KR101419689B1 - Decoding apparatus for performing inverse quantiation \and inverse transform in h.264/avc decoder and decoding method using decoding apparatus - Google Patents

Abstract

The present invention reduces the control complexity by performing inverse transform after performing inverse quantization irrespective of the DC coefficient and the AC coefficient included in the compressed image information received from the encoder and processing the inverse quantization operation using a common operator, And an inverse quantization and inverse transform in an H.264 / AVC decoder capable of reducing an inverse quantization error and a decoding method using the same.
A decoding apparatus for performing inverse quantization and inverse transform in an H.264 / AVC decoder according to an embodiment of the present invention restores frequency components of DC (Direct Current) coefficients and AC (Alternate Current) coefficients in a quantized frequency component An inverse quantization (IQ) operation unit; An inverse transform (IT) operation unit for inversely transforming the frequency component reconstructed by the inverse quantization unit into a pixel component; A rounding unit for removing a rounding error preventing coefficient multiplied by the inverse quantization in the inverse quantization operation unit; And a decoding control unit for controlling the macroblock scan order of the image input according to the DC coefficient or the AC coefficient, the presence / absence of inverse Hadamard transformation, and the timing of the inverse quantization and inverse transformation.

Description

[0001] The present invention relates to a decoding apparatus for performing inverse quantization and inverse transform in an H.264 / AVC decoder, and a decoding method using the same, and a decoding method using the same. BACKGROUND ART < RTI ID = 0.0 >

The present invention relates to a decoding apparatus for inverse quantization and inverse transform and a decoding method using the same, and more particularly, to a decoding apparatus for performing inverse quantization without performing DC quantization and AC coefficient included in compressed image information received from an encoder, To a decoding apparatus for performing inverse quantization and inverse transformation in an H.264 / AVC decoder capable of reducing computational complexity by reducing the control complexity and processing the inverse quantization operation using a common operator, and a decoding method using the same will be.

H.264 / AVC is a digital video codec standard with a very high data compression rate. It is developed to obtain a low bit rate at the same picture quality compared with the existing standards MPEG-1, MPEG-2 and MPEG-4 Part 2 .

H.264 / AVC uses motion compensation in 4x4 block units, motion prediction in 1 / 4pixel units, enhanced entropy coding, integer-based conversion, deblocking filter, etc. to improve video compression performance.

In particular, since the integer-based conversion of H.264 / AVC uses integer unit operations, it is possible to solve the problem of mismatch of the conversion coefficients between the encoder and the decoder, and the multiplication operation of the conversion is integrated into the quantization, This is possible.

While the H.264 / AVC has higher compressibility and flexibility than the existing video compression standard, the complexity of the encoder and the decoder can be increased compared to the existing standard. In H.264 / AVC encoder, the parameter and predictive coding modes are increased and the computational complexity can be increased due to the addition of integer-based DCT and quarter-pixel motion compensation and deblocking filter.

Therefore, efficient hardware design is needed to reduce the computational complexity of H.264 / AVC.

In connection with the technology for encoding and decoding such images, the disclosure of Japanese Patent Application Laid-Open No. 10-2010-0000066 discloses a decoding process by organically connecting a predetermined unit of partial decoding process necessary for decoding a bitstream encoded in various forms A decoding process can be configured adaptively based on an encoding format of input data.

However, as shown in FIG. 9, in the case of decoding the image information using the conventional technique, the luminance and chrominance components corresponding to the AC coefficients are subjected to inverse transform (IT) after performing inverse quantization (IQ) In the intra 16X16 mode, the luminance and chrominance components corresponding to the DC coefficient are inversely transformed and inverse quantized.

That is, since the execution order of the DC coefficient and the AC coefficient is different, there is a problem that complexity of execution and control can be increased when decryption is implemented.

Disclosure of Invention Technical Problem [8] The present invention has been conceived to solve the problems described above, and it is an object of the present invention to provide an apparatus and a method for performing inverse quantization after inverse quantization regardless of a DC coefficient and an AC coefficient to reduce control complexity, The present invention also provides a decoding apparatus and a decoding method for performing inverse quantization and inverse transformation on compressed image information received from an encoder capable of reducing computation complexity in an H.264 / AVC decoder.

