CN111953999B

CN111953999B - Inverse transformation method and device

Info

Publication number: CN111953999B
Application number: CN202010851690.7A
Authority: CN
Inventors: 罗昆
Original assignee: Beijing QIYI Century Science and Technology Co Ltd
Current assignee: Beijing QIYI Century Science and Technology Co Ltd
Priority date: 2020-07-31
Filing date: 2020-08-21
Publication date: 2022-03-04
Anticipated expiration: 2040-08-21
Also published as: CN111953999A

Abstract

The embodiment of the invention provides an inverse transformation method and a device, which relate to the technical field of video coding and decoding, wherein the method comprises the following steps: determining the maximum ordinate value and the maximum abscissa value of a non-zero coefficient block CG in a frequency domain transformation coefficient matrix of an image block in a video frame; determining a target inverse transformation mode according to the size relation between the maximum ordinate value and the maximum abscissa value; according to a target inverse transformation mode, performing Inverse Discrete Cosine Transform (IDCT) on each first coefficient in a target inverse transformation unit with a target coordinate value not being 0 to obtain an inverse transformation result corresponding to each first coefficient; and setting the inverse transformation result corresponding to the second coefficient to be 0. By applying the scheme provided by the embodiment of the invention to carry out inverse transformation on the video data, the accuracy of the determined non-zero CG in the IDCT process is improved, and the calculated amount in the IDCT process is further reduced.

Description

Inverse transformation method and device

Technical Field

The present invention relates to the field of video encoding and decoding technologies, and in particular, to an inverse transform method and an inverse transform device.

Background

In the process of video encoding and decoding, a frequency domain Transform coefficient matrix is obtained by sequentially performing prediction, Discrete Cosine Transform (DCT), quantization, entropy coding and inverse quantization on image blocks in a video frame. After obtaining the frequency domain Transform coefficient matrix, IDCT (Inverse Discrete Cosine Transform) is also performed on the frequency domain Transform coefficient matrix. The coefficients in the frequency domain transform coefficient matrix are divided into CGs (coefficient blocks), and after the IDCT is performed, the inverse transform result corresponding to the coefficient included in each CG in the frequency domain transform coefficient matrix is obtained. When the image block is larger, the coefficients in the frequency domain transform coefficient matrix are more, the number of CG is also more, and the calculation amount consumed in the IDCT process is larger. For example, if the image block is a 32 × 32 image block, the frequency domain transform coefficient matrix includes 1024 coefficients, and if each CG includes 4 × 4 coefficients, the image block includes 64 CGs; if the image block is a 64 × 64 image block, the frequency domain transform matrix includes 4096 coefficients, and if each CG includes 2 × 2 coefficients, the image block includes 1024 CGs.

Since all-zero CGs in which the coefficients included in the frequency domain transform coefficient matrix are all 0 exist, the inverse transform result corresponding to the coefficients included in these all-zero CGs is often also 0. Therefore, in the related art, IDCT is generally performed only on CG in a fixed region preset in the frequency domain transform coefficient matrix, and the inverse transform result corresponding to a coefficient included in CG in another region is directly set to 0. For example, the frequency domain transform matrix is a 64 × 64 matrix including 16 × 16 CGs, and a region of 8 × 8 CGs at the upper left corner of the frequency domain transform matrix may be used as the fixed region.

However, the position of the non-zero CG containing the coefficient whose value is not 0 in the frequency domain transform coefficient matrix is not fixed, so that the fixed region may contain all-zero CGs or may be difficult to contain all the non-zero CGs, and therefore, the accuracy of the non-zero CGs determined in the process of performing the IDCT on the frequency domain transform coefficient matrix by applying the above method is low.

