KR102437718B1 - Method for quantization based on fast rate-distortion optimization, and apparatus for the same - Google Patents

Abstract

The present invention relates to a fast rate-distortion optimization-based quantization method performed by a fast rate-distortion optimization-based quantization apparatus. According to the present invention, the method comprising: initializing a value of first context information in a transform block to 1, and initializing a value of second context information in the transform block to 0 to initialize context information; scanning transform coefficients in the transform block and quantizing the transform coefficients; calculating a quantized level value of the transform coefficient and determining an optimal quantized level value for a position of some transform coefficient in the transform block or a subblock within the transform block; and determining an optimal last significant coefficient with respect to the position of the partial transform coefficient or the subblock, and applying scalar quantization using a dead zone to the position of the partial transform coefficient or the subblock. A fast rate-distortion optimization based quantization method is provided. According to the present invention, it is possible to reduce the complexity of quantization.

Description

Quantization method and apparatus based on fast rate-distortion optimization

The present invention relates to a quantization method and apparatus based on fast rate-distortion optimization.

Recently, as broadcasting services having a high definition (HD) resolution are expanding not only in Korea but also worldwide, many users are getting used to high-resolution and high-definition images.

Accordingly, many organizations are spurring the development of next-generation imaging devices. In addition, as interest in Ultra High Definition (UHD) having a resolution four times higher than that of HDTV increases along with HDTV, a compression technology for higher resolution and high-definition images is required.

In addition, for image compression, inter prediction technology for predicting pixel values included in the current picture from temporally previous or subsequent pictures, predicting pixel values included in the current picture using pixel information in the current picture Intra prediction techniques may be used. Also, for image compression, an entropy encoding technique in which a short code is assigned to a symbol having a high frequency of occurrence and a long code is assigned to a symbol having a low frequency of occurrence may be used.

However, the existing technology has a problem in that the complexity of quantization is large. Accordingly, it is intended to propose a technique capable of reducing the complexity of quantization.

An object of the present invention is to solve all of the above problems.

Another object of the present invention is to provide for reducing the complexity of quantization.

Another object of the present invention is to reduce the complexity of rate-distortion optimization based quantization.

In order to achieve the object of the present invention as described above and to realize the characteristic effects of the present invention to be described later, the characteristic configuration of the present invention is as follows.

In the fast rate-distortion optimization-based quantization method performed by the fast rate-distortion optimization-based quantization apparatus according to an embodiment, the fast rate-distortion optimization-based quantization method performed by the fast rate-distortion optimization-based quantization apparatus includes: initializing the value of the first context information in the block to 1 and initializing the value of the second context information in the transformation block to 0 to initialize the context information; scanning transform coefficients in the transform block and quantizing the transform coefficients; calculating a quantized level value of the transform coefficient and determining an optimal quantized level value for a position of some transform coefficient in the transform block or a subblock within the transform block; and determining an optimal last significant coefficient with respect to the position of the partial transform coefficient or the subblock, and applying scalar quantization using a dead zone to the position of the partial transform coefficient or the subblock. may include

In the fast rate-distortion optimization-based quantization method performed by the fast rate-distortion optimization-based quantization apparatus according to another embodiment, the quantizing includes: scanning transform coefficients in the transform block in a reverse diagonal order and quantizing the transform coefficient by using a quantization offset value of 0.5.

In the fast rate-distortion optimization-based quantization method performed by the fast rate-distortion optimization-based quantization apparatus according to another embodiment, applying the scalar quantization includes: a rounding operator, a sign, a quantization step size, and The transform coefficient is mapped to a discrete quantized level using a quantization offset, and the dead zone is a section in which an input value is output as 0, and the quantization offset may include adjusting a section of the dead zone. .

In the fast rate-distortion optimization-based quantization method performed by the fast rate-distortion optimization-based quantization apparatus according to another embodiment, the step of determining the optimal quantized level value comprises: the last significant coefficient in the transform block fixing; and calculating rate-distortion costs for the non-zero remaining transform coefficients.

In the fast rate-distortion optimization-based quantization method performed by the fast rate-distortion optimization-based quantization apparatus according to another embodiment, the determining of the optimal last significant coefficient includes: an initial rate-distortion with respect to the transform block initializing to a cost value; and scanning in an inverse diagonal order using the quantized level values in the transform block and calculating a rate-distortion cost.

In the fast rate-distortion optimization-based quantization apparatus according to an embodiment, the fast rate-distortion optimization-based quantization apparatus initializes a value of first context information in a transform block to 1 and second context information in the transform block a context information initialization unit initializing the context information by initializing a value of 0; a transform coefficient quantizer that scans the transform coefficients in the transform block and quantizes the transform coefficients; a quantization level determiner for calculating a quantized level value of the transform coefficient and determining an optimal quantized level value for a position of a partial transform coefficient within the transform block or a subblock within the transform block; and an optimal last significant coefficient determining unit that determines the position of the partial transform coefficient or an optimal last significant coefficient with respect to the subblock, wherein scalar quantization using a dead zone is applied to the position of the partial transform coefficient or the subblock. It may further include a scalar quantization unit.

In the fast rate-distortion optimization-based quantization apparatus according to another embodiment, the transform coefficient quantizer scans transform coefficients in the transform block in a reverse diagonal order, and uses a quantization offset value of 0.5 to transform the transform coefficient. quantizing the coefficients.

In the quantization apparatus based on fast rate-distortion optimization according to another embodiment, the scalar quantization unit maps the transform coefficient to a discrete quantized level using a rounding operator, a sign, a quantization step size, and a quantization offset, , the dead zone may be a section in which an input value is output as 0, and the quantization offset may include adjusting a section of the dead zone.