Also, a decoding apparatus for performing inverse quantization and inverse transform in an H.264 / AVC decoder capable of reducing the number of execution cycles required for processing a macroblock by constituting an inverse transformer and an inverse quantizer in a three-stage pipeline, and a decoding apparatus And to provide a decoding method.

However, the object of the present invention is not limited to the above-mentioned objects, and other objects not mentioned can be clearly understood by those skilled in the art from the following description.

In order to achieve the above object, a decoding apparatus for performing inverse quantization and inverse transform in an H.264 / AVC decoder according to an embodiment of the present invention includes a direct current (DC) coefficient and an alternate current ) Inverse quantization (IQ) operation unit for recovering a frequency component of a coefficient; An inverse transform (IT) operation unit for inversely transforming the frequency component reconstructed by the inverse quantization unit into a pixel component; A rounding unit for removing a rounding error preventing coefficient multiplied by the inverse quantization in the inverse quantization operation unit; And a decoding control unit for controlling the macroblock scan order of the image input according to the DC coefficient or the AC coefficient, the presence / absence of inverse Hadamard transformation, and the timing of the inverse quantization and inverse transformation.

Also, the decoding apparatus according to the present invention is characterized in that the dequantization operation unit, the inverse transformation operation unit, and the rounding unit are configured as a three-stage pipeline.

The decoding apparatus according to the present invention is characterized in that the dequantization operation unit multiplies a quantized coefficient by a quantization parameter (QP) value and a scaling coefficient, and restores the orthogonally transformed component.

Also, the decoding apparatus according to the present invention is characterized in that the scaling coefficient is stored as an LUT (Lookup Table).

In the decoding apparatus according to the present invention, the inverse quantization operation unit generates a shift length by dividing the quantization parameter (QP) by 6 and performs a left shift operation on the scaling coefficient using the generated shift length A global quantization (Pre_IQ) module for calculating a LevelScale value; A Multi module for multiplying the DC coefficient and the AC coefficient by a calculated level scale value; And a post-IQ module for performing a right shift operation after the inverse transform operation of the inverse transform operation unit.

Also, the decoding apparatus according to the present invention is characterized in that the inverse quantization operation unit performs inverse quantization on the AC coefficient using an inverse quantization relation of the following 4x4 AC coefficients.

(Wherein, W _{D (i, j)} are inverse quantized transform coefficients, Z _{QD (i, j)} is the quantized transform coefficient, V _{(i, j)} represents a scaling factor)

The decoding apparatus according to the present invention is characterized in that the inverse quantization operation unit performs inverse quantization on the luminance DC coefficient and inverse quantization on the color difference DC coefficient.

Also, the decoding apparatus according to the present invention is characterized in that the inverse quantization operation unit performs inverse quantization on the luminance DC coefficient using an inverse quantization relation of the following 4x4 luminance DC coefficient.

,

,

(Here, QP is a quantization parameter, W _{D (i, j)} are inverse quantized transform coefficients, Z _{QD (i, j)} is the quantized transform coefficient, V _{(i, j)} represents a scaling factor)

Also, the decoding apparatus according to the present invention is characterized in that the inverse quantization operation unit performs inverse quantization on the chrominance DC coefficient using an inverse quantization relation of the 2x2 color difference DC coefficient.

,

,

(Here, QP is a quantization parameter, W _{D (i, j)} are inverse quantized transform coefficients, Z _{QD (i, j)} is the quantized transform coefficient, V _{(i, j)} represents a scaling factor)

The decoding apparatus according to the present invention is characterized in that the decoding control section divides a luminance component into a macroblock encoded in the intra 16x16 mode and a macroblock encoded in the intra 4x4 mode or the inter mode.

In addition, the decoding apparatus according to the present invention may further comprise an inverse Hadamard Transform (IHDT) process for the macroblocks coded in the intra 16x16 mode separated through the decoding control unit, IDCT, and Inverse Discrete Cosine Transform).

In a decoding method using a decoding apparatus for performing inverse quantization and inverse transform in an H.264 / AVC decoder according to an embodiment of the present invention, a decoding controller divides a frequency component of the compressed image information by a direct current (DC) Alternate Current) coefficients; A second step of the inverse quantization unit performing inverse quantization (IQ) on the DC coefficient and the AC coefficient classified in the first step to recover a frequency component; A third step of performing an inverse transform (IT) on the frequency component restored in the second step to transform the frequency component into a pixel component; And a fourth step of outputting the pixel component converted in the third step.