Disclosure of Invention

The embodiment of the invention aims to provide an inverse transformation method and device to improve the accuracy of non-zero CG determined in the IDCT process. The specific technical scheme is as follows:

in a first aspect of the present invention, there is provided an inverse transform method, including:

determining the maximum ordinate value and the maximum abscissa value of a non-zero coefficient block CG in a frequency domain transformation coefficient matrix of an image block in a video frame;

determining a target inverse transformation mode according to the magnitude relation between the maximum ordinate value and the maximum abscissa value, wherein the target inverse transformation mode is as follows: representing an inverse transformation mode of a one-dimensional horizontal IDCT and a one-dimensional vertical IDCT execution sequence in an IDCT process;

according to the target inverse transformation mode, performing IDCT on each first coefficient in a target inverse transformation unit with a target coordinate value not being 0 to obtain an inverse transformation result corresponding to each first coefficient, wherein the target inverse transformation unit is as follows: the target inverse transformation mode represents a CG row or a CG column which is subjected to IDCT preferentially, and when a target inverse transformation unit is a CG row, the target coordinate value is as follows: the maximum ordinate value of the non-zero CG in the target inverse transformation unit, the first coefficient being: the longitudinal coordinate value in the target inverse transformation unit is less than or equal to a coefficient included in the CG of the target coordinate value, and when the target inverse transformation unit is a CG column, the target coordinate value is: the maximum abscissa value of the non-zero CG in the target inverse transformation unit, the first coefficient is: a coefficient included in CG, whose abscissa value is equal to or less than the target coordinate value, in the target inverse transformation unit;

setting an inverse transformation result corresponding to a second coefficient to be 0, wherein the second coefficient is: and transforming the coefficients except the first coefficient in the coefficient matrix of the frequency domain.

In one embodiment of the present invention, when the maximum ordinate value is equal to or greater than the maximum abscissa value, the target inverse transformation method is: in the characterization process, a one-dimensional horizontal IDCT is firstly carried out, and then an inverse transformation mode of a one-dimensional vertical IDCT is carried out;

the performing IDCT on each first coefficient in the target inverse transform unit with the target coordinate value not being 0 according to the target inverse transform mode to obtain an inverse transform result corresponding to each first coefficient includes:

performing one-dimensional horizontal IDCT on the first coefficient in the CG row of which each target longitudinal coordinate value is not 0 to obtain an intermediate value matrix, wherein the target longitudinal coordinate value is as follows: the maximum longitudinal coordinate value of the non-zero CG in the CG line, and the intermediate value matrix is as follows: the number of included CG rows is the same as the frequency domain transform coefficient matrix, the number of included CG columns is the same as the frequency domain transform coefficient matrix, the initial element value of each element is a 0 matrix, and the element value of the element corresponding to each first coefficient in the intermediate value matrix is: one-dimensional horizontal inverse transformation results of the respective first coefficients;

and performing one-dimensional vertical IDCT on coefficients contained in the CG, of which the abscissa values in each CG row are less than or equal to the maximum abscissa value, in the intermediate value matrix to obtain an inverse transformation result corresponding to each first coefficient.

In an embodiment of the present invention, when the maximum ordinate value is smaller than the maximum abscissa value, the target inverse transformation method is: in the characterization process, a one-dimensional vertical IDCT is firstly carried out, and then a one-dimensional horizontal IDCT inverse transformation mode is carried out;

performing one-dimensional vertical IDCT on a first coefficient in a CG row of which each target abscissa value is not 0 to obtain an intermediate value matrix, wherein the target abscissa value is: the maximum abscissa value of non-zero CG in the CG row, and the intermediate value matrix is as follows: the number of included CG rows is the same as the frequency domain transform coefficient matrix, the number of included CG columns is the same as the frequency domain transform coefficient matrix, the initial element value of each element is a 0 matrix, and the element value of the element corresponding to each first coefficient in the intermediate value matrix is: a one-dimensional vertical inverse transform result of each first coefficient;

and performing one-dimensional horizontal IDCT on coefficients contained in the CG, of which the vertical coordinate value in each CG row is less than or equal to the maximum vertical coordinate value, in the intermediate value matrix to obtain an inverse transformation result corresponding to each first coefficient.