In the quantization apparatus based on fast rate-distortion optimization according to another embodiment, the quantization level determiner fixes a last significant coefficient in the transform block, and calculates rate-distortion costs for the remaining non-zero transform coefficients. may include doing

In the fast rate-distortion optimization-based quantization apparatus according to another embodiment, the coefficient determiner initializes the transform block to an initial rate-distortion cost value, and uses the quantized level value in the transform block to inverse diagonal This may include scanning in sequence and calculating the rate-distortion cost.

The present invention can provide for reducing the complexity of quantization. Therefore, the present invention is effective in reducing the complexity of quantization.

The present invention can provide for reducing the complexity of rate-distortion optimization based quantization. Therefore, the present invention is effective in reducing the complexity of rate-distortion optimization-based quantization.

1 is a flowchart illustrating a quantization method based on fast rate-distortion optimization according to an embodiment.
2 is a block diagram illustrating a quantization apparatus based on fast rate-distortion optimization according to an embodiment.
3 is a quantization apparatus based on high-speed rate-distortion optimization according to an embodiment, and shows a detailed configuration thereof.
4 is a diagram illustrating an image decoding apparatus to which a quantization method based on a fast rate-distortion optimization according to an embodiment is applied.
5 is a diagram schematically illustrating a split structure of an image when an image is encoded and decoded by a quantization method based on a high-speed rate-distortion optimization according to an embodiment.
6 is a diagram illustrating a form of a prediction unit that may be included in a coding unit to which a quantization method based on a fast rate-distortion optimization according to an embodiment is applied.
7 is a diagram illustrating a form of a transform unit that may be included in a coding unit to which a quantization method based on a fast rate-distortion optimization according to an embodiment is applied.
8 illustrates a step of determining an optimal quantized level value in a quantization method based on a fast rate-distortion optimization according to an embodiment.
9 is a diagram illustrating a step of determining an optimal last significant coefficient in a quantization method based on a fast rate-distortion optimization according to an embodiment.
10 is a diagram illustrating a transform block to which a quantization method based on a fast rate-distortion optimization according to an embodiment is applied.
11 illustrates a quantization method based on fast rate-distortion optimization according to an embodiment.
12 is a diagram illustrating a quantization method based on fast rate-distortion optimization using a transform block according to an embodiment.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 is a flowchart illustrating a quantization method based on fast rate-distortion optimization according to an embodiment.

According to an embodiment, an image encoder, which is a quantization apparatus based on fast rate-distortion optimization, performs quantization on a transform coefficient during an image encoding process to obtain a quantized transform coefficient level. In this case, quantizing the transform coefficient may reduce a bit depth of the coefficient. Also, the quantized transform coefficient level may be a quantized level.

An image encoder, which is a quantization device based on high-speed rate-distortion optimization, uses a scalar quantization method using a deadzone or a rate-distortion optimized quantization method when performing quantization. Can be used.

Also, the high-speed rate-distortion optimization-based quantization apparatus may optimize a tradeoff between a rate, which is a bit rate of encoded image data, and distortion, which is a difference between a reconstructed image and an original image.

Referring to FIG. 1 , the fast rate-distortion optimization-based quantization method performed by the fast rate-distortion optimization-based quantization apparatus includes the following steps. Of course, depending on the case, each step may be performed simultaneously or the order may be changed, but is not limited thereto.

In step S101, the fast rate-distortion optimization-based quantization apparatus initializes the value of the first context information in the transform block to 1, and initializes the value of the second context information in the transform block to 0 to initialize the context information. can

For example, an image encoder, which is a quantization apparatus based on fast rate-distortion optimization, initializes context information. In this case, the value of the context information C1 may be initialized to 1, and the value C2 may be initialized to 0.

In step S102, the fast rate-distortion optimization-based quantization apparatus may scan transform coefficients in a transform block and quantize the transform coefficients. In this case, fast rate-distortion optimization-based quantization in which transform coefficients within a transform block are scanned in a reverse diagonal order and the transform coefficients are quantized using a quantization offset value of 0.5 may be performed.

For example, an image encoder, which is a quantization apparatus based on fast rate-distortion optimization, may quantize a transform coefficient while scanning a transform coefficient within a transform block and calculate a quantized level value. In addition, an image encoder, which is a quantization apparatus based on fast rate-distortion optimization, scans transform coefficients in a transform coefficient block in a reverse diagonal scan order. In this case, the transform coefficient may be quantized using a quantization offset value of 0.5 while scanning the transform coefficient, and a quantized level value may be calculated.

In step S103, the fast rate-distortion optimization-based quantization apparatus calculates the quantized level value of the transform coefficient, and obtains an optimal quantized level value for the position of some transform coefficient in the transform block or the sub-block within the transform block. can decide

For example, an image encoder, which is a quantization apparatus based on fast rate-distortion optimization, may determine optimal quantized level values for transform coefficients in a transform coefficient block.

Specifically, the fast rate-distortion optimization-based quantization apparatus may fix the last significant coefficient in the transform block.

Next, the fast rate-distortion optimization-based quantization apparatus may calculate rate-distortion costs for the remaining non-zero transform coefficients.

In step S104 , the fast rate-distortion optimization-based quantization apparatus may determine an optimal last significant coefficient for a position or subblock of some transform coefficients.

For example, an image encoder, which is a quantization apparatus based on a fast rate-distortion optimization, may determine an optimal last significant coefficient.

Specifically, the fast rate-distortion optimization-based quantization apparatus may initialize a transform block with an initial rate-distortion cost value. Next, the fast rate-distortion optimization-based quantization apparatus may scan in an inverse diagonal order using the quantized level values in the transform block and calculate the rate-distortion cost.