According to the H.264 / AVC decoder for inverse quantization and inverse transformation of the compressed image information received from the encoder of the present invention and the decoding method using the same, inverse quantization is performed irrespective of DC coefficient and AC coefficient, There is an advantage in that the computational complexity can be reduced by reducing the complexity and processing the inverse quantization operation using the pain operator.

In addition, according to the present invention, the number of execution cycles required for processing a macroblock can be reduced by configuring the inverse transformer and the inverse quantizer as a three-stage pipeline.

In particular, it takes 253 cycles to process 4x4 macroblocks encoded in the intra 16x16 mode, whereas the decoding apparatus according to the present invention requires 124 cycles, and takes 244 cycles in the intra 4x4 mode and the inter mode, In the present invention, 124 cycles were required and the function was improved to 49% as compared with the conventional method.

1 is a block diagram schematically illustrating a decoding apparatus for performing inverse quantization and inverse transform in an H.264 / AVC decoder according to the present invention.
2 is a block diagram schematically showing an inverse quantization operation unit constituting a decoding apparatus according to the present invention.
3 is an exemplary diagram showing orthogonal transformation of a luminance component in H.264 / AVC.
4 is an exemplary diagram illustrating a processing cycle for the calculation of the DC coefficient and the AC coefficient according to the present invention.
5 is an exemplary diagram showing a configuration of a macroblock.
FIG. 6 is a diagram illustrating a processing cycle of the three-stage pipeline structure according to the present invention.
7 is a flowchart illustrating a decoding method using a decoding apparatus that performs inverse quantization and inverse transform in an H.264 / AVC decoder according to the present invention.
8 is a flowchart showing an inverse quantization process according to the present invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, a detailed description of preferred embodiments of the present invention will be given with reference to the accompanying drawings. In the following description of the present invention, detailed description of known functions and configurations incorporated herein will be omitted when it may make the subject matter of the present invention rather unclear.

Embodiments in accordance with the concepts of the present invention can make various changes and have various forms, so that specific embodiments are illustrated in the drawings and described in detail in this specification or application. It is to be understood, however, that it is not intended to limit the embodiments according to the concepts of the present invention to the particular forms of disclosure, but includes all modifications, equivalents, and alternatives falling within the spirit and scope of the invention.

It is to be understood that when an element is referred to as being "connected" or "connected" to another element, it may be directly connected or connected to the other element, . On the other hand, when an element is referred to as being "directly connected" or "directly connected" to another element, it should be understood that there are no other elements in between. Other expressions that describe the relationship between components, such as "between" and "between" or "neighboring to" and "directly adjacent to" should be interpreted as well.

The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The singular expressions include plural expressions unless the context clearly dictates otherwise. In this specification, the terms "comprises ",or" having ", or the like, specify that there is a stated feature, number, step, operation, , Steps, operations, components, parts, or combinations thereof, as a matter of principle.

FIG. 1 is a block diagram schematically illustrating a decoding apparatus for performing inverse quantization and inverse transform in an H.264 / AVC decoder according to the present invention. FIG. 2 is a block diagram schematically showing an inverse quantization operation unit constituting a decoding apparatus according to the present invention. .

Generally, since there is almost no change in the image and the monotonous portion has a very high correlation between the pixel and the pixel, when a block is encoded in a block unit, much energy is concentrated on the DC (direct current) coefficient. Therefore, as shown in FIG. 3, in the normalization of H.264 / AVC, a 4x4 discrete cosine transform (DCT) process is performed after encoding a monotonous portion in the intra 16x16 mode and the DC coefficient obtained in the discrete cosine transform Hadamard transform (HDT) processing can be further performed to enhance the compression effect.

Therefore, a 16x16 macroblock is divided into 16 4x4 blocks and an integer discrete cosine transform is performed. In the case of a macroblock coded in the intra 16x16 mode, the Hadamard transform can be performed by constructing 16 DC coefficients existing at (0, 0) position into one 4x4 block.