In one embodiment of the present invention, the size of the CG is one of the following sizes:

1×1、2×2、4×4、8×8、16×16、32×32、64×64。

in a second aspect of the present invention, there is also provided an inverse transform apparatus, the apparatus including:

the coordinate value determining module is used for determining the maximum ordinate value and the maximum abscissa value of a non-zero coefficient block CG in a frequency domain transformation coefficient matrix of an image block in a video frame;

a mode determining module, configured to determine a target inverse transformation mode according to a magnitude relationship between the maximum ordinate value and the maximum abscissa value, where the target inverse transformation mode is: representing an inverse transformation mode of a one-dimensional horizontal IDCT and a one-dimensional vertical IDCT execution sequence in an IDCT process;

a result obtaining module, configured to perform IDCT on each first coefficient in a target inverse transformation unit with a target coordinate value different from 0 according to the target inverse transformation manner, to obtain an inverse transformation result corresponding to each first coefficient, where the target inverse transformation unit is: the target inverse transformation mode represents a CG row or a CG column which is subjected to IDCT preferentially, and when a target inverse transformation unit is a CG row, the target coordinate value is as follows: the maximum ordinate value of the non-zero CG in the target inverse transformation unit, the first coefficient being: the longitudinal coordinate value in the target inverse transformation unit is less than or equal to a coefficient included in the CG of the target coordinate value, and when the target inverse transformation unit is a CG column, the target coordinate value is: the maximum abscissa value of the non-zero CG in the target inverse transformation unit, the first coefficient is: a coefficient included in CG, whose abscissa value is equal to or less than the target coordinate value, in the target inverse transformation unit;

a result setting module, configured to set an inverse transformation result corresponding to a second coefficient to 0, where the second coefficient is: and transforming the coefficients except the first coefficient in the coefficient matrix of the frequency domain.

the result obtaining module comprises:

a first matrix obtaining submodule, configured to perform one-dimensional horizontal IDCT on a first coefficient in a CG row where each target ordinate value is not 0, to obtain an intermediate value matrix, where the target ordinate value is: the maximum longitudinal coordinate value of the non-zero CG in the CG line, and the intermediate value matrix is as follows: the number of included CG rows is the same as the frequency domain transform coefficient matrix, the number of included CG columns is the same as the frequency domain transform coefficient matrix, the initial element value of each element is a 0 matrix, and the element value of the element corresponding to each first coefficient in the intermediate value matrix is: one-dimensional horizontal inverse transformation results of the respective first coefficients;

and the first result obtaining submodule is used for carrying out one-dimensional vertical IDCT on coefficients contained in the CG of which the abscissa in each CG row is less than or equal to the maximum abscissa value in the intermediate value matrix to obtain an inverse transformation result corresponding to each first coefficient.

the result obtaining module comprises:

a second matrix obtaining submodule, configured to perform one-dimensional vertical IDCT on a first coefficient in the CG column where each target abscissa value is not 0, to obtain an intermediate value matrix, where the target abscissa value is: the maximum abscissa value of non-zero CG in the CG row, and the intermediate value matrix is as follows: the number of included CG rows is the same as the frequency domain transform coefficient matrix, the number of included CG columns is the same as the frequency domain transform coefficient matrix, the initial element value of each element is a 0 matrix, and the element value of the element corresponding to each first coefficient in the intermediate value matrix is: a one-dimensional vertical inverse transform result of each first coefficient;

and the second result obtaining submodule is used for carrying out one-dimensional horizontal IDCT on the coefficients contained in the CG of which the longitudinal coordinate value in each CG row is less than or equal to the maximum longitudinal coordinate value in the intermediate value matrix to obtain the inverse transformation results corresponding to each first coefficient.

1×1、2×2、4×4、8×8、16×16、32×32、64×64。

in a third aspect of the present invention, an electronic device is provided, which includes a processor, a communication interface, a memory, and a communication bus, where the processor, the communication interface, and the memory complete communication with each other through the communication bus;

a memory for storing a computer program;

a processor for implementing the method steps of any of the first aspect when executing a program stored in the memory.

In yet another aspect of the present invention, there is also provided a computer-readable storage medium having a computer program stored therein, which when executed by a processor implements the method steps of any one of the above first aspects.

