to the coefficients in each subset are the same.

Preferably, the intra prediction unit adaptively filters a prediction block according to the selected intra prediction mode, and if the selected intra prediction mode is a DC mode, upper and left sides of the prediction block using reference pixels adjacent to the prediction block Filter the pixels in the line.

An apparatus for encoding a residual signal in intra prediction according to the present invention generates a prediction block close to the original image by generating a reference pixel to generate a prediction block close to the original image, and adaptively filtering the reference pixel. . In addition, by adaptively filtering the prediction block according to the intra prediction mode, it is possible to further reduce the residual signal, thereby increasing the image compression rate.

1 is a block diagram illustrating a video encoding apparatus according to the present invention.
2 is a flowchart illustrating an operation process of a scanning unit according to the present invention.
3 is a block diagram illustrating a video decoding apparatus according to the present invention.
4 is a block diagram illustrating an intra prediction unit according to the present invention.
5 is a diagram illustrating positions of reference pixels used for intra prediction according to the present invention.
6 is a diagram illustrating a method of generating a reference pixel according to the present invention.
7 is a block diagram illustrating an intra prediction unit of a decoding apparatus according to the present invention.

Hereinafter, various embodiments of the present invention will be described in detail with reference to exemplary drawings. The present invention is not intended to limit the present invention to a specific embodiment, as various changes can be made and various embodiments can be made, and all changes, equivalents or substitutes included in the spirit and scope of the present invention are included. should be understood as including In describing each figure, like reference numerals have been used for like elements.

For video encoding, each picture consists of a plurality of slices, and each slice consists of a plurality of coding units. Since there are many regions in which the image is relatively flat in an image of HD level or higher, encoding using coding units of various sizes may increase the image compression rate.

The coding unit according to the present invention may be a block of a size of 32x32 or a block of a size of 64x64 as well as an MB of a size of 16x16. Also, a block having a size of 8x8 or smaller may be a coding unit. For convenience, the largest coding unit is referred to as a super macroblock (SMB). In addition, the size of the super macroblock may be determined by information indicating the size of the smallest coding unit and depth information. The depth information indicates a difference in size between the super macroblock and the smallest coding unit.

Accordingly, the size of coding units to be used for encoding all pictures of the image sequence may be the size of the SMB and its subblocks. The size of a coding unit to be used for an image sequence may be designated as a default or may be designated in a sequence header.

The SMB (ie, the largest coding unit) is composed of one or more coding units. Accordingly, the SMB has the form of a recursive coding tree to include the coding unit and the split structure of the coding unit. Therefore, the description is based on the SMB as follows. First, if the SMB is not split into four lower coding units, the coding tree may be composed of information indicating that the SMB is not split and one coding unit. However, when the SMB is divided into four sub-coding units, the coding tree may be composed of information indicating the split and four sub-coding trees. Similarly, each sub-encoding tree has the same structure as the coding tree of the SMB unit. However, it is not divided below the minimum coding unit.

Meanwhile, in each coding unit in the coding tree, intra prediction or inter prediction is performed in units of the coding unit itself or sub-partitions. A unit for performing the intra prediction or inter prediction is referred to as a prediction unit. A prediction unit for intra prediction may be 2Nx2N or NxN. A prediction unit for inter prediction may be 2Nx2N, 2NxN, Nx2N, or NxN. Here, 2N means the horizontal and vertical length of the coding unit .

The coding unit includes a prediction mode of a prediction unit in the coding unit and size information (partmode) for the prediction unit. In order to increase encoding efficiency, joint coding may be performed by combining prediction mode information and size information of a prediction unit. In this case, a joint-coded prediction type (pred_type) is included for each coding unit.

The coding unit includes an additional information container for each prediction unit. The additional information container includes additional information necessary to generate a prediction block of a prediction unit. In the case of intra prediction, the side information includes encoded intra prediction information. In the case of inter prediction, the side information includes coded motion information. The motion information includes a motion vector and a reference picture index.

The encoding unit includes a residual signal container for containing the encoded residual signal of the encoding unit. The residual signal container includes one transform tree, one luminance residual signal container and two chrominance residual signal containers. The transform tree represents a split structure of transform units with respect to the residual signal included in the coding unit. The transform tree also indicates whether the residual signal of each transform unit is zero. The residual signal container consists of a recursive coding tree structure. That is, a description will be made based on the residual signal container of the coding unit as follows. First, if the coding unit is not divided into four lower coding units, the residual signal container includes quantization information (residual quantization parameter) and the coded residual signal. If the coding unit is divided into four lower coding units, the residual signal container includes quantization information and four lower residual signal containers. Each lower residual signal container has the same structure as the residual signal container of the coding unit, except that quantization information is not included.

On the other hand, the prediction unit may be partitioned unevenly. In the above description, only the prediction units in which the coding units are equally divided have been described. However, when the above-described equal division is used when the image has a boundary in a specific direction or a specific location, different prediction units are used for similar data at the boundary, so that a residual signal cannot be effectively reduced.