1 and 2, the decoding apparatus 10 of the present invention includes an inverse quantization (IQ) operation unit 100, an inverse transform (IT) operation unit 200, a rounding unit 300 And a decoding control unit 400. [

The inverse quantization operation unit 100 can recover the frequency components of a direct current (DC) coefficient and an alternate current (AC) coefficient from a frequency component quantized by an encoder (not shown). In this case, the dequantization operator 100 may multiply the quantized coefficients by a quantization parameter (QP) value at the time of encoding and a scaling coefficient (V) stored in a lookup table (LUT) to restore orthogonally transformed components .

2, the dequantization operator 100 includes a three-stage configuration of a pre-IQ module 110, a multipl module 120, and a post-IQ module 130, . ≪ / RTI >

The total dequantization module 110 generates a shift length by dividing the quantization parameter QP by 6 and a level scale LevelScale uses the generated shift length to perform a left shift operation on the scaling coefficient V . The multiplication module 120 may multiply the input quantized DC coefficient and the AC coefficient by the level scale value calculated by the global quantization module 110. The post-quantization module 130 may multiply the dequantization operation unit 100, And may be performed after the inverse transformation operation result of the inverse transform operation unit 200. [

In particular, the dequantization process of H.264 / AVC through the inverse quantization operation unit 100 may be performed differently depending on the luminance DC coefficient, chrominance DC coefficient, and AC coefficient. In order to recover unencoded luminance conversion coefficients and chrominance AC coefficients in the intra 16X16 mode, the inverse quantization of 4X4 AC coefficients can be performed using the following relation.

[Formula 1]

Here, W _{D (i, j)} represents an inversely quantized transform coefficient, Z _{QD (i, j)} represents a quantized transform coefficient, and V _{(i, j)} represents a scaling coefficient.

Further, when restoring the DC transform coefficients encoded in the intra 16X16 mode, when the quantization parameter QP is greater than or equal to 12, the inverse quantization of the 4X4 luminance DC coefficient can be performed using the following relational expression.

[Formula 2]

On the other hand, when the quantization parameter value is smaller than 12, the inverse quantization of the 4X4 luminance DC coefficient can be performed using the following relational expression.

[Formula 3]

When the quantization parameter value is equal to or greater than 6 in the inverse quantization of the chrominance DC coefficient, the inverse quantization of the 2X2 chrominance DC coefficient can be performed using the following relational expression.

[Formula 4]

On the other hand, if the quantization parameter value is smaller than 6, the inverse quantization of the 2X2 color difference DC coefficient can be performed using the following relational expression.

[Formula 5]

In addition, the inverse transform operation unit 200 may include a process of inversely transforming the coefficients of the frequency component reconstructed by the inverse quantization operation unit 100 into pixel components and reconstructing the original image.

The frequency component can be converted into the pixel component through inverse transformation of the macroblock coded in the intra 16X16 mode. That is, after restoring the 4 × 4 DC block to 16 DC coefficients through inverse Hadamard Transform (IHDT), the recovered DC coefficients are added to the (0,0) position of the AC coefficient. Also, the combined AC coefficients can be reconstructed as residual data through an inverse discrete cosine transform (IDCT).

The integer IDCT of H.264 / AVC can be obtained by the following relation.

[Formula 6]

Where Y is a dequantized transform coefficient matrix, E _i is a discrete scale matrix for generating an integer IDCT, and C _i is an integer IDCT matrix C ^T is a transpose matrix of C _i . As shown in Equation 6, the integer IDCT of H.264 / AVC approximates the IDCT to perform an integer unit operation, so that the results before and after the orthogonal transformation are completely matched and the computation after the decimal point is unnecessary, can do. In addition, the core portion of the orthogonal transformation can be implemented only by addition and shift operations, and the multiplication process as a part of the inverse transformation process can be integrated in the inverse quantization operation to reduce the total number of multiplications.

The rounding unit 300 may perform a process of removing 64 multiplied in order to prevent a rounding error at the time of performing quantization through the inverse quantization operation unit 100 on pixel components reconstructed by the inverse transformation operation unit 200 . The rounding unit 300 may implement a right shift operation to remove 64 from the pixel component to reduce computational complexity.

The decoding control unit 400 may decode the encoded mode of the macroblock to divide the macroblock into a macroblock encoded in the intra 16x16 mode and a macroblock encoded in the intra 4x4 mode or the inter mode, Block scan order can be controlled.

