KR101283577B1 - Device and method for encodinh video image - Google Patents

Abstract

According to an embodiment of the present invention, in a video image encoding apparatus, a current macro block to be encoded in a video image, wherein the macro block has a plurality of pixels in a matrix form and a plurality of pixels in a row direction or a column direction forms a line. The intra predicted value is encoded after performing intra prediction on all the lines using the reconstructed pixel of the line before the specific line as the reference pixel for the specific line of-.

Description

A video image encoding method and a video image encoding apparatus performing the same {DEVICE AND METHOD FOR ENCODINH VIDEO IMAGE}

The present invention relates to a video image encoding method and a video image encoding apparatus for performing the same.

HDTV broadcasts are available in major countries around the world, and HD video content is commonly found in storage media, including Blu-ray Discs.

Typically, HD video content has screen resolutions such as 1024x768, 1920x1080. This HD video content offers higher image quality than SD video with 720x480 screen resolution.

As the service of high-definition video in 4K (3840x2160) and 8K (7680x4320) resolutions including HD is expected to become commonplace, the related technology is likely to become a new growth engine.

Recently, ISO / IEC Moving Picture Experts Group (MPEG) and ITU-T Video Coding Experts Group (VCEG) formed Joint Collaborative Team on Video Coding (JCT-VC) to create next generation video compression technology for high quality and ultra high quality video. Efficiency Video Coding (HEVC) standardization began.

Currently, technology development is underway with the aim of releasing the standard draft in 2012, and as of September 2010, a test model under Consideration (TMuC) has been produced and discussed.

High-definition video coding is based on the technology of H.264 / AVC, the latest video coding standard. Intra picture coding of H.264 / AVC utilizes already encoded pixels around a current macroblock as a reference block for prediction to remove spatial correlation and to improve coding efficiency.

However, H.264 / AVC is the latest video coding standard, but it has been a long time since its enactment and was not developed for high quality video. Therefore, in order to encode high quality video, it is necessary to introduce element technology specialized for high quality video encoding.

Currently, the intra prediction used in H.264 / AVC prediction includes in-screen 16x16, in-screen 8x8, and in-screen 4x4 modes. In H.264 / AVC, in picture 4 × 4 mode provides nine prediction modes, while in picture 16 × 16 mode provides only four prediction modes: vertical, horizontal, average and planar prediction. This in-picture 16 × 16 prediction provides four modes as shown in FIG.

1 shows four modes of intra prediction 16x16 prediction.

Referring to FIG. 1, four prediction modes defined as vertical prediction, horizontal prediction, average prediction, and plane prediction have a predetermined direction.

Among them, the vertical, horizontal, and average modes are the same as those in the in-picture 4x4 mode, which is larger in size, and the planar prediction mode is similar to the double linear transformation with only integer operations.

Therefore, in the complex region, the intra-picture 4x4 mode having various prediction modes is more selected as the optimum mode than the intra-picture 16x16 mode, and the intra-picture 16x16 mode is selected in the relatively flat region.

However, when encoding high-definition video in a near-lossless environment, the complex characteristics of the high-definition video are preserved as they are, and thus, 4x4 mode with high prediction efficiency is selected.

FIG. 2 shows an optimal mode distribution after performing intra picture coding in a lossy and near lossless environment, respectively, in an H.264 / AVC standard for an arbitrary video image.

At this time, a basketball video (1920 × 1080) video was used as an arbitrary video image.

Referring to Fig. 2A, in the lossy coding using the quantization coefficients (QP = 22, 27, 32, 37), the ratio of the 16x16 mode in the picture is high. However, referring to FIG. 2 (b), it can be seen that in the close lossless coding using the quantization coefficients (QP = 9, 12, 15, and 18), the ratio of the 4 × 4 mode in the picture is much higher.

In this case, since the intra 4 × 4 mode needs to transmit information on the prediction mode determined for each subblock to the decoder, it consumes a lot of bits.

Therefore, by increasing the prediction accuracy of the 16x16 mode in the screen and increasing the ratio of the 16x16 mode selected as the optimal mode, the coded bits can be efficiently reduced. Therefore, a method for improving the prediction accuracy of the 16x16 mode is required. It is true.

Accordingly, an object of the present invention is to encode a video image capable of increasing the accuracy of 16 × 16 prediction in a line-based screen that predicts a current pixel value from a pixel closest to the current pixel, and a video image encoding apparatus that performs the same. To provide.