In addition, in a step between steps S101 to S104, the fast rate-distortion optimization-based quantization apparatus may apply scalar quantization using a dead zone to positions or subblocks of some transform coefficients. Also, in some cases, in a step before step S101, the fast rate-distortion optimization-based quantization apparatus may apply scalar quantization using a dead zone to positions or subblocks of some transform coefficients.

In this case, the fast rate-distortion optimization-based quantization apparatus may map the transform coefficient to the discrete quantized level using a rounding operator, a sign, a quantization step size, and a quantization offset.

Of course, the dead zone may be a period in which an input value is output as 0, and the quantization offset may adjust the period of the dead zone.

For example, the fast rate-distortion optimization-based quantization apparatus may map transform coefficients to discrete quantized levels using Equation 1 below.

In this case, operator · is an operation for rounding to the nearest integer value, and Δ is the quantization step size. Also, f is the quantization offset, and the function sign() outputs the sign for the input value W. In this case, the quantization offset f may adjust the section of the dead zone, and z may be a quantized level value.

In addition, when an image decoder, which is a quantization device based on high-speed rate-distortion optimization, performs inverse quantization, Equation 2 below may be used.

In this case, W' may be an inverse quantized coefficient, and Z may be a quantized level. The inverse quantization method performed by the high-speed rate-distortion optimization-based quantization apparatus may be used not only for the quantized level to which the rate-distortion optimization-based quantization method is applied, but also for the quantized level to which the scalar quantization method using the dead zone is applied.

2 is a block diagram illustrating a quantization apparatus based on fast rate-distortion optimization according to an embodiment.

2, the fast rate-distortion optimization-based quantization apparatus 200 includes a context information initializer 210, a transform coefficient quantizer 220, a quantization level determiner 230, and an optimal last significant coefficient determiner ( 240), and a scalar quantization unit 250 may be included. In this case, the context information initializer 210 , the transform coefficient quantizer 220 , the quantization level determiner 230 , the optimal last significant coefficient determiner 240 , and the scalar quantizer 250 include an electronic circuit, an electric circuit, and an integrated circuit. It may be configured to include a circuit. In addition, the context information initializer 210 , the transform coefficient quantizer 220 , the quantization level determiner 230 , the optimal last significant coefficient determiner 240 , and the scalar quantizer 250 include a processor, a memory, and a data transceiver. It may be configured to include, but is not limited thereto.

The context information initializer 210 may initialize the context information by initializing the value of the first context information in the transform block to 1 and the value of the second context information in the transform block to 0.

The transform coefficient quantization unit 220 may scan a transform coefficient within a transform block and quantize the transform coefficient. Also, the transform coefficient quantizer 220 may scan transform coefficients in a transform block in a reverse diagonal order, and quantize the transform coefficients using a quantization offset value of 0.5.

The quantization level determiner 230 may calculate a quantized level value of a transform coefficient and determine an optimal quantized level value with respect to a position of some transform coefficient within a transform block or a subblock within the transform block. Also, the quantization level determiner 230 may fix the last significant coefficient in the transform block and calculate rate-distortion costs for the remaining non-zero transform coefficients.

The optimal last significant coefficient determiner 240 may determine an optimal last significant coefficient with respect to the position or subblock of some transform coefficients. The optimal last significant coefficient determiner 240 may initialize the transform block to an initial rate-distortion cost value, scan the transform block in an inverse diagonal order using the quantized level value, and calculate the rate-distortion cost.

The scalar quantization unit 250 may apply scalar quantization using a dead zone to positions or subblocks of some transform coefficients. Also, the scalar quantization unit 250 may map the transform coefficient to the discrete quantized level using a rounding operator, a sign, a quantization step size, and a quantization offset. In this case, the dead zone may be a period in which an input value is output as 0, and the quantization offset may adjust the period of the dead zone.

3 is a quantization apparatus based on high-speed rate-distortion optimization according to an embodiment, and shows a detailed configuration thereof.

Referring to FIG. 3 , it is a block diagram showing the configuration of an image encoding apparatus, which is a quantization apparatus based on fast rate-distortion optimization. The image encoding apparatus 300 , which is a quantization apparatus based on high-speed rate-distortion optimization, may include a motion predictor 311 , a motion compensator 312 , an intra predictor 320 , and a switch 315 . In addition, the image encoding apparatus 300, which is a quantization apparatus based on high-speed rate-distortion optimization, includes a subtractor 325, a transform unit 330, a quantization unit 340, an entropy encoding unit 350, an inverse quantization unit 360, It may include an inverse transform unit 370 , an adder 375 , a filter unit 380 , and a reference picture buffer 390 .

The image encoding apparatus 300 , which is a quantization apparatus based on fast rate-distortion optimization, may perform encoding on an input image in an intra mode or an inter mode and output a bitstream. Also, in the case of the intra mode, the switch 315 may be switched to the intra mode, and in the case of the inter mode, the switch 315 may be switched to the inter mode. Also, the image encoding apparatus 300 , which is a quantization apparatus based on fast rate-distortion optimization, may generate a prediction block for an input block of an input image, and then encode a residual between the input block and the prediction block.

In the intra mode, the intra prediction unit 320 may generate a prediction block by performing spatial prediction using pixel values of an already encoded block around the current block. Also, in the inter mode, the motion prediction unit 311 may find a region that best matches the input block in the reference image stored in the reference picture buffer 390 during the motion prediction process to obtain a motion vector. Also, the motion compensator 312 may generate a prediction block by performing motion compensation using a motion vector. In this case, the motion vector is a two-dimensional vector used for inter prediction, and may indicate an offset between the current encoding and decoding target image and the reference image.