The decoding control unit 400 outputs Hadamard inverse transform performance information for carrying out an inverse transformation process of the luminance DC coefficient and the coefficient information obtained by dividing the luminance DC coefficient and the luminance AC coefficient for the processing of the macroblock coded in the intra 16x16 mode, The quantization operation unit 100 and the inverse transformation operation unit 200 can be controlled. As shown in FIG. 4, the timing of the inverse quantization unit 100 and the inverse transformation unit 200 are controlled so that the inverse quantization process and the inverse transformation process can be performed simultaneously, and the luminance order and the scan order of the chrominance coefficients are subjected to luminance control .

In particular, as shown in FIG. 4, the decoding apparatus 10 according to the present invention can reduce the complexity by performing inverse quantization after performing inverse quantization irrespective of the DC coefficient and the AC coefficient.

In addition, the dequantization operation unit 100, the inverse transformation operation unit 200, and the rounding unit 300 according to the present invention are configured as a three-stage pipeline, thereby reducing the entire execution cycle.

As shown in FIG. 5, one macroblock may be composed of Cb and Cr representing the luminance and the chrominance representing the brightness of the image. The macroblocks to be processed when performing the inverse quantization and inverse transform can be classified into a macroblock coded in the intra 16x16 mode and a macroblock coded in the intra 4x4 mode or the inter mode.

The blocks coded in the intra 16x16 mode consist of one luminance 4x4 DC block, 16 luminance 4x4 AC blocks, one chrominance Cb 2x2 DC block, four Cb 4x4 AC blocks, one chrominance Cr 2x2 DC block and four Cr 4x4 blocks And a macroblock encoded in the intra 4x4 mode or the inter mode may be composed of blocks other than one 4x4 DC block in the intra 16x16 mode.

Table 1 shows the number of cycles required for 4x4 / 2x2 blocks in each step.

Generally, the number of cycles required to decode one macroblock coded in the intra 16x16 mode at the time of inverse quantization and inverse transform is calculated as follows.

It takes nine cycles to process one 4x4 luminance DC coefficient, four cycles to process two chrominance DC coefficients Cb / Cr, 240 cycles to process 24 4x4 luminance AC coefficients and 4x4 chrominance AC coefficients.

Therefore, it takes 253 cycles to process one macroblock.

Also, calculating the number of cycles for processing one macroblock coded in the intra 4x4 mode or the inter mode processes four cycles, 24 4x4 luminance AC coefficients and 4x4 chrominance AC coefficients to process two chrominance DC coefficients Cb / Cr , It takes 240 cycles, and it takes 244 cycles to process one macroblock.

Therefore, as shown in Fig. 6, by employing the three-stage pipeline structure according to the present invention, there is a feature that the number of processing cycles of the macroblock can be reduced.

6 (a) shows that 5 cycles are required to process a 4x4 luminance AC coefficient when a 3-stage pipeline is processed, and 5 cycles are less than 10 cycles when a 3-stage pipeline structure is not used. As shown in Fig. 6 (b), it takes 5 cycles to process the 4x4 luminance DC coefficient, which means that 4 cycles are reduced from 9 cycles when the 3-stage pipeline structure is not employed.

FIG. 7 is a flowchart illustrating a decoding method using a decoding apparatus for performing inverse quantization and inverse transform in the H.264 / AVC decoder according to the present invention, and FIG. 8 is a flowchart illustrating an inverse quantization process according to the present invention.

Referring to the drawings, in the decoding method using the decoding apparatus 10 according to the present invention, a frequency component of compressed image information is classified into a direct current (DC) coefficient and an alternate current (AC) coefficient by the decoding control unit 400 S101), the inverse quantization operation unit 200 performs inverse quantization (IQ) on the classified DC coefficient and the AC coefficient to recover the frequency component (S102).

At this time, the reconstruction of the frequency component may be performed by multiplying the quantized coefficient by a quantization parameter (QP) value and a scaling factor (V) to reconstruct the orthogonally transformed component.

The inverse transform unit 200 performs an inverse transform (IT) on the reconstructed frequency component in the inverse quantization unit 100 to convert the reconstructed frequency component into a pixel component (S103) (S104).