According to an aspect of the present invention, an encoding method is provided. In the method for encoding a video image, the current macroblock to be encoded in the video image, wherein the macroblock is a plurality of pixels in a matrix form, a plurality of pixels in a row direction or a column direction form a line Performing intra prediction on all lines using a reconstructed pixel of the line preceding the specific line as a reference pixel for a specific line of the box; And encoding each intra predicted value for all the lines.

According to another feature of the present invention, an encoding apparatus is provided. In the apparatus for encoding a video image, the current macroblock to be encoded in the video image, wherein the macroblock has a plurality of pixels in a matrix form, and the plurality of pixels in a row direction or a column direction form a line. An intra predictor configured to perform intra prediction on all lines using a reconstructed pixel of the line preceding the specific line as a reference pixel with respect to a specific line of the box; And an encoder for encoding each intra predicted value for all the lines.

According to an exemplary embodiment of the present invention, since 16x16 prediction that predicts a current pixel value from a pixel closest to a current pixel is performed based on line prediction instead of conventional block-based prediction to use a close pixel as a reference pixel. It is possible to effectively reduce the bits required for high-quality video encoding by improving the accuracy of 16 × 16 prediction within a screen without modifying syntax elements of H.264 / AVC and transmitting additional information.

1 shows four modes of intra prediction 16x16 prediction.
FIG. 2 shows an optimal mode distribution after performing intra picture coding in a lossy and near-lossless environment using a H.264 / AVC standard for an arbitrary image.
3 is a block diagram illustrating a configuration of a video image encoding apparatus according to an embodiment of the present invention.
4 is a block diagram illustrating a detailed configuration of an intra predictor according to an exemplary embodiment of the present invention.
5 is a flowchart illustrating a video image encoding method according to an embodiment of the present invention.
6 to 9 illustrate a line-based intra prediction process in the vertical mode according to an embodiment of the present invention.
10 to 13 illustrate a line-based intra prediction process in a horizontal mode according to an embodiment of the present invention.

DETAILED DESCRIPTION Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those skilled in the art may easily implement the present invention. The present invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. In the drawings, parts irrelevant to the description are omitted in order to clearly describe the present invention, and like reference numerals designate like parts throughout the specification.

Throughout the specification, when an element is referred to as "comprising ", it means that it can include other elements as well, without excluding other elements unless specifically stated otherwise.

Hereinafter, a video image encoding method and a video image encoding apparatus for performing the same according to an embodiment of the present invention will be described in detail with reference to the accompanying drawings.

The video image encoding apparatus may be an H.264 / AVC encoder.

In this case, the video image encoding apparatus divides a frame of an input video image into macro blocks (MBs), and each of the divided macro blocks is encoded in an intra mode. In particular, it is divided into macroblock units having a size of 16x16 pixels.

Here, the configuration of the division will not be described separately, but a configuration of encoding the divided macroblock will be described.

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

Referring to FIG. 3, the video image encoding apparatus 100 may include an intra predictor 110, a transformer 120, a quantizer 130, an inverse quantizer 140, an inverse transformer 150, and a filter 160. , The frame storage unit 170, the decoder 180, and the encoder 190.

The intra predictor 110 performs intra prediction on a current macroblock having a size of 16 × 16 pixels.

A macro block intra predicted by the intra prediction unit 110 is subjected to a discrete cosine transform (hereinafter, collectively referred to as 'DCT') through the transform unit 120 and then quantized through the quantization unit 130. (Quantization). In order to generate the reference image, the quantized macro block is inverse quantized through the inverse quantization unit 140 and then inverse discrete cosine transform (IDCT) through the inverse transform unit 150. Collectively decoded to generate a reconstructed pixel. The reconstructed pixel is stored in the frame storage unit 170 after passing through the loop filter 160 and used as a reference image of the intra predictor 110, that is, a reference pixel.

Meanwhile, the macroblock quantized through the quantizer 130 is entropy coded through the decoder 180 and the encoder 190 and is output as an output image.

Here, the configuration of the intra prediction unit 110 will be described with reference to FIG. 4.

4 is a block diagram illustrating a detailed configuration of an intra predictor according to an exemplary embodiment of the present invention.

Referring to FIG. 4, the intra prediction unit 110 includes an intra prediction module 111, a calculation module 112, a selection module 113, a line based prediction module 114, and a block based prediction module 115. .

The intra prediction module 111 performs intra prediction for four 16 × 16 intra modes on the current macroblock divided into 16 × 16 units.

In this case, intra prediction in the H.264 / AVC compression region may be applied in units of 4 × 4 blocks and 16 × 16 blocks, 9 prediction modes are applied to the 4 × 4 blocks, and 4 types of 16 × 16 blocks are used. Prediction mode may be applied.