The subtractor 325 may generate a residual block by the difference between the input block and the generated prediction block.

The transform unit 330 may perform transform on the residual block to output a transform coefficient. Here, the transform coefficient may mean a coefficient value generated by performing transform on the residual block or the residual signal. Of course, in some cases, a quantized transform coefficient level generated by applying quantization to a transform coefficient may also be referred to as a transform coefficient.

The quantization unit 340 may quantize the input transform coefficient according to the quantization parameter and output a quantized transform coefficient level. In this case, the quantization unit 340 may quantize the input transform coefficient using a quantization matrix. In this case, the quantized transform coefficient level may be a quantized level, or may be a part of the transform coefficient in some cases.

The entropy encoding unit 350 may output a bit stream by performing entropy encoding based on the values calculated by the quantization unit 340 or the encoding parameter values calculated during the encoding process.

In this case, when entropy encoding is applied, a small number of bits are allocated to a symbol having a high probability of occurrence and a large number of bits are allocated to a symbol having a low probability of occurrence, so that the symbol can be expressed. Of course, the size of the bit stream for the encoding target symbols may be reduced. In addition, compression performance of image encoding may be improved through entropy encoding.

In addition, the entropy encoder 350 derives a binarization method of the target symbol and a probability model of the target symbol or bin, and then arithmetic encoding using the derived binarization method or probability model. can also be performed. Also, the entropy encoder 350 may use an encoding method such as exponential golomb, Context-Adaptive Variable Length Coding (CAVLC), or Context-Adaptive Binary Arithmetic Coding (CABAC) for entropy encoding.

Since the image encoding apparatus 300, which is a quantization apparatus based on a high rate rate-distortion optimization according to an embodiment, performs inter prediction encoding, that is, inter prediction encoding, the currently encoded image needs to be decoded and stored to be used as a reference image. there is In addition, the quantized level is inverse quantized by the inverse quantization unit 360 and inversely transformed by the inverse transform unit 370 . In addition, the inverse-quantized and inverse-transformed coefficients are added to the prediction block through the adder 375 and a reconstructed block is generated.

The reconstructed block passes through the filter unit 380, and the filter unit 380 applies at least one of a deblocking filter, a sample adaptive offset (SAO), and an adaptive loop filter (ALF) to the reconstructed block or the reconstructed picture. can do. In this case, the filter unit 380 may be an adaptive in-loop filter. In addition, the deblocking filter can remove block distortion generated at the boundary between blocks. The SAO may add an appropriate offset value to a pixel value to compensate for a coding error. The ALF may perform filtering based on a value obtained by comparing the reconstructed image and the original image. The reconstructed block passing through the filter unit 380 may be stored in the reference picture buffer 390 .

4 is a diagram illustrating an image decoding apparatus to which a quantization method based on a fast rate-distortion optimization according to an embodiment is applied.

Referring to FIG. 4 , the image decoding apparatus 400 to which the quantization method based on the fast rate-distortion optimization is applied includes an entropy decoding unit 410 , an inverse quantization unit 420 , an inverse transform unit 430 , and an intra prediction unit 440 . , a motion compensator 450 , an adder 455 , a filter unit 460 , and a reference picture buffer 470 .

The image decoding apparatus 400 to which the high-speed rate-distortion optimization-based quantization method is applied receives the bitstream output by the encoder, which is a fast rate-distortion optimization-based quantization apparatus, as an input, performs decoding in intra mode or inter mode, and decodes the reconstructed image , that is, the restored image may be output. Specifically, in the intra mode, the switch may be switched to the intra mode, and in the inter mode, the switch may be switched to the inter mode.

Also, the image decoding apparatus 400 to which the quantization method based on the fast rate-distortion optimization is applied may obtain a reconstructed residual block from the received bitstream and generate a prediction block. Next, the image decoding apparatus 400 to which the quantization method based on the fast rate-distortion optimization is applied may generate a reconstructed block that is a reconstructed block by adding the reconstructed residual block and the prediction block.

The entropy decoding unit 410 entropy-decodes the input bitstream according to a probability distribution to generate symbols including symbols in the form of quantized levels. Also, the entropy decoding method may be performed as a reverse process of the entropy encoding.

The quantized level is inverse quantized by the inverse quantization unit 420 and inversely transformed by the inverse transformation unit 430 . As a result of inverse quantization and inverse transformation of the quantized level, a reconstructed residual block may be generated. In this case, the inverse quantizer 420 may apply a quantization matrix to the quantized level.

In the intra mode, the intra prediction unit 440 may generate a prediction block by performing spatial prediction using pixel values of an already encoded block around the current block. In the inter mode, the motion compensator 450 may generate a prediction block by performing motion compensation using a motion vector and a reference image stored in the reference picture buffer 470 .

The reconstructed residual block and the prediction block are added by the adder 455 , and the added block may pass through the filter unit 460 . The filter unit 460 may apply at least one of a deblocking filter, SAO, and ALF to a reconstructed block or a reconstructed picture. The filter unit 460 may output a reconstructed image, that is, a reconstructed image. In some cases, the reconstructed image may be stored in the reference picture buffer 470 and used for inter prediction.

5 is a diagram schematically illustrating a split structure of an image when an image is encoded and decoded by a quantization method based on a high-speed rate-distortion optimization according to an embodiment.

The high-speed rate-distortion optimization-based quantization apparatus may perform encoding and decoding with a coding unit (CU) to efficiently segment an image. The unit may be configured by combining a syntax element and a block including image samples. In addition, a unit may be divided, and in this case, a block corresponding to the unit may be divided depending on the case.