The inverse quantization in step S102 generates a shift length by dividing the quantization parameter QP by 6 through a global quantization (Pre_IQ) module 110 and generates a shift length by dividing the quantization parameter QP by 6 A left shift operation is performed to calculate a level scale value (S201).

The Mult (Multi) module 120 multiplies the DC coefficient and the AC coefficient by the level scale value calculated by the global quantization module 110 (S202).

Further, the post-quantization (Post_IQ) can perform a right shift operation after the inverse transform operation through the post-quantization module 130 (S203).

In particular, the inverse quantization and inverse transform process according to the present invention is characterized in that the entire execution cycle can be reduced through the configuration of the three-stage pipeline.

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 embodiments, but, on the contrary, is intended to cover various modifications and equivalent arrangements included within the spirit and scope of the appended claims. will be. Accordingly, the true scope of the present invention should be determined by the technical idea of the appended claims.

10: Decoding apparatus 100: Inverse quantization operation unit
110: a full quantization (Pre_IQ) module 120: a multiplication module
130: post-inverse quantization (Post_IQ) module 200:
300: Rounding unit 400: Decoding control unit

Claims (19)

A decoding apparatus for performing inverse quantization and inverse transform in an H.264 / AVC decoder,
An inverse quantization (IQ) operation unit for restoring a frequency component of a direct current (DC) coefficient as a luminance component and an AC (Alternate Current) coefficient as a chrominance component in a quantized frequency component;
An inverse transform (IT) operation unit for inversely transforming the frequency component reconstructed by the inverse quantization operation unit into a pixel component;
A rounding unit for removing a rounding error preventing coefficient multiplied by the inverse quantization in the inverse quantization operation unit; And
And a decoding controller for controlling a macroblock scan order of the image input according to the DC coefficient or the AC coefficient, determining whether there is Hadamard inverse transform, inverse quantization and inverse transformation,
The inverse quantization operation unit multiplies a quantization parameter (QP) value by a scaling coefficient stored in an LUT (Lookup Table)
Wherein the inverse quantization operation unit includes:
(Pre_IQ) module for generating a shift length by a quotient obtained by dividing a quantization parameter (QP) by 6 and calculating a level scale value by performing a left shift operation on the scaling coefficient using the generated shift length A Multi module for multiplying the DC coefficient and the AC coefficient by the calculated level scale value, and a post-IQ module for performing a right shift operation after the inverse transform operation of the inverse transform operation part,
Wherein the decoding control unit includes:
And a macroblock coded in an intra 16x16 mode and a macroblock coded in an intra 4x4 mode or an inter mode with respect to a luminance component, and the inverse transform operation unit performs inverse Hadamard transform (IHDT) Hadamard Transform), and performs inverse discrete cosine transform (IDCT).
The method according to claim 1,
Wherein the inverse quantization operation unit, the inverse transform operation unit, and the rounding unit are configured in a three-stage pipeline.
delete delete delete The method according to claim 1,
Wherein the inverse quantization unit performs inverse quantization on the AC coefficient using the inverse quantization relation of the 4x4 AC coefficients.

( _{I, j)} is the scaling factor, and floor () is the transformed value of (), where W _{D (i, j)} is the dequantized transform coefficient, Z _{QD (i, j)} is the quantized transform coefficient, QD is the quantized data, QP is the quantization parameter, and D is the data (dequantized data)
delete The method according to claim 1,
Wherein the inverse quantization unit performs inverse quantization on the luminance DC coefficient using an inverse quantization relation of the following 4x4 luminance DC coefficient.

,

(Here, QP is a quantization _parameter, W _{D (i, j)} are inverse quantized transform _{coefficients,} Z _{QD (i, j)} is the quantized transform _coefficient, V _{(i, j)} is the scaling factor, floor () is QD is the quantized data, and D is the data (inverse-quantized data).
The method according to claim 1,
Wherein the inverse quantization unit performs inverse quantization on chrominance DC coefficients using an inverse quantization relation of the following 2x2 chrominance DC coefficients.

,

(Here, QP is a quantization _parameter, W _{D (i, j)} are inverse quantized transform _{coefficients,} Z _{QD (i, j)} is the quantized transform _coefficient, V _{(i, j)} is the scaling factor, floor () is (QD) represents quantized data, and D represents data (dequantized data). In this case, delete delete delete delete delete delete delete delete delete delete

Images