In the 16 × 16 block, four prediction modes are composed of vertical prediction, horizontal prediction, average prediction, and plane prediction, as described with reference to FIG. 1.

Accordingly, the intra prediction module 111 performs all of the vertical prediction, the horizontal prediction, the average prediction, and the plane prediction for the current macroblock. In this case, the reference pixel uses a reconstruction pixel of an already encoded macro block, and the reconstruction pixel is previously stored in the frame storage unit 170.

The calculation module 112 calculates a sum of absolute transform difference (SATD) between the predicted current macroblock and the current macroblock before prediction for each mode performed by the intra prediction module 111.

The selection module 113 compares the SATD for each intra mode calculated by the calculation module 112 and selects the mode having the highest value as the optimal prediction mode.

In this case, when the horizontal mode or the vertical mode is selected as the optimal mode, the operation of the line-based prediction module 114 is commanded.

However, when the optimal mode is the average prediction or the planar prediction, the operation of the block-based prediction module 115 is instructed.

The line-based prediction module 114 performs intra prediction on a line basis and will be described in detail with reference to FIGS. 5 to 13.

The block-based prediction module 115 performs 16 × 16 intra prediction on a block basis in conventional H.264 / AVC. That is, intra prediction is performed by using all pixels of the current macro block as reconstruction pixels of the already encoded macro block as reference pixels.

Next, a video image encoding process will be described based on the configuration of FIGS. 3 and 4. The same reference numerals are used in the steps performed by the components of FIGS. 3 and 4.

5 is a flowchart illustrating a video image encoding method according to an embodiment of the present invention.

Referring to FIG. 5, the intra prediction module 111 performs vertical prediction, horizontal prediction, and average values using reconstructed pixels of a previously encoded macro block stored in the frame storage unit 170 with respect to a 16 × 16 unit current macroblock to be encoded. Both prediction and plane prediction are performed (S101).

Then, the calculation module 112 calculates the SATD for each prediction block for the four modes generated in step S101 (S103).

The selection module 113 compares the SATD calculated in step S103 (S105) and determines whether the smallest mode, that is, the optimal mode is the horizontal prediction mode or the vertical prediction mode (S107).

If it is determined in step S107 that the optimal mode is the horizontal prediction mode or the vertical prediction mode, the selection module 113 determines the intra prediction mode on a line basis (S109).

Then, when the line based on the intra prediction target of the current macroblock is the i-th line, the line-based prediction module 114 determines whether i is 1 (S111).

In this case, when i is 1, that is, for the first line of the current macroblock, the line-based prediction module 114 performs intra prediction by using the reconstructed pixels of the already encoded macroblock stored in the frame storage unit 170 as a reference pixel. It performs (S113).

On the other hand, when i is not 1, that is, from the second line of the current macroblock, the line-based prediction module 114 uses the reconstructed pixel for the previous line, i.e., the i-1th line, stored in the frame storage unit 170 as a reference pixel. Intra prediction is performed (S115).

The line-based prediction module 114 calculates a difference value between the predicted pixel value of the line predicted intra in step S113 or S115 and the pixel value of the corresponding line before the prediction (S117). Here, the difference value means a residual component predicted and remaining.

The line-based prediction module 114 generates the difference value calculated in step S117 as a 4 × 4 macroblock (S119) and inputs it to the converter 120.

The converter 120 inputs the DCT transform S121 to the quantizer 130 after performing the DCT transform S121 on the 4 × 4 macroblock generated in step S119.

The quantization unit 130 performs quantization on the DTC-converted macroblock in operation S121 (S123), and outputs the same to the decoder 180 and the inverse quantization unit 140.

In this case, the decoder 180 first calculates a coded block pattern (hereinafter, referred to as 'CBP') value for the quantized macro block (S125). At this time, the CBP value is calculated for each line of the current macro block and accumulated.

The encoder 190 uses the accumulated CBP value in the bit stream during entropy encoding, and generates an output image by encoding the quantized prediction block (S127).

Here, CBP is information on the entire macroblock.

When CBP is '0', AC coefficients after the transform and quantization of 16 lines in the 16 × 16 unit current macroblock are '0'.

In addition, when the CBP is not '0', the CBP is 15 when the AC coefficient is present in any of the 16 × 16 unit current macroblocks.

Thus, all sixteen 4x4 blocks are scanned for entropy encoding. The CBP is defined as a syntax element and transmitted to a decoder stage, that is, a video image decoding device (not shown).