Referring to FIG. 5 , the high-speed rate-distortion optimization-based quantization apparatus sequentially divides an image 500 in units of a largest coding unit (LCU) in HEVC, and then determines a division structure in units of LCUs. . In some cases, the partition structure may indicate a distribution of encoding units (CUs) for efficiently encoding an image in the LCU 510 .

In this case, the fast rate-distortion optimization-based quantization apparatus may determine according to whether to divide the distribution of CUs into four CUs reduced by half of the horizontal and vertical sizes of one CU. Of course, the divided CU may be recursively divided into four CUs whose horizontal size and vertical size are reduced by half with respect to the divided CU in the same manner.

In addition, the fast rate-distortion optimization-based quantization apparatus may recursively divide the division of a CU to a predefined depth. In this case, the depth information is information indicating the size of a CU, and may be stored for each CU.

For example, the depth of the LCU may be 0, and the depth of the Smallest Coding Unit (SCU) may be a predefined maximum depth. Of course, the LCU may be a coding unit having the largest coding unit size as described above, and the Smallest Coding Unit (SCU) may be a coding unit having the smallest coding unit size.

The depth of the CU increases by 1 whenever division is performed from the LCU 510 to half of the horizontal and vertical sizes. Also, for each depth, a CU that does not perform division has a size of 2Nx2N, and a CU that performs division can be divided into four CUs having a size of NxN from a CU of 2Nx2N size. In this case, the size of N may be reduced by half whenever the depth increases by 1.

Also, for example, the size of an LCU having a minimum depth of 0 may be 64x64 pixels, and a size of an SCU having a maximum depth of 3 may be 8x8 pixels. In this case, a CU (LCU) of 64x64 pixels may be expressed as depth 0, a CU of 32x32 pixels as depth 1, a CU of 16x16 pixels as depth 2, and a CU (SCU) of 8x8 pixels as depth 3 may be expressed.

In addition, information on whether to split a specific CU may be expressed through 1-bit split information for each CU. This partition information may be included in all CUs except the SCU, and in some cases, when the fast rate-distortion optimization-based quantization apparatus does not split a CU, 0 may be stored in the partition information, and the fast rate-distortion optimization-based quantization When the device divides a CU, 1 may be stored in the division information.

6 is a diagram illustrating a form of a prediction unit that may be included in a coding unit to which a quantization method based on a fast rate-distortion optimization according to an embodiment is applied.

Referring to FIG. 6 , the shape of the prediction unit PU that the encoding unit CU may include may be known. The fast rate-distortion optimization-based quantization apparatus may divide a CU that is not divided any more among CUs divided from an LCU into one or more prediction units.

In this case, the prediction unit is a basic unit for performing prediction, and is encoded and decoded by a fast rate-distortion optimization-based quantization apparatus as any one of a skip mode, an inter mode, and an intra mode, and each Depending on the mode, it may be partitioned in various forms. For example, in the case of the skip mode, the fast rate-distortion optimization-based quantization apparatus may support the 2Nx2N mode 410 having the same size as the CU without a partition in the CU.

In addition, in the case of the inter mode, the fast rate-distortion optimization-based quantization apparatus may support eight partitioned types in a CU. For example, the fast rate-distortion optimization-based quantization apparatus includes a 2Nx2N mode 410, a 2NxN mode 415, an Nx2N mode 420, an NxN mode 425, a 2NxnU mode 430, a 2NxnD mode 435, An nLx2N mode 440 and an nRx2N mode 445 may be supported. Also, in the case of the intra mode, the fast rate-distortion optimization-based quantization apparatus may support the 2Nx2N mode 410 and the NxN mode 425 in the CU.

7 is a diagram illustrating a form of a transform unit that may be included in a coding unit to which a quantization method based on a fast rate-distortion optimization according to an embodiment is applied.

Referring to FIG. 7 , a form of a transform unit (TU) that the encoding unit (CU) may include may be known. In this case, the transform unit may be a basic unit used for transform, quantization, inverse transform, and inverse quantization processes in the CU. Also, the TU may be in the form of a square or rectangle.

A CU that is no longer split among CUs divided from an LCU by the fast rate-distortion optimization-based quantization apparatus may be divided into one or more TUs by the fast rate-distortion optimization-based quantization apparatus. In this case, the partition structure of the TU may be a quad-tree structure. For example, in some cases, one CU 710 may be divided one or more times according to a quadtree structure to be configured with TUs of various sizes.

8 illustrates a step of determining an optimal quantized level value in a quantization method based on a fast rate-distortion optimization according to an embodiment.

Referring to FIG. 8 , the step of determining the optimal quantized level value among the fast rate-distortion optimization-based quantization methods performed by the fast rate-distortion optimization-based quantization apparatus is specifically performed as follows.

In step S801 , the image encoder, which is a quantization device based on fast rate-distortion optimization, may fix the last significant coefficient in the transform block. In this case, the significant coefficient may be a quantized level value having a non-zero value, and the last significant coefficient may be present at the last position among quantized levels having a non-zero value in the transform block. Also, the last significant coefficient may be a first coefficient whose quantized level value is not 0 when the transform coefficients are quantized while scanning in an inverse diagonal scan order.

In step S801, the video encoder, which is a quantization device based on fast rate-distortion optimization, applies the quantized levels of each transform coefficient to the remaining transform coefficients whose values are non-zero except for the last significant coefficient in the transform block. We can calculate the rate-distortion cost for In this case, the image encoder, which is a quantization apparatus based on a fast rate-distortion optimization, determines an optimal quantization level based on the calculated rate-distortion cost. Of course, the remaining transform coefficients may be transform coefficients occurring after the last significant coefficient in the inverse diagonal scan order.