In a further aspect of the present invention, there is also provided a computer program product comprising instructions which, when run on a computer, cause the computer to perform the method steps of any one of the above first aspects.

As can be seen from the above, when performing inverse transform by applying the scheme provided by the embodiment of the present invention, a maximum ordinate value and a maximum abscissa value of a non-zero CG in a frequency domain transform coefficient matrix of an image block in a video frame are determined, a target inverse transform mode representing an execution order of a one-dimensional horizontal IDCT and a one-dimensional vertical IDCT in an IDCT process is determined according to a magnitude relationship between the maximum ordinate value and the maximum abscissa value, a CG row or a CG column which performs the IDCT preferentially according to the target inverse transform mode is used as a target inverse transform unit, the non-zero CG in the target inverse transform unit is determined, the IDCT is performed according to a coefficient included in the non-zero CG in the target inverse transform unit whose each target coordinate value is not 0, instead of performing the IDCT on a coefficient included in a fixed region by the CG. The coefficient actually used for IDCT is changed according to the specific position of the non-zero CG in the coefficient matrix of different frequency domain transformation, so that only the coefficient contained in each non-zero CG is accurately subjected to IDCT, and the inverse transformation results corresponding to the coefficients contained in other all-zero CGs are directly set to be 0, thereby improving the accuracy of the non-zero CG determined in the IDCT process and further reducing the calculation amount in the IDCT process.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below.

Fig. 1 is a schematic flowchart of a first inverse transformation method according to an embodiment of the present invention;

FIG. 2 is a diagram illustrating a first frequency domain transform coefficient matrix according to an embodiment of the present invention;

fig. 3 is a schematic flowchart of a second inverse transformation method according to an embodiment of the present invention;

FIG. 4A is a diagram illustrating a first intermediate value matrix according to an embodiment of the present invention;

fig. 4B is a schematic diagram of a matrix change process in a first inverse transform process according to an embodiment of the present invention;

FIG. 5 is a diagram illustrating a second frequency domain transform coefficient matrix according to an embodiment of the present invention;

FIG. 6 is a diagram illustrating a second intermediate value matrix according to an embodiment of the present invention;

fig. 7 is a schematic flowchart of a third inverse transformation method according to an embodiment of the present invention;

FIG. 8 is a diagram illustrating a third frequency domain transform coefficient matrix according to an embodiment of the present invention;

FIG. 9A is a diagram illustrating a third intermediate value matrix according to an embodiment of the present invention;

fig. 9B is a schematic diagram of a matrix change process in a second inverse transform process according to an embodiment of the present invention;

fig. 9C is a schematic flowchart of a fourth inverse transformation method according to an embodiment of the present invention;

fig. 10 is a schematic structural diagram of a first inverse transform device according to an embodiment of the present invention;

fig. 11 is a schematic structural diagram of a second inverse transform apparatus according to an embodiment of the present invention;

fig. 12 is a schematic structural diagram of a third inverse transform device according to an embodiment of the present invention;

fig. 13 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.

Detailed Description

The technical solutions in the embodiments of the present invention will be described below with reference to the drawings in the embodiments of the present invention.

Because the accuracy of the non-zero CG determined in the process of performing IDCT on the frequency domain transform coefficient matrix is low at present, to solve this problem, embodiments of the present invention provide an inverse transform method and apparatus.

In one embodiment of the present invention, there is provided an inverse transform method including:

and determining the maximum ordinate value and the maximum abscissa value of the non-zero coefficient block CG in the frequency domain transformation coefficient matrix of the image block in the video frame.

Determining a target inverse transformation mode according to a magnitude relation between the maximum ordinate value and the maximum abscissa value, wherein the target inverse transformation mode is as follows: and representing the inverse transformation mode of the execution sequence of the one-dimensional horizontal IDCT and the one-dimensional vertical IDCT in the IDCT process.