Therefore, in this case, it is more effective to compress the residual signal to perform intra or inter prediction by dividing the image in a specific direction according to the shape of the boundary portion of the image.

The simplest method is to divide the coding unit into two blocks using a straight line to extract the statistical dependence of the local topography of the prediction region. That is, it is divided by matching the boundary part of the image with a straight line. In this case, it is preferable to limit the divisible directions to a predetermined number. For example, a method of dividing a block may be limited to four directions: horizontal, vertical, upward diagonal, and downward diagonal. In addition, it may be limited only horizontally and vertically. The divisible number can be 3, 5, 7, etc. The number of divisions may vary depending on the size of the block. For example, a relatively divisible number of coding units having a large size may be increased.

In the case of inter prediction, when one coding unit is divided into two prediction blocks for more adaptive prediction, motion prediction and motion compensation must be performed for each prediction block. Accordingly, motion information is obtained for each divided region, and a residual signal with a prediction block obtained based on the motion information is encoded.

After obtaining a residual signal for each of the two prediction blocks divided for one coding unit, the two residual signals are added to form a residual signal of one coding unit, and transcoding is performed. In this case, since there is a high probability that the distribution of the residual signal of each divided region around the boundary has a high probability that the overall distribution is different, the residual signal for one coding unit can be generated by multiplying the value of one region by a certain value. have. In addition, one residual signal may be generated by smoothing the boundary portions by overlapping the boundary portions of the two residual signals.

As another method, a block may be generated and encoded through padding for each divided region of the block. That is, when encoding the current partition among the two partitions, another partition constituting the block may be padded with the values of the current partition to form one block, and then 2D transform encoding may be performed.

1 is a block diagram illustrating a video encoding apparatus according to the present invention.

Referring to FIG. 1 , a video encoding apparatus 100 according to the present invention includes a picture divider 110 , a transform unit 120 , a quantization unit 130 , a scanning unit 131 , an entropy encoding unit 140 , and an intra Prediction unit 150 , inter prediction unit 160 , inverse quantization unit 135 , inverse transform unit 125 , post-processing unit 170 , picture storage unit 180 , subtraction unit 190 and addition unit 195 . includes

The picture splitter 110 analyzes the input video signal, divides the picture into coding units of a predetermined size for each largest coding unit, determines a prediction mode, and determines the size of a prediction unit for each coding unit. Then, the picture splitter 110 transmits the prediction unit to be encoded to the intra prediction unit 150 or the inter prediction unit 160 according to the prediction mode. Also, the picture splitter 110 transmits a prediction unit to be encoded to the subtractor 190 .

The transform unit 120 transforms a residual signal between the original block of the input prediction unit and the prediction block generated by the intra prediction unit 150 or the inter prediction unit 160 . The residual block may have a size of a coding unit. The residual block is divided into optimal transform units and transformed. Different transform matrices may be determined according to prediction modes (intra or inter). A transform unit of the residual signal may be transformed by a horizontal one-dimensional transform matrix and a vertical one-dimensional transform matrix. For example, in the case of inter prediction, one predetermined transform matrix is applied. On the other hand, in the case of intra prediction, when the intra prediction mode of the current prediction unit is horizontal, the probability that the residual signal has a vertical direction increases, so a DCT-based integer matrix is applied in the vertical direction, and the horizontal direction DST-based or KLT-based integer matrix is applied. When the intra prediction mode is vertical, a DST-based or KLT-based integer matrix is applied in a vertical direction and a DCT-based integer matrix is applied in a horizontal direction. In the case of DC mode, DCT-based integer matrix is applied in both directions. In addition, in the case of intra prediction, the transform matrix may be adaptively determined depending on the size of the transform unit.

The quantization unit 130 determines a quantization step size for quantizing transform coefficients of the residual block for each coding unit. Then, the coefficients of the transform block are quantized using a quantization matrix determined according to the determined quantization step size and prediction mode. The quantizer 130 uses a quantization step size of a coding unit adjacent to the current coding unit as a quantization step size predictor of the current coding unit. The quantization unit 130 searches in the order of the left coding unit, the upper coding unit, and the upper left coding unit of the current coding unit, determines the quantization step size of the effective coding unit as the quantization step size predictor of the current coding unit, and sets the difference value as entropy It is transmitted to the encoder 140 .

Meanwhile, when a slice is divided into coding units, there is a possibility that all of the left coding unit, the upper coding unit, and the upper left coding unit of the current coding unit do not exist. On the other hand, a previously existing coding unit in a coding order within the largest coding unit may exist. Accordingly, within the coding units adjacent to the current coding unit and the largest coding unit, the coding unit immediately preceding in the coding order may be a candidate. In this case, in order of 1) the left coding unit of the current coding unit, 2) the upper coding unit of the current coding unit, 3) the upper left coding unit of the current coding unit, and 4) the coding unit immediately preceding in the coding order, can The order may be changed, and the upper left coding unit may be omitted.