Specifically, the video encoder, which is a quantization apparatus based on fast rate-distortion optimization, first determines whether a quantized level value for a corresponding transform coefficient is less than 3 in order to determine an optimal quantized level for one transform coefficient. do.

At this time, when the quantized level value of the corresponding transform coefficient is less than 3, the image encoder, which is a quantization apparatus based on fast rate-distortion optimization, replaces the transform coefficient with a Level value, a Level-1 value, and a value of 0. Then we can calculate the rate-distortion cost for each. In this case, the Level value may be the quantized level value calculated in step 2 above. In addition, when the quantized level value of the corresponding transform coefficient is greater than 3, the image encoder, which is a quantization apparatus based on fast rate-distortion optimization, replaces the corresponding transform coefficient with a Level value and a value of 0, and then rate-distortions for each of the transform coefficients. cost can be calculated.

An image encoder, which is a quantization device based on fast rate-distortion optimization, may update rate-distortion cost. In this case, the image encoder, which is a quantization apparatus based on high-speed rate-distortion optimization, may update rate-distortion costs for the case of encoding 0 as quantized level values for all transform coefficients in a transform block. In addition, an image encoder, which is a quantization apparatus based on a high-speed rate-distortion optimization, may update a rate-distortion cost for encoding each transform coefficient in a transform block. Also, an image encoder, which is a quantization apparatus based on fast rate-distortion optimization, may update a rate-distortion cost for encoding a significant map of a transform block. In this case, the significance map may be information on whether a quantized level value for each transform coefficient in the transform block is 0 or not.

An image encoder, which is a quantization apparatus based on fast rate-distortion optimization, may update context information on an optimal quantized level for a transform coefficient. Also, the image encoder, which is a quantization apparatus based on fast rate-distortion optimization, may determine a quantized level having the smallest rate-distortion cost among quantized levels as an optimal quantized level. In this case, the updated context information may be used when determining a quantized level for the next transform coefficient.

Also, in some cases, the image encoder, which is a quantization apparatus based on fast rate-distortion optimization, may perform the above step S802 on the remaining non-zero transform coefficients in the transform block.

9 is a diagram illustrating a step of determining an optimal last significant coefficient in a quantization method based on a fast rate-distortion optimization according to an embodiment.

Referring to FIG. 9 , the step of determining the most optimal last important coefficient among the fast rate-distortion optimization-based quantization methods performed by the fast rate-distortion optimization-based quantization apparatus may be performed as follows.

In step S901, the video encoder, which is a fast rate-distortion optimization-based quantization device, and a fast rate-distortion optimization-based quantization device, converts the rate-distortion cost value, d64BestCost, into a rate-distortion cost value when the transform block is not coded. can be initialized.

In step S902, the video encoder, which is a fast rate-distortion optimization-based quantization device, scans in an inverse diagonal scan order until a quantized level value in a transform block meets a transform coefficient greater than 1, and reduces the rate-distortion cost. can be calculated

Specifically, the image encoder, which is a quantization device based on high-speed rate-distortion optimization, considers a transform coefficient having a quantized level value greater than 1 as the last significant coefficient while scanning according to step S902, and rate-distortion cost for the entire transform block. You can calculate the totalCost which is .

Next, when totalCost is less than d64BestCost, the video encoder, which is a quantization device based on fast rate-distortion optimization, may set the position of the last significant coefficient to iBestLastIdxP1 indicating the optimal last position indicator, and set d64BestCost to totalCost.

In this case, when an image encoder, which is a quantization apparatus based on high-speed rate-distortion optimization, calculates the rate-distortion cost, the rate-distortion cost may be calculated using Equation 3 below. D may be a mean square error of difference values between original transform coefficients and reconstructed transform coefficients in a transform block, and R may be a bit rate using related context information.

10 is a diagram illustrating a transform block to which a quantization method based on a fast rate-distortion optimization according to an embodiment is applied.

According to an embodiment, the high-speed rate-distortion optimization-based quantization apparatus may reduce the complexity of rate-distortion optimization-based quantization. In this case, the quantization apparatus based on fast rate-distortion optimization may determine an optimal quantized level and an optimal last significant coefficient only for some transform coefficient positions or subblocks within a transform block.

In addition, the high-speed rate-distortion optimization-based quantization apparatus applies the scalar quantization method using the dead zone to only some transform coefficient positions or subblocks within the transform block, and performs rate-distortion optimization-based quantization according to the applied result to the entire transform block. It can provide a way to apply it.

Referring to FIG. 10 , the transform block may have a size of 4x4, 8x8, 16x16, 32x32, etc. according to the transform size. In this case, the fast rate-distortion optimization-based quantization apparatus may use an inverse diagonal scan, an inverse horizontal scan, or an inverse vertical scan to scan each transform block.

In addition, a transform block having a size of 8x8 or larger may be divided into 4x4 sub-blocks. 9 shows an 8x8 transform block having a 4x4 transform block and four 4x4 unit subblocks. In the 4x4 transform block, transform coefficients or quantized levels may be scanned in an inverse diagonal scan order as shown. In addition, in an 8x8 transform block, transform coefficients or quantized levels in subblocks may be scanned in an inverse diagonal scan order as shown, and scans between subblocks in a transform block may also be scanned in an inverse diagonal scan order as shown. have. In this case, the scan method within the transform block and the subblock and the scan between the subblocks are not limited to the inverse diagonal scan method, and the aforementioned scan methods such as inverse horizontal scan and inverse vertical scan may also be used. Of course, rate-distortion optimization-based quantization may be performed in units of transform blocks by the high-speed rate-distortion optimization-based quantization apparatus.