According to the target inverse transformation mode, performing IDCT on each first coefficient in a target inverse transformation unit with a target coordinate value not being 0 to obtain an inverse transformation result corresponding to each first coefficient, wherein the target inverse transformation unit is: the CG row or CG column which is characterized by the target inverse transformation method and which preferentially performs IDCT, wherein when the target inverse transformation unit is a CG row, the target coordinate value is: the maximum ordinate value of the non-zero CG in the target inverse transform unit is given by: in the target inverse transformation unit, the vertical coordinate value is equal to or less than a coefficient included in CG of the target coordinate value, and when the target inverse transformation unit is a CG row, the target coordinate value is: the maximum abscissa value of the non-zero CG in the target inverse transform unit, wherein the first coefficient is: the target inverse transformation means includes a coefficient included in CG having an abscissa value equal to or less than the target coordinate value.

Setting an inverse transformation result corresponding to a second coefficient to be 0, wherein the second coefficient is: and the coefficients except the first coefficient in the frequency domain transformation coefficient matrix.

As is apparent from the above, IDCT is performed on the basis of the coefficient included in the non-zero CG in the target inverse transform unit for which each target coordinate value is not 0, instead of IDCT performed on the coefficient included in the CG within the fixed region. The coefficient actually used for IDCT is changed according to the specific position of the non-zero CG in the coefficient matrix of different frequency domain transformation, so that only the coefficient contained in each non-zero CG is accurately subjected to IDCT, and the inverse transformation results corresponding to the coefficients contained in other all-zero CGs are directly set to be 0, thereby improving the accuracy of the non-zero CG determined in the IDCT process and further reducing the calculation amount in the IDCT process.

The inverse transformation method and apparatus provided by the embodiments of the present invention are described below with specific embodiments.

Referring to fig. 1, an embodiment of the present invention provides a flowchart of a first inverse transformation method, and specifically, the method includes the following steps S101 to S104.

S101: and determining the maximum ordinate value and the maximum abscissa value of the CG in the frequency domain transformation coefficient matrix of the image block in the video frame.

Specifically, after performing prediction, transformation, quantization, entropy coding, and inverse quantization on image blocks in a video frame in sequence, a frequency domain transform coefficient matrix of the image blocks in the video frame may be obtained, where the size of the frequency domain transform coefficient matrix is the same as the size of the image blocks in the video frame, for example, if the image blocks are 64 × 64 image blocks, the frequency domain transform coefficient matrix is 64 × 64 matrix, and if the image blocks are 32 × 32 image blocks, the frequency domain transform coefficient matrix is 32 × 32 matrix.

Before entropy encoding operation, the transform coefficients of the video frame obtained by quantization need to be scanned, and two-dimensional transform coefficients need to be converted into one-dimensional transform coefficients.

In one embodiment of the present invention, the size of the CG (Coefficient block) may be one of the following sizes: 1 × 1, 2 × 2, 4 × 4, 8 × 8, 16 × 16, 32 × 32, 64 × 64.

For example, when the frequency domain transform coefficient matrix is a 32 × 32 matrix and the CG size is 4 × 4, the frequency domain transform coefficient matrix includes 8 × 8 CGs, and when the CG size is 2 × 2, the frequency domain transform coefficient matrix includes 16 × 16 CGs.

Specifically, the maximum vertical coordinate value of the non-zero CG in each CG line may be determined, and then the maximum value of the maximum vertical coordinate values of each CG line may be determined as the maximum vertical coordinate value. For example, the maximum vertical coordinate value of the non-zero CG in each CG row may be 5, 3, 1, 0, respectively, wherein the maximum vertical coordinate value is 5.

The maximum abscissa value of the non-zero CG in each CG column may be determined, and then the maximum of the maximum abscissa values of each CG column may be determined as the above-mentioned maximum abscissa value. For example, the maximum abscissa value of the non-zero CG in each CG row may be 3, 2, 1, 0, respectively, wherein the maximum abscissa value is 3.

In addition, the maximum vertical coordinate value of each CG row may be stored in a vertical coordinate series having the same number of elements as the number of CG rows, wherein the maximum vertical coordinate value of the first CG row may be stored in a first element of the vertical coordinate series, the maximum vertical coordinate value of the second CG row may be stored in a second element of the vertical coordinate series, and so on, the maximum vertical coordinate value of each CG row may be stored in the vertical coordinate series.