The quantized transform block is provided to the inverse quantization unit 135 and the scanning unit 131 .

The scanning unit 131 scans the coefficients of the quantized transform block and transforms them into one-dimensional quantized coefficients. Since the coefficient distribution of the transform block after quantization may depend on the intra prediction mode, the scanning method is determined according to the intra prediction mode. Also, the coefficient scanning method may be determined differently according to the size of the transform unit.

The inverse quantization 135 inversely quantizes the quantized quantization coefficient. The inverse transform unit 125 restores the inverse quantized transform coefficient to a residual block in the spatial domain. The adder generates a reconstructed block by combining the residual block reconstructed by the inverse transform unit and the prediction block from the intra prediction unit 150 or the inter prediction unit 160 .

The post-processing unit 160 performs a deblocking filtering process to remove a blocking effect occurring in the reconstructed picture, an adaptive offset application process to compensate for a difference value from the original image in units of pixels, and a difference from the original image in units of coding. An adaptive loop filter process is performed to complement the values.

It is preferable to apply the deblocking filtering process to a boundary between a prediction unit and a transform unit having a size greater than or equal to a predetermined size. The size may be 8x8. The deblocking filtering process includes determining a boundary to be filtered, determining a boundary filtering strength to be applied to the boundary, determining whether to apply a deblocking filter, and the deblocking filter and selecting a filter to be applied to the boundary when it is determined to apply .

Whether the deblocking filter is applied is determined by i) whether the boundary filtering intensity is greater than 0, and ii) a value indicating the degree of change in pixel values at the boundary of two blocks (P block, Q block) adjacent to the boundary to be filtered. It is determined by whether it is smaller than a first reference value determined by the quantization parameter.

The filter is preferably at least two or more. When the absolute value of the difference value between two pixels located at the block boundary is greater than or equal to the second reference value, a filter that performs relatively weak filtering is selected. The second reference value is determined by the quantization parameter and the boundary filtering strength.

The adaptive loop filter process may perform filtering based on a value obtained by comparing the original image with the reconstructed image that has undergone the deblocking filtering process or the adaptive offset application process. The adaptive loop filter is detected through one Laplacian activity value based on a block of 4x4 size. The determined ALF may be applied to all pixels included in a block having a size of 4x4 or 8x8. Whether to apply the adaptive loop filter may be determined for each coding unit. The size and coefficient of the loop filter to be applied may vary according to each coding unit. Information indicating whether the adaptive loop filter is applied to each coding unit, filter coefficient information, filter form information, etc. may be included in each slice header and transmitted to the decoder. In the case of a color difference signal, it may be determined whether the adaptive loop filter is applied in units of pictures. The shape of the loop filter may also have a rectangular shape unlike the luminance.

The picture storage unit 180 receives the post-processed image data from the post-processing unit 160, restores the image in units of pictures, and stores it. A picture may be an image in units of frames or images in units of fields. The picture storage unit 180 includes a buffer (not shown) capable of storing a plurality of pictures.

The inter prediction unit 150 performs motion estimation using at least one or more reference pictures stored in the picture storage unit 180 , and determines a reference picture index and a motion vector indicating the reference picture. Then, according to the determined reference picture index and motion vector, from a reference picture used for motion estimation among a plurality of reference pictures stored in the picture storage unit 150, a prediction block corresponding to a prediction unit to be encoded is extracted and output. .

The intra prediction unit 140 performs intra prediction encoding by using the reconstructed pixel value inside the picture including the current prediction unit. The intra prediction unit 140 receives a current prediction unit to be prediction-encoded, selects one of a preset number of intra prediction modes according to the size of the current block, and performs intra prediction. The intra prediction unit adaptively filters reference pixels to generate an intra prediction block. When the reference pixel is not available, the reference pixels may be generated using the available reference pixels.

The entropy encoding unit 130 entropy-encodes the quantized coefficients quantized by the quantization unit 130 , intra prediction information received from the intra prediction unit 140 , motion information received from the inter prediction unit 150 , and the like.

FIG. 2 is a flowchart illustrating a scanning operation process of the scanning unit 131 of FIG. 1 .

First, it is determined whether to divide the currently quantized coefficient blocks into a plurality of subsets (S110). The determination of whether to divide depends on the size of the current transformation unit. That is, when the size of the transform unit is greater than the first reference size, the coded quantization coefficients are divided into a plurality of subsets. Preferably, the first reference size is 4x4 or 8x8. The predetermined size information may be transmitted to the decoder through a picture header or a slice header.

If it is determined that the quantized coefficient blocks are not divided into a plurality of subsets, a scan pattern to be applied to the quantized coefficient blocks is determined ( S120 ). Then, the coefficients of the quantized coefficient block are scanned according to the determined scan pattern (S130). The scan pattern may be adaptively determined according to a prediction mode and an intra prediction mode.