11 illustrates a quantization method based on fast rate-distortion optimization according to an embodiment.

When the transform is performed by the high-speed rate-distortion optimization-based quantization apparatus, the transform coefficient is generated at the upper left side in the transform block, and the transform coefficient generated at the upper left side after the transform may have a relatively low frequency. Also, the transform coefficient generated at the lower right side may have a high frequency. Of course, the low-frequency component has a greater effect on the encoding efficiency than the high-frequency component, so that the low-frequency component can play a more important role than the high-frequency component in the image encoder and decoder, which is a quantization device based on fast rate-distortion optimization.

Referring to FIG. 11 , a total of 16 transform coefficients may exist in a 4x4 transform block, a total of 64 transform coefficients in an 8x8 transform block, a total of 256 transform coefficients in a 16x16 transform block, and a total of 1024 transform coefficients in a 32x32 transform block may exist.

In this case, when a rate-distortion optimization-based quantization method performed by a high-speed rate-distortion optimization-based quantization apparatus is used, an optimal quantized level and an optimal last important coefficient for a maximum of 1024 transform coefficients for a 32x32 transform block must be determined. . Of course, in some cases, since the complexity of the image encoder, which is a quantization apparatus based on fast rate-distortion optimization, may be significantly increased, the fast rate-distortion optimization-based quantization apparatus can prevent this.

A fast rate-distortion optimization-based quantization apparatus, in order to reduce the number of optimal quantized levels and optimal last significant coefficients to be determined in the transform block, some transform coefficient positions 1101 or subblocks 1102 in the transform block. ) can only determine the optimal quantized level and the optimal last significant coefficient.

For example, the fast rate-distortion optimization-based quantization apparatus calculates an optimal quantized level and an optimal last significant coefficient only for transform coefficients existing in some transform coefficient positions in a transform block or some sub-blocks according to the size of the transform block. can decide In this case, the quantization apparatus based on fast rate-distortion optimization may set all quantized level values to 0 for some transform coefficient positions in a transform block or transform coefficients that do not correspond to some sub-blocks.

Also, in some cases, the fast rate-distortion optimization-based quantization apparatus may determine an optimal quantized level and an optimal last significant coefficient only for some transform coefficient positions 1101 for an 8x8 transform block. Also, in some cases, the fast rate-distortion optimization-based quantization apparatus may determine an optimal quantized level and an optimal last significant coefficient for only some subblocks 1102 for a 16×16 transform block.

Of course, the fast rate-distortion optimization-based quantization device can determine the optimal quantized level and the optimal last significant coefficient only for transform coefficients existing in some transform coefficient positions or some subblocks, similarly to the 4x4 transform block and the 32x32 transform block. have. In addition, the high-speed rate-distortion optimization-based quantization device, for an 8x8, 16x16, or 32x32 transform block, has two each from the last scan order according to the inverse horizontal scan, inverse vertical scan, or inverse diagonal scan order, which is the inter-subblock scan order, The above method may be performed only for 9 and 35 subblocks.

12 is a diagram illustrating a quantization method based on fast rate-distortion optimization using a transform block according to an embodiment.

According to an embodiment, the fast rate-distortion optimization-based quantization apparatus optimizes only the transform coefficients present in some transform coefficient positions in the transform block or some sub-blocks according to the intra prediction mode of the prediction unit including the transform block. It is possible to determine the quantized level and the optimal last significant coefficient. The fast rate-distortion optimization-based quantization apparatus provides an optimal quantized level and optimal quantization level only for two subblocks present at the top of an 8x8 transform block when the intra prediction mode is a vertical mode or a mode having a similar direction to the vertical mode. The last important factor can be determined. In addition, when the intra prediction mode is a horizontal mode or a mode having a similar direction to the horizontal mode, the quantization apparatus based on the fast rate-distortion optimization provides an optimal quantized level only for the two subblocks present on the left side of the 8x8 transform block. and an optimal last critical coefficient.

According to an embodiment, a quantization apparatus based on fast rate-distortion optimization performs optimal quantization only for transform coefficients existing in some transform coefficient positions or some subblocks in a transform block according to a prediction mode of a coding unit including the transform block. It is possible to determine the level and the optimal last important factor.

Also, when the prediction mode is the intra prediction mode, more transform coefficients may exist in a relatively low frequency region than in the inter prediction mode. In this case, in the case of the intra prediction mode, the fast rate-distortion optimization based quantization apparatus may perform the above method for more transform coefficient positions or subblocks in the low frequency region than in the inter prediction mode.

According to an embodiment, the fast rate-distortion optimization-based quantization apparatus provides an optimal quantized level and An optimal last critical factor can be determined. Since the luminance component is more likely to have transform coefficients than the chrominance component, the fast rate-distortion optimization-based quantization apparatus has more transform coefficient positions or subblocks in the transform block for the luminance component than in the transform block for the chrominance component. The above method can be performed.

Also, the high-speed rate-distortion optimization-based quantization apparatus does not apply rate-distortion optimization-based quantization to the entire transform block as in the conventional rate-distortion optimization-based quantization method. A quantization apparatus based on high-speed rate-distortion optimization applies a scalar quantization method using a dead zone mainly to a low-frequency component, which is a relatively important component in a transform block, and then, when a quantized level value for the low-frequency component exists, rate-distortion optimization Apply based quantization. Also, when a quantized level value for a low frequency component does not exist, the fast rate-distortion optimization-based quantization apparatus may determine a quantized level value for all transform coefficients in a corresponding transform block as 0.