The maximum abscissa value of each CG column may be stored in an abscissa array including the same number of elements as the CG columns, wherein the maximum abscissa value of the first CG column may be stored in a first element of the abscissa array, the maximum abscissa value of the second CG column may be stored in a second element of the abscissa array, and so on, and the maximum abscissa value of each CG column is stored in the abscissa array.

Referring to fig. 2, a schematic diagram of a first frequency domain transform coefficient matrix is provided.

Fig. 2 illustrates an example of the frequency domain transform coefficient matrix including 8 × 8 CGs, where a gray square is a non-zero CG and a white square is a full-zero CG. Here, each element in the matrix is represented by a square, and other shapes such as a rectangle may be used to represent each element in the matrix.

The maximum vertical coordinate values of the non-zero CGs in each CG row of the frequency domain transform coefficient matrix shown in fig. 2 are 5, 3, 1, 0, and 0, respectively, so that the maximum vertical coordinate value of the frequency domain transform coefficient matrix is 5, and the maximum horizontal coordinate values of the non-zero CGs in each CG column of the frequency domain transform coefficient matrix are 3, 2, 1, 0, and 0, respectively, so that the maximum horizontal coordinate value of the frequency domain transform coefficient matrix is 3.

S102: and determining a target inverse transformation mode according to the magnitude relation between the maximum ordinate value and the maximum abscissa value.

Wherein, the target inverse transformation mode is as follows: the Inverse transformation mode of the execution sequence of the one-dimensional horizontal IDCT and the one-dimensional vertical IDCT in the IDCT (Inverse Discrete Cosine Transform) characterization process.

Specifically, the inverse transformation results obtained finally are the same no matter whether one-dimensional horizontal IDCT is performed first and then one-dimensional vertical IDCT is performed, or one-dimensional vertical IDCT is performed first and then one-dimensional horizontal IDCT is performed. But the number of the non-zero CGs contained in the CG rows and the CG columns is different due to the different positions of the non-zero CGs in the different frequency domain transform coefficient matrices. Therefore, in the process of performing the IDCT according to the non-zero CG, according to the rule of matrix multiplication calculation in the IDCT process, the execution sequence of the one-dimensional horizontal IDCT and the one-dimensional horizontal IDCT in the inverse transformation process is different, and the calculation amount in the IDCT process is different.

Wherein, when the maximum ordinate value is greater than or equal to the maximum abscissa value, the target inverse transformation mode is as follows: in the characterization process, a one-dimensional horizontal IDCT is firstly carried out, and then an inverse transformation mode of a one-dimensional vertical IDCT is carried out.

In the case where the maximum ordinate value is equal to or greater than the maximum abscissa value, in the process of performing the one-dimensional horizontal IDCT, since the non-zero CGs are located in the CG rows having the abscissa value equal to or less than the maximum abscissa value in the frequency domain transform coefficient matrix, it is necessary to perform the IDCT on the non-zero CGs in the CG rows having the abscissa value equal to or less than the maximum abscissa value.

On the contrary, in the process of performing the one-dimensional vertical IDCT, since the non-zero CG is located in the CG column having the ordinate value equal to or less than the maximum ordinate value in the frequency domain transform coefficient matrix, the non-zero CG in the CG column having the abscissa value equal to or less than the maximum ordinate value needs to be subjected to the IDCT.

Since the maximum abscissa value is less than or equal to the maximum ordinate value, the calculation amount for performing the one-dimensional horizontal IDCT first and then performing the one-dimensional vertical IDCT is low.

When the maximum ordinate value is smaller than the maximum abscissa value, the target inverse transformation method is as follows: in the characterization process, a one-dimensional horizontal IDCT is firstly carried out, and then an inverse transformation mode of a one-dimensional vertical IDCT is carried out.