In the case of the inter prediction mode, only one predetermined scan pattern (eg, zigzag scan) may be applied. Also, it is possible to select and scan any one of a plurality of predetermined scan patterns and transmit scan pattern information to the decoder.

In the case of intra prediction, a predetermined scan pattern may be applied according to the intra prediction mode. For example, horizontal priority scan is applied to a vertical intra prediction mode and a predetermined number of intra prediction modes in an adjacent prediction direction. A vertical scan is applied to a horizontal intra prediction mode and a predetermined number of intra prediction modes having adjacent prediction directions. The predetermined number varies depending on the number of intra prediction modes (or the number of directional intra prediction modes) allowed for each prediction unit or the size of the prediction unit. For example, when the number of directional intra prediction modes allowed for the current prediction unit is 16, it is preferable that the number of directional intra prediction modes is two on both sides of the horizontal or vertical intra prediction mode. When the number of directional intra prediction modes allowed for the current prediction unit is 33, it is preferable that the number of directional intra prediction modes is 4 on both sides based on the horizontal or vertical intra prediction modes.

Meanwhile, zigzag scan is applied to non-directional modes. The non-directional mode may be a DC mode or a planar mode.

If it is determined that the currently quantized coefficient blocks are divided into a plurality of subsets, the transform quantized coefficient blocks are divided into a plurality of subsets (S140). The plurality of subsets includes one main subset and at least one remaining subset. The main subset is located on the upper left side including the DC coefficient, and the remaining subset covers a region other than the main subset.

Next, a scan pattern to be applied to the subset is determined (S150). The scan pattern is applied to all subsets. The scan pattern may be adaptively determined according to a prediction mode and an intra prediction mode. When the size of the current transform quantized coefficient block (size of the transform block) is larger than the second reference size (eg, 8x8), only the zigzag scan may be applied. Accordingly, the above step is performed only when the first reference size is smaller than the second reference size.

In the inter prediction mode, the scan pattern in the subset may be one predetermined scan pattern (eg, zigzag scan). In the case of intra prediction, it is the same as in S120.

The scan order of the quantization coefficients in the subset is reversed. That is, entropy encoding may be performed by scanning non-zero quantization coefficients in the reverse direction from the last non-zero quantization coefficient in the subset according to a scan pattern.

Next, the quantization coefficients are scanned according to the scan pattern for each subset ( S160 ). The scan order of the quantization coefficients in the subset is reversed. That is, entropy encoding may be performed by scanning non-zero quantization coefficients in the reverse direction from the last non-zero quantization coefficient in the subset according to a scan pattern.

In addition, a zigzag scan is applied to the scan pattern between the subsets. The scan pattern is preferably scanned from the main subset to the remaining subsets in a forward direction, but the reverse direction is also possible. Also, the scan pattern between the subsets may be applied in the same way as the scan pattern applied within the subset.

Meanwhile, the encoder transmits information indicating the position of the last non-zero quantization coefficient in the transform unit to the decoder. Information capable of indicating the position of the last non-zero quantization coefficient in each subset is also transmitted to the decoder. The information may be information indicating the position of the last non-zero quantization coefficient in each subset.

3 is a block diagram illustrating a video decoding apparatus according to an embodiment of the present invention.

Referring to FIG. 3 , the video decoding apparatus according to the present invention includes an entropy decoding unit 210 , an inverse scanning unit 220 , an inverse quantization unit 230 , an inverse transform unit 240 , an intra prediction unit 250 , and an inter It includes a prediction unit 260 , a post-processing unit 270 , a picture storage unit 280 , an adder 290 , and an intra/inter switching switch 295 .

The entropy decoding unit 210 decodes the received encoded bit stream, and separates it into intra prediction information, inter prediction information, quantization coefficient information, and the like. The entropy decoding unit 210 supplies the decoded inter prediction information to the inter prediction unit 260 . The entropy decoding unit 210 supplies the intra prediction information to the intra prediction unit 250 and the inverse transform unit 240 . Also, the entropy decoding 210 supplies the inverse quantization coefficient information to the inverse scan unit 220 .

The inverse scanning unit 220 converts the inverse quantization coefficient information into a two-dimensional array of inverse quantization blocks. For the transformation, one of a plurality of inverse scanning patterns is selected. The coefficient inverse scanning pattern is determined based on at least one of a prediction mode and an intra prediction mode. The operation of the reverse scanning unit 220 is the same as the reverse process of the operation of the scanning unit 131 described above.