Meanwhile, the inverse quantization unit 140 performs inverse quantization on the prediction block quantized by the quantization unit 130 (S129). The inverse transform unit 150 performs IDCT on the inverse quantized prediction block output by the inverse quantizer 140 (S131).

The IDCT predicted block passing through the filter 160 is stored in the frame storage unit 170. Then, the line-based prediction module 114 rearranges the prediction blocks stored in the frame storage unit 170 in the form of lines (S133) and updates the corresponding blocks of the current macroblock (S135).

The line-based prediction module 114 determines whether the updated line is the last line (S137).

At this time, if not the last line, the step after step S111 is repeated.

On the other hand, if it is the last line, the step ends.

Meanwhile, when the mode having the smallest SATD is the average prediction mode or the planar prediction mode in operation S107, the selection module 113 determines the intra-block intra prediction mode in operation S139.

Then, the block-based prediction module 115 performs intra prediction on the current macro block using the reconstructed pixel of the already encoded macro block stored in the frame storage unit 170 as a reference pixel (S141).

The transformer 120 performs a DCT on the prediction block intra predicted in step S141 (S143).

The quantization unit 130 performs quantization (S145) on the prediction block DCT in step S143 (S145).

In this case, the block-based prediction module 115 calculates the CBP (S147), and the encoder 190 uses the calculated CBP value in the bit stream during entropy encoding, and encodes and outputs the intra-predicted prediction block (S149). Create an image.

Here, the line-based intra prediction process will be described in more detail by dividing the intra prediction mode into the horizontal prediction mode and the vertical prediction mode as follows.

First, FIGS. 6 to 9 are diagrams for describing a line-based intra prediction process in a vertical mode according to an embodiment of the present invention, and FIGS. 10 to 13 are line-based intra prediction in a horizontal mode according to an embodiment of the present invention. It is a figure for demonstrating a process.

6 or 10, in a 16 × 16 unit current macroblock, a plurality of pixels form a matrix, 16 pixels are positioned in a row direction, and 16 pixels are positioned in a column direction. At this time, sixteen pixels in the row direction or the horizontal direction or the column direction or the vertical direction form one line.

In this case, referring to FIG. 6, intra prediction is performed using a reconstructed pixel of a macroblock that is already encoded with respect to a line, that is, the first line, as a reference pixel.

Referring to FIG. 7, the prediction block of FIG. 6 in a line form is converted into a 4 × 4 macroblock. After DCT and quantization of the 4 × 4 macroblock, the result is converted back into a line form as shown in FIG. 8. After the inverse quantization and the inverse DCT of the transformed prediction block of the line type, the original line, that is, the first line is rearranged and updated. The reconstructed pixel of the updated first line is used as a reference pixel in intra prediction for the second line as shown in FIG. 9.

Meanwhile, referring to FIG. 10, intra prediction is performed on the A line, that is, the first line, using the reconstructed pixel of the macro block already encoded as shown in FIG. 6 as a reference pixel. Referring to FIG. 11, a prediction block of a column direction line is converted into a 4 × 4 macroblock. After the DCT and the quantization of the 4 × 4 macroblock, the result is converted into a column direction line form again as shown in FIG. 12. After the inverse quantization and the inverse DCT of the transformed prediction block of the line type, the original line, that is, the first line is rearranged and updated. The updated reconstructed pixel of the first line is used as a reference pixel in intra prediction for the second line as shown in FIG. 13.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the invention is not limited to the disclosed exemplary embodiments, It belongs to the scope of right.

100: video image encoding apparatus 110: intra prediction unit
120: transform unit 130: quantization unit
140: inverse quantization unit 150: inverse transform unit
160: filter 170: frame storage unit
180: decoder 190: encoder

Claims (14)