In the case where the maximum vertical coordinate value is smaller than the maximum horizontal coordinate value, in the process of performing the one-dimensional vertical IDCT, since the non-zero CG is located in the CG column having the vertical coordinate value equal to or smaller than the maximum vertical coordinate value in the frequency domain transform coefficient matrix, it is necessary to perform the IDCT on the non-zero CG in the CG column having the vertical coordinate value equal to or smaller than the maximum vertical coordinate value.

On the contrary, in the process of performing the one-dimensional horizontal IDCT, since the non-zero CGs are located in the CG rows having the abscissa value equal to or less than the maximum abscissa value in the frequency domain transform coefficient matrix, the non-zero CGs in the CG rows having the abscissa value equal to or less than the maximum abscissa value need to be subjected to the IDCT.

Since the maximum ordinate value is smaller than the maximum abscissa value, the calculation amount for performing the one-dimensional vertical IDCT first and then performing the one-dimensional horizontal IDCT is low.

The specific IDCT mode is described in detail in step S103 below, and will not be described in detail here.

S103: according to the target inverse transformation mode, each first coefficient in the target inverse transformation unit with the target coordinate value not being 0 is subjected to IDCT, and an inverse transformation result corresponding to each first coefficient is obtained.

Wherein, the target inverse transformation unit is: the CG rows or CG columns which are characterized by the target inverse transformation mode and are subjected to IDCT preferentially.

Specifically, since each element in the frequency domain transform coefficient matrix may represent a CG, each row in the frequency domain transform coefficient matrix may be referred to as a CG row, and each column in the frequency domain transform coefficient matrix may be referred to as a CG column. If one-dimensional horizontal IDCT is performed first, then each CG row is preferentially subjected to IDCT, the target inverse transformation unit is the CG row, and if one-dimensional vertical IDCT is performed first, then each CG row is preferentially subjected to IDCT, then the target inverse transformation unit is the CG row.

In the case where the target inverse transformation unit is a CG line, the target coordinate values are: the maximum ordinate value of the non-zero CG in the target inverse transform unit is given by: the vertical coordinate value of the target inverse transformation means is equal to or less than a coefficient included in CG of the target coordinate value.

When the target inverse transformation unit is a CG column, the target coordinate value is: the maximum abscissa value of the non-zero CG in the target inverse transform unit, wherein the first coefficient is: the target inverse transformation means includes a coefficient included in CG having an abscissa value equal to or less than the target coordinate value.

Specifically, taking the frequency domain transform coefficient matrix shown in fig. 2 as an example, the first coefficient included in the first CG row is: a coefficient included in the CG whose ordinate value is 5 or less, and a first coefficient included in the second CG line: a coefficient included in the CG whose ordinate value is 3 or less, and a first coefficient included in the third CG line: and a coefficient included in CG whose ordinate value is 1 or less.

In an embodiment of the present invention, the IDCT may be performed according to a target inverse transform mode according to each first coefficient in the frequency domain transform coefficient matrix, the preset transform matrix, and a transposed matrix of the preset transform matrix.

The number of CG rows contained in the preset transformation matrix is the same as that of the frequency domain transformation coefficient matrix, and the number of CG columns contained in the preset transformation matrix is the same as that of the frequency domain transformation coefficient matrix.

Specifically, the step S103 can be realized by steps S103A-S103B or S103C-S103D respectively for different magnitude relations between the maximum abscissa value and the maximum ordinate value, which will not be detailed herein for the moment.

S104: and setting the inverse transformation result corresponding to the second coefficient to be 0.

Wherein the second coefficient is: and the coefficients except the first coefficient in the frequency domain transformation coefficient matrix.

Specifically, since the second coefficient is a coefficient other than the first coefficient in the frequency domain transform coefficient matrix, and the first coefficient is a coefficient included in each non-zero CG in the frequency domain transform coefficient matrix, the second coefficient is a coefficient included in each all-zero CG, and the second coefficient is 0. Since the inverse transform result obtained after IDCT is performed on the second coefficient is 0, the inverse transform result corresponding to the second coefficient is set to 0 without IDCT being performed on the second coefficient. Thereby reducing the amount of calculation of the IDCT process.