In this case, the fast rate-distortion optimization-based quantization apparatus applies the scalar quantization method using the dead zone only to some transform coefficient positions or sub-blocks in the transform block, and when there is a quantized level value in the transform coefficient position or sub-block Quantization based on rate-distortion optimization can be applied to In addition, the quantization apparatus based on fast rate-distortion optimization may determine a quantized level value of all transform coefficients in a corresponding transform block as 0 when a quantized level value does not exist in the corresponding transform coefficient position or subblock.

Referring to FIG. 12 , a transform coefficient position 1201 or subblock 1202 to which a scalar quantization method using a dead zone is to be applied may be determined according to the size of a transform block by a quantization apparatus based on a fast rate-distortion optimization. In this case, the fast rate-distortion optimization-based quantization apparatus may apply the scalar quantization method using the dead zone to the position 1201 or the sub-block 1202 and determine whether to apply the rate-distortion optimization-based quantization to the entire transform block.

According to an embodiment, the transform coefficient position 1201 or subblock 1202 to which the scalar quantization method using the dead zone is to be applied may be determined according to the prediction mode of the coding unit by the fast rate-distortion optimization-based quantization apparatus. For example, when the prediction mode is the intra prediction mode, more transform coefficients may exist in a relatively low frequency region than in the inter prediction mode. In this case, the high-speed rate-distortion optimization-based quantization apparatus, in the intra-prediction mode, places more transform coefficient positions or sub-blocks in the low-frequency region than the inter-prediction mode into the region to which the scalar quantization method using the dead zone is to be applied. can decide

Also, according to an embodiment, the transform coefficient position 1201 or subblock 1202 to which the scalar quantization method using the dead zone is to be applied may be determined according to a luminance component or a chrominance component that is a color component. In this case, since the luminance component is more likely to have a transform coefficient than the chrominance component, the fast rate-distortion optimization-based quantization apparatus generates more transform coefficient positions or sub-blocks in the transform block for the luminance component than in the transform block for the chrominance component. It may be determined as a region to which a scalar quantization method using a dead zone is to be applied.

The device described above may be implemented as a hardware component, a software component, and/or a combination of the hardware component and the software component. For example, devices and components described in the embodiments may include, for example, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable array (FPA), It may be implemented using one or more general purpose or special purpose computers, such as a programmable logic unit (PLU), microprocessor, or any other device capable of executing and responding to instructions. The processing device may execute an operating system (OS) and one or more software applications running on the operating system. The processing device may also access, store, manipulate, process, and generate data in response to execution of the software. For convenience of understanding, although one processing device is sometimes described as being used, one of ordinary skill in the art will recognize that the processing device includes a plurality of processing elements and/or a plurality of types of processing elements. It can be seen that may include For example, the processing device may include a plurality of processors or one processor and one controller. Other processing configurations are also possible, such as parallel processors.

The software may comprise a computer program, code, instructions, or a combination of one or more thereof, which configures a processing device to operate as desired or is independently or collectively processed You can command the device. The software and/or data may be any kind of machine, component, physical device, virtual equipment, computer storage medium or device, to be interpreted by or to provide instructions or data to the processing device. , or may be permanently or temporarily embody in a transmitted signal wave. The software may be distributed over networked computer systems and stored or executed in a distributed manner. Software and data may be stored in one or more computer-readable recording media.

The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded in a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, etc. alone or in combination. The program instructions recorded on the medium may be specially designed and configured for the embodiment, or may be known and available to those skilled in the art of computer software. Examples of the computer-readable recording medium include magnetic media such as hard disks, floppy disks and magnetic tapes, optical media such as CD-ROMs and DVDs, and magnetic such as floppy disks. - includes magneto-optical media, and hardware devices specially configured to store and execute program instructions, such as ROM, RAM, flash memory, and the like. Examples of program instructions include not only machine language codes such as those generated by a compiler, but also high-level language codes that can be executed by a computer using an interpreter or the like. The hardware devices described above may be configured to operate as one or more software modules to perform the operations of the embodiments, and vice versa.

As described above, although the embodiments have been described with reference to the limited embodiments and drawings, various modifications and variations are possible by those skilled in the art from the above description. For example, the described techniques are performed in a different order than the described method, and/or the described components of the system, structure, apparatus, circuit, etc. are combined or combined in a different form than the described method, or other components Or substituted or substituted by equivalents may achieve an appropriate result. Therefore, other implementations, other embodiments, and equivalents to the claims are also within the scope of the following claims.

Claims (3)

In the video decoding method,
obtaining a residual signal for a current transform block;
performing inverse quantization on the residual signal of the current transform block; and
performing an inverse transform on the inverse quantized signal of the current transform block,
The residual signal is obtained only for a partial region of the current transform block,
The image decoding method, characterized in that the partial region is determined based on the size of the current transform block.
In the video encoding method,
deriving a residual signal for a current transform block;
performing transform on the residual signal of the current transform block;
performing quantization on the transformed signal of the current transform block; and
encoding the quantized signal of the current transform block;
Only a partial region of the quantized signal of the current transform block is encoded,
The image encoding method, characterized in that the partial region is determined based on the size of the current transform block.
A computer-readable recording medium storing a bitstream generated by an image encoding method, comprising:
The video encoding method comprises:
deriving a residual signal for a current transform block;
performing transform on the residual signal of the current transform block;
performing quantization on the transformed signal of the current transform block; and
encoding the quantized signal of the current transform block;
Only a partial region of the quantized signal of the current transform block is encoded,
The partial area is characterized in that determined based on the size of the current transformation block, computer-readable recording medium.

Images