The inverse quantizer 230 determines a quantization step size predictor of the current coding unit. The process of determining the predictor is the same as the process of determining the predictor of the quantization unit 130 of FIG. 1 , and thus is omitted. The inverse quantization unit 230 adds the determined quantization step size predictor and the received residual quantization step size to obtain a quantization step size applied to the current inverse quantization block. The inverse quantization unit 230 restores the inverse quantization coefficients using the quantization matrix to which the quantization step size is applied. Different quantization matrices are applied according to the size of the current block to be reconstructed, and a quantization matrix is selected based on at least one of a prediction mode and an intra prediction mode of the current block even for a block of the same size.

The inverse transform unit 240 inverse transforms the inverse quantized block to reconstruct the residual block. Then, the residual block is reconstructed by inverse transforming the reconstructed quantization coefficient. An inverse transform matrix to be applied to the inverse quantization block may be adaptively determined according to a prediction mode (intra or inter) and an intra prediction mode. Since the inverse transform matrix of the transform matrix applied to the transform unit 120 of FIG. 1 is determined, a detailed description thereof will be omitted.

The adder 290 reconstructs the image block by adding the residual block reconstructed by the inverse transform unit 240 and the prediction block generated by the intra prediction unit 250 or the inter prediction unit 260 .

The intra prediction unit 250 reconstructs the intra prediction mode of the current block based on the intra prediction information received from the entropy decoding unit 210 . Then, a prediction block is generated according to the reconstructed intra prediction mode.

The inter prediction unit 260 reconstructs a reference picture index and a motion vector based on the inter prediction information received from the entropy decoding unit 210 . Then, a prediction block for the current block is generated using the reference picture index and the motion vector. When motion compensation of fractional precision is applied, a prediction block is generated by applying the selected interpolation filter.

Since the operation of the post-processing unit 270 is the same as that of the post-processing unit 160 of FIG. 1 , it is omitted.

The picture storage unit 280 stores the decoded image post-processed by the post-processing unit 270 in units of pictures.

4 is a block diagram illustrating the intra prediction unit 150 of the encoding apparatus 100 according to the present invention.

Referring to FIG. 4 , the intra prediction unit 150 includes a reference pixel generation unit 151 , a reference pixel filtering unit 152 , a prediction mode determination unit 153 , a prediction block generation unit 154 , and a prediction block filtering unit ( 155) and a prediction mode encoder 156.

The reference pixel generator 151 determines whether to generate reference pixels for intra prediction, and generates a reference pixel when necessary.

5 is a diagram illustrating positions of reference pixels used for intra prediction of a current prediction unit. As shown in FIG. 5 , the upper reference pixels of the current prediction unit are pixels (areas C and D) that extend twice the horizontal length of the current prediction unit, and the left reference pixels are the vertical length of the current prediction unit. Pixels (areas A and B) that exist over twice.

The reference pixel generating unit 151 determines whether the reference pixels at all the positions described above are available. When some of the reference pixels are not available, reference pixels in the unavailable locations are generated by using the available reference pixels.

First, a case in which reference pixels in any one of an upper or left region of a current prediction unit to be encoded are not available will be described.

For example, when the current prediction unit is located at an upper boundary of a picture or a slice, reference pixels (regions C and D) above the current prediction unit do not exist. Similarly, when the current prediction unit is located at the left boundary of a picture or a slice, left reference pixels (regions A and B) do not exist. As such, when either one of the reference pixels is not available, the reference pixel may be generated by copying the nearest available reference pixel. In the former case, the nearest available reference pixel is the upper left reference pixel (ie, the uppermost reference pixel in region A). In the latter case, the closest available reference pixel is the upper leftmost reference pixel (that is, the leftmost reference pixel in region C). The above-described method may be applied by default, but may be adaptively applied for each sequence unit, screen unit, or slice unit, if necessary.

Next, a case in which some of the upper or left reference pixels of the current prediction unit to be encoded are not available will be described. Based on the unavailable reference pixel, there are two cases: 1) there are reference pixels that are usable in only one direction, and 2) there are reference pixels that are usable in both directions.

Case 1) will be described.

For example, when the current prediction unit is located at a right boundary of a picture or a slice or a right boundary of a maximum coding unit, reference pixels of the region D are not available. Similarly, when the current prediction unit is located at the lower boundary of the picture or slice or the lower boundary of the largest coding unit, the reference pixels of the region B are not available.

* In this case, reference pixels can be created by copying the available reference pixels that exist in the nearest position. As another method, reference pixels may be generated by using a plurality of available reference pixels present at the closest positions.

2) will be described.

For example, when the current prediction unit is located at the upper boundary of the slice and the right-right prediction unit of the current prediction unit is available, reference pixels corresponding to the area C of the current prediction unit are not available, but the area A and the area Reference pixels located in D are available. In this way, when there are reference pixels usable in both directions, one usable reference pixel present at the nearest position in each direction is selected one by one, and these reference pixels) to generate reference pixels at positions that are not available.