In a method of encoding a video image,
16 × 16 current macroblock to be encoded in the video image, wherein the current macroblock includes a plurality of pixels in a matrix form, and a plurality of pixels in a row direction or a column direction are sequentially arranged in the row direction or the column direction. Performing intra prediction on a line-by-line basis for forming a plurality of lines listed as;
Accumulating a coded block pattern (CBP) value calculated for each line of the current macroblock; And
Encoding each intra predicted value for every line of the current macro block,
Wherein the encoding comprises:
Using the accumulated value of the code block pattern value in a bit stream during entropy encoding;
Performing the intra prediction,
The intra prediction of a specific line among a plurality of lines constituting the current macroblock uses a reconstructed pixel of a previous line of the specific line as a reference pixel. The method of claim 1,
Wherein the performing comprises:
Performing intra prediction on the specific line;
Decoding the intra predicted value for the specific line and updating the specific line; And
Performing the intra prediction and the update on a line following the particular line,
The reconstructed pixels updated in the specific line are used as reference pixels in intra prediction of the next line. The method of claim 2,
Before the step of performing,
Performing intra prediction using a reconstructed pixel of a macro block that is already encoded with respect to the first line of the current macro block as a reference pixel; And
Decoding the intra prediction value for the first line and updating the first line;
The performing of the encoding method comprises a target from a second line to a last line. The method of claim 3,
Wherein the updating comprises:
Calculating difference values between the intra prediction value and the pre-reconstructed pixel value of the corresponding line;
Generating the difference values into a 4x4 macroblock;
Performing Discrete Cosine Transform (DCT) and quantization on the 4x4 macroblock;
Performing inverse quantization and inverse discrete cosine transform on the quantized 4x4 macroblock;
Rearranging an inverse discrete cosine transformed 4x4 macroblock into a line form of the current macroblock; And
Updating the inverse discrete cosine transformed 4x4 macroblock rearranged in line form to the corresponding line
Encoding method comprising a. delete The method according to claim 3 or 4,
Before updating to the first line,
Performing prediction for all 16x16 intra modes for the current macroblock of type 16x16;
Calculating a Sum of Absolute Transform Difference (SATD) between a predicted value for each intra mode and the current macroblock before prediction;
Comparing SATDs for the respective intra modes with each other to determine whether a mode having the smallest value is a horizontal mode or a vertical mode; And
And performing the intra prediction and updating from the first line to the last line of the current macroblock when it is determined as the horizontal mode or the vertical mode. The method according to claim 6,
After the determining step,
Performing intra prediction using the reconstructed pixels of the macroblock, which are already coded as all the pixels of the current macroblock, as reference pixels, if the horizontal mode and the vertical mode do not correspond to both
Encoding method further comprising. An apparatus for encoding a video image,
16 × 16 current macroblock to be encoded in the video image, wherein the current macroblock includes a plurality of pixels in a matrix form, and a plurality of pixels in a row direction or a column direction are sequentially arranged in the row direction or the column direction. An intra prediction unit configured to perform intra prediction on a line-by-line basis, and to calculate and accumulate a coded block pattern (CBP) value for each line of the current macroblock; And
An encoding unit for encoding each intra predicted value for all lines of the current macroblock, and using the accumulated value of the code block pattern value in a bit stream during entropy encoding;
The intra-
Intra-prediction of a specific line from among a plurality of lines constituting the current macroblock uses a reconstructed pixel of a previous line of the specific line as a reference pixel. 9. The method of claim 8,
The intra-
After performing the intra prediction on the specific line, the intra predicted value is decoded and updated in the specific line, and used as a reference pixel in the intra prediction of the next line of the specific line by using the reconstructed pixels updated in the specific line. Video image encoding device. 10. The method of claim 9,
The intra-
Intra prediction is performed by using a reconstructed pixel of a macro block that is already coded with respect to the first line of the current macroblock as a reference pixel, and decodes the intra prediction value for the first line and updates the first line, and the second line is the last. A video image encoding apparatus using reconstructed pixels of a previous line decoded and updated after intra prediction as a reference pixel. The method of claim 10,
The intra-
After intra prediction on the corresponding line, the difference between the intra prediction value and the pre-reconstructed pixel value of the corresponding line is calculated and generated as a 4 × 4 macroblock, and discrete cosine transform is performed for the 4 × 4 macroblock. Transform, DCT), quantization, inverse quantization, and inverse discrete cosine transformation, and then rearranging into a line form of the current macroblock and updating the corresponding line of the current macroblock. delete delete The method according to any one of claims 8 to 11,
The intra-
An intra prediction module that performs prediction for all 16 × 16 intra modes for the current macroblock of 16 × 16 form;
A calculation module for calculating a sum of absolute transform transform (SATD) between a predicted value for each intra mode performed by the intra prediction module and the current macroblock before prediction;
A line based prediction module for performing intra prediction using the reconstructed pixel of the previous line as a reference pixel for each line of the current macro block;
A block-based prediction module configured to perform intra prediction by using all pixels of the current macro block as reconstruction pixels of an already encoded macro block as a reference pixel; And
The line-based prediction module determines whether the mode having the smallest value is the horizontal mode or the vertical mode by comparing the SATDs for the respective intra modes calculated by the arithmetic module with each other. A selection module for determining an operation of the controller and determining an operation of the block-based prediction module when the operation does not correspond to both the horizontal mode and the vertical mode.
Video encoding apparatus comprising a.

Images