The rounded average value of the two reference pixels (pixels located closest to each other in each direction) may be generated as the reference pixel value. However, when an unavailable reference pixel area is large, a step difference between an available pixel and a generated pixel is highly likely to occur, so it is more useful to generate the reference pixel using a linear interpolation method. Specifically, an unavailable reference pixel of the current location may be generated by considering a location with two available reference pixels.

Next, a case in which both upper and left reference pixels of the current prediction unit to be encoded are not available will be described. For example, when the current prediction unit is adjacent to the upper-left boundary of a picture or slice, there are no available reference pixels.

In this case, some or all of the reference pixels may be generated using two or more pixels existing inside the current prediction unit. The number of pixels present in the current prediction unit used to generate the reference pixel may be 2 or 3.

6 is a diagram illustrating a method of generating a reference pixel according to the present invention.

A case in which two pixels are used will be first described with reference to FIG. 6 . This is a case of using a pixel different from the upper-left pixel (○) of the current prediction unit. The other pixel may be a right pixel (□), a lower left pixel (Δ), and a lower right pixel (▽). When the upper-left pixel (○) and the right-right pixel (□) of the current prediction unit are used, the upper-left pixel and the upper-right pixel are copied to the corresponding upper position, and the copied upper-left pixel and the right-right pixel are used. Thus, reference pixels in the region (region C) between them are generated. The generated reference pixels may be a rounded average value of the two reference pixels (the copied upper-left pixel and the right-right pixel) or a value generated by a linear interpolation method. Reference pixels of the region D are generated by copying the right pixel (□) or using a plurality of upper generated reference pixels. In the case of using the upper-left pixel (○) and the lower-left pixel (Δ) of the current prediction unit, the same method as above is applied. When the upper left pixel (○) and the lower right pixel (▽) of the current prediction unit are used, the lower right pixel pixel (▽) is copied to the reference pixel position at the same position in the horizontal and vertical directions. After that, the same method as above is applied to the reference pixel generation method.

A case in which three pixels are used will be described first. The upper left pixel (○), the right pixel (□), and the lower left pixel (Δ) of the current prediction unit may be used. In this case, after each pixel is copied to a corresponding reference pixel position, the remaining reference pixels are generated using the pixels. The generation method of the remaining reference pixels is the same as in the case of using the above two pixels.

Meanwhile, in the case of using the above method, pixels in the current prediction unit used to generate the reference pixel must be transmitted to the decoder. In this case, in order to reduce the amount of information transmitted, a value other than the upper-left pixel (○) transmits a difference value from the upper-left pixel (○). Also, the upper-left pixel value used to generate the reference pixel may be a quantized value, or may be entropy-encoded and transmitted.

The above-described method of generating reference pixels using two or more pixels existing inside the current prediction unit is effective when the slice type is intra(I).

Another method of generating a reference pixel when both reference pixels above and to the left of the current prediction unit to be encoded are not available will be described. This method is effective when the slice type is not intra (I).

First, it is determined whether pixels exist in the same position as the positions of the reference pixels of the current prediction unit in a previously coded reference picture of the current block. If present, the reference pixel of the current prediction unit is generated by copying the pixels in the reference picture.

If it does not exist, it is determined whether pixels are present in the previously encoded reference picture at the position closest to the position of the reference pixels of the current prediction unit (a position one pixel apart), and if present, they are copied to determine the size of the current prediction unit. used as a reference pixel.

The reference pixel filtering unit 152 adaptively filters reference pixels of the current prediction unit. This is for smoothing the amount of change in pixel values between reference pixels, and a low-pass filter is applied. The low-pass filter may be a 3-tap filter [1, 2, 1] or a 5-tap filter [1, 2, 4, 2, 1].

Also, the filter may be adaptively applied according to the size of the current prediction unit. For example, when the current prediction unit has a predetermined size or larger, the filter may not be applied. Specifically, when the size of the current prediction unit is 64x64, the filter is not applied.

Also, the low-pass filter may be adaptively applied according to the size of the current prediction unit and the intra prediction mode.

When the intra prediction mode is horizontal or vertical, one reference pixel is used to generate a pixel of a prediction block. Therefore, in this case, the filter is not applied. In addition, when the intra prediction mode is DC, since the average value of the reference pixels is used, the step difference between the reference pixels is not affected. Therefore, no filter is applied even in DC mode.

On the other hand, the filter is applied to modes in which the intra prediction direction is tilted by 45° with respect to the horizontal or vertical direction regardless of the size of the current prediction unit. In this case, a first filter may be used for a prediction unit smaller than a predetermined size, and a second filter having a greater smoothing effect may be used for a prediction unit having a size greater than or equal to a predetermined size. The predetermined size may be 16x16.

Meanwhile, in modes other than the above-mentioned modes (6 types) in which at least two or more reference pixels are used to generate a pixel of a prediction block, it may be adaptively applied according to the size of the current prediction unit and the intra prediction mode. However, in the planar mode, filtering is not performed on reference pixels.

Also, a filter may not be applied to the reference pixels generated by the linear combination.

The prediction block generator 153 generates a prediction block corresponding to the intra prediction mode. The prediction block is generated using reference pixels or a linear combination of reference pixels based on an intra prediction mode. The reference pixels used to generate the prediction block may be reference pixels filtered by the reference pixel filtering unit 152 .

The prediction block filtering unit 154 adaptively filters the prediction block generated by the prediction block generation unit 153 according to the used intra prediction mode. The operation is to minimize the residual signal between the generated prediction block and the current prediction unit to be encoded. That is, a difference value between reference pixels and pixels in an adjacent prediction block may vary according to the intra prediction mode. Accordingly, the residual can be reduced by filtering pixels in the prediction block generated by the intra prediction mode that generates a large number of steps.

In the DC mode, since the prediction block is made of an average value of reference pixels, a step may occur between reference pixels and pixels in an adjacent prediction block. Accordingly, pixels on the upper line and pixels on the left line in the prediction block adjacent to the reference pixels are filtered using the reference pixels. Specifically, since the pixel located in the upper-left corner in the prediction block has two adjacent reference pixels (the upper reference pixel and the left reference pixel), the pixel in the upper-left corner is filtered (or smoothed) using a 3-tap filter. )do. And, since the pixels other than those (ie, pixels on the upper line and pixels on the left line in the prediction block) have only one reference pixel, they are filtered using a 2-tap filter. In order to compensate for the step difference, some pixels of the prediction block are adaptively filtered even for a directional mode other than DC. Meanwhile, it may be adaptively applied according to the size of the current prediction unit. For example, a filter may not be applied to a prediction unit of a predetermined size or less for each intra prediction mode.

The prediction mode determiner 153 determines the intra prediction mode of the current prediction unit by using the reference pixels. The prediction mode determiner 153 may determine a mode in which the encoding amount of the residual block for each intra prediction mode is minimized as the intra prediction mode of the current prediction unit. To obtain a residual block, a prediction block is generated according to each intra prediction mode. The prediction block may be a prediction block using pixels filtered by the reference pixel filtering unit according to each intra prediction mode or a prediction block obtained by filtering the generated prediction block by the prediction block filtering unit 155 according to a predetermined condition. .

The prediction block transmitter 157 transmits the prediction block generated in response to the intra prediction mode determined by the prediction mode determiner 155 to the subtraction unit 190 .

The prediction mode encoder 156 encodes the intra prediction mode of the current prediction unit determined by the prediction mode determiner 155 . The prediction mode encoder 156 may be included in the intra predictor 150 and may be performed, or may be performed by the entropy encoder 140 .

The prediction mode encoder 156 encodes the intra prediction mode of the current prediction unit by using the upper intra prediction mode of the current prediction unit and the left intra prediction mode of the current prediction unit.

First, upper and left intra prediction modes of the current prediction unit are derived. When there are a plurality of upper prediction units of the current prediction unit, the intra prediction mode of the first effective prediction unit is set as the upper intra prediction mode while scanning in a predetermined direction (eg, right to left). Even when there are a plurality of left prediction units of the current prediction unit, the intra prediction mode of the first effective prediction unit may be set as the left intra prediction mode while scanning in a predetermined direction (eg, from bottom to top). Alternatively, the smallest mode number among the mode numbers of the plurality of valid prediction units may be set as the upper intra prediction mode.

When the upper intra prediction mode or the left intra prediction mode is not valid, the DC mode (mode number 2) may be set as the upper or left intra prediction mode. A case in which the upper or left intra prediction mode is not valid is a case in which a corresponding intra prediction unit does not exist.

Next, if the number of the upper or left intra prediction mode is greater than or equal to the number of intra prediction modes allowed in the current prediction unit, the upper or left intra prediction mode is converted into one of a predetermined number of intra prediction modes. The predetermined number varies depending on the size of the current prediction unit. For example, if the size of the current prediction unit is 4x4, it is mapped to one of nine modes (modes 0 to 8), and if the size of the current prediction unit is 64x64, it is mapped to one of three modes (modes 0 to 2). do. Also, it may be converted into one of intra prediction modes allowed in the current prediction unit.

Next, if the intra prediction mode number of the current prediction unit is the same as one of the left and upper intra prediction mode numbers, a flag indicating the sameness and a flag indicating any one of the upper and left intra prediction modes are transmitted. . In this case, if the left and upper intra prediction modes are the same, only a flag indicating the same may be transmitted. In addition, even if only one of the upper and left intra prediction modes is valid and is the same as the intra prediction mode of the current block, only a flag indicating the same as the mode of the adjacent block may be transmitted.

However, when the intra prediction mode of the current prediction unit is different from the left and upper intra prediction modes, the intra prediction mode number of the current prediction unit is compared with the left and upper intra prediction mode numbers. The number of cases not greater than the intra prediction mode number of the current prediction unit among the left and upper intra prediction mode numbers is calculated, and the value obtained by reducing the intra prediction mode number of the current prediction unit by the number of cases is the last of the current prediction unit to be transmitted. Determined in intra prediction mode. Here, when the left and upper intra prediction mode numbers are the same, it is treated as one.

A table for entropy-encoding the determined final intra prediction mode is determined according to whether the upper and left intra prediction modes are the same.

7 is a block diagram illustrating the intra prediction unit 250 of the decoding apparatus 200 according to the present invention.

The intra prediction unit 250 according to the present invention includes a prediction mode decoding unit 251 , a reference pixel generation unit 252 , a reference pixel filtering unit 253 , a prediction block generation unit 254 , and a prediction block filtering unit 255 . and a prediction block transmitter 256 .

The prediction mode decoder 251 reconstructs the intra prediction mode of the current prediction unit through the following process.

Additional information for generating a prediction block included in an additional information container in the received coding unit is received. The additional information includes a predictable flag and residual prediction mode information. The predictable flag indicates whether the current prediction unit is the same as any one intra prediction mode of adjacent prediction units. The residual prediction mode information includes other information according to the predictable flag. If the predictable flag value is 1, the residual prediction mode information may include a prediction mode candidate index. The prediction mode candidate index is information designating an intra prediction mode candidate. If the predictable flag value is 0, the residual prediction mode information includes a residual intra prediction mode number.

An intra prediction mode candidate of the current prediction unit is derived. The intra prediction mode candidate derives an intra prediction mode of an adjacent prediction unit. Here, for convenience, the intra prediction mode candidates of the current prediction unit are limited to upper and left intra prediction modes. When there are a plurality of upper or left prediction units of the current prediction unit, upper and left intra prediction modes of the current prediction unit are derived in the same manner as the prediction mode encoder 156 of the encoding apparatus 100 described above. In addition, if the number of the upper or left intra prediction mode is greater than or equal to the number of intra prediction modes allowed in the current prediction unit, it is converted in the same manner as that of the prediction mode encoder 156 .

Next, if the received predictable flag indicates the same as the intra prediction mode of the adjacent prediction unit, if the prediction mode candidate index exists, the prediction mode indicated by the prediction mode candidate index is determined as the intra prediction mode of the current prediction unit do.

If the received predictable flag indicates the same as the intra prediction mode of the neighboring prediction unit, but the prediction mode candidate index does not exist and there is one valid intra prediction mode of the neighboring prediction unit, this is the intra prediction of the current prediction unit. restore mode.

When the received predictable flag indicates that it is not the same as the intra prediction mode of the neighboring prediction unit, the intra prediction mode of the current prediction unit is determined by comparing the received residual intra prediction mode value with the intra prediction mode numbers of the valid intra prediction mode candidates. restore

The reference pixel generator 252 generates a reference pixel in the same manner as the reference pixel generator 151 of the encoding apparatus 100 . However, the reference pixel generator 252 and the reference pixel generator 151 of the encoding apparatus 100 adaptively generate a reference pixel according to the intra prediction mode restored by the prediction mode decoder 251. different. That is, the reference pixel generator 252 generates a reference pixel only when reference pixels to be used for generating a prediction block are not valid using the reconstructed intra prediction mode.

The reference pixel filtering unit 253 adaptively filters the reference pixels based on the intra prediction mode restored by the prediction mode decoding unit 251 and the size information of the current prediction unit. The filtering conditions and filters are the same as the filtering conditions and filters of the reference pixel filtering unit 152 of the encoding apparatus 100 .

The prediction block generation unit 254 is A prediction block is generated using reference pixels according to the intra prediction mode reconstructed by the prediction mode decoder 251 . ,

The prediction block filtering unit 255 adaptively filters according to the intra prediction mode reconstructed by the prediction mode decoding unit 251 . The operation is the same as that of the prediction block filtering unit 154 of the encoding apparatus 100 .

The prediction block transmitter 256 transmits the prediction block received from the prediction mode generator 254 or the prediction block filtering unit 255 to the adder 290 .

Claims (2)

a transform unit that transforms the residual signal using a transform matrix determined according to a size of a transform unit to generate a transform block;
a quantization unit that quantizes the transform block using a quantization step size to generate a quantized block;
a scanning unit that scans the coefficients of the quantization block to generate one-dimensional quantization coefficient information;
an intra prediction unit generating a prediction block corresponding to an intra prediction mode selected from among a plurality of intra prediction modes; and
and an entropy encoding unit for entropy encoding the one-dimensional quantization coefficient information.
The intra prediction unit adaptively filters the reference pixel according to the size of the prediction unit, but if the size of the prediction unit is larger than the preset size, the filter is not applied to the reference pixel regardless of the intra prediction mode of the prediction block A video encoding apparatus characterized in that. The method of claim 1,
The video encoding apparatus, characterized in that the preset size is 64×64.

Images