So far the IDCT process is finished.

In the case that the maximum ordinate value is equal to or greater than the maximum abscissa value, referring to fig. 3, a flow chart of a second inverse transformation method is provided. In the above case, the target inverse transform method is: in the characterization process, a one-dimensional horizontal IDCT is firstly carried out, and then an inverse transformation mode of a one-dimensional vertical IDCT is carried out. Compared to the previous embodiment shown in FIG. 1, the above step S103 can be realized by the following steps S103A-S103B.

S103A: and performing one-dimensional horizontal IDCT on the first coefficient in the CG row of which the longitudinal coordinate value of each target is not 0 to obtain a middle value matrix.

Wherein, the target ordinate value is: the maximum vertical coordinate value of a non-zero CG in a CG line.

The median matrix is: the number of CG rows included is the same as the frequency domain transform coefficient matrix, the number of CG columns included is the same as the frequency domain transform coefficient matrix, and the initial element value of each element is a matrix of 0, and the element value of the element corresponding to each first coefficient in the intermediate value matrix is: and the one-dimensional horizontal inverse transformation result corresponding to each first coefficient.

In an embodiment of the present invention, the frequency domain transform coefficient matrix may be multiplied by a predetermined transform matrix, and the one-dimensional horizontal IDCT may be performed on the first coefficient.

Specifically, for each CG row with a target ordinate value different from 0, matrix multiplication is performed on the first coefficient in the CG row and the element included in the CG row with an abscissa value less than or equal to the target ordinate value in each CG column of the preset transformation matrix, respectively, so as to obtain a first-dimensional inverse transformation result of each first coefficient in the CG row.

Since the maximum ordinate value of the frequency domain transform coefficient matrix shown in fig. 2 is greater than the maximum abscissa value, taking fig. 2 as an example, matrix multiplication calculation may be performed on the first coefficient included in the CG having the ordinate value of 5 or less in the first CG row and the element included in the CG having the abscissa value of 5 or less in each CG column of the preset transform matrix.

And performing matrix multiplication calculation on a first coefficient contained in the CG with the ordinate value less than or equal to 3 in the second CG row and an element contained in the CG with the abscissa value less than or equal to 3 in each CG column of the preset transformation matrix.

And performing matrix multiplication calculation on a first coefficient contained in the CG of which the ordinate value is less than or equal to 1 in the third CG row and an element contained in the CG of which the abscissa value is less than or equal to 1 in each CG column of the preset transformation matrix.

Because the first coefficient is positioned in the CG row of which the abscissa value is less than or equal to the maximum abscissa value in the frequency domain transformation coefficient matrix, according to the rule of a matrix multiplication calculation mode, the one-dimensional horizontal IDCT is carried out according to the first coefficient, and each CG in the CG row of which the abscissa value is less than or equal to the maximum abscissa value in the obtained intermediate value matrix is non-zero CG.

Referring to fig. 4A, a schematic diagram of a first intermediate value matrix is provided.

Specifically, fig. 4A is a middle value matrix obtained by performing one-dimensional horizontal IDCT on the first coefficient in the frequency domain transform coefficient matrix shown in fig. 2, where a gray square is non-zero CG and a white square is all-zero CG. Here, each element in the matrix is represented by a square, and other shapes such as a rectangle may be used to represent each element in the matrix.

As can be seen from fig. 4A, in the intermediate value matrix shown in the figure, the CGs included in the CG row whose abscissa value is equal to or less than 3 are non-zero CGs, and the CGs included in the other CG rows are all-zero CGs.

Referring to fig. 4B, an embodiment of the present invention provides a schematic diagram of a matrix change process in a first inverse transform process.

Where each square represents an element in the matrix, the grey squares are non-zero CGs and the white squares are all-zero CGs. Here, each element in the matrix is represented by a square, and other shapes such as a rectangle may be used to represent each element in the matrix.

Each set of squares represents a matrix, where the matrix U is a frequency domain transform coefficient matrix, the matrix U is the same as the frequency domain transform coefficient matrix shown in fig. 2, the matrix V is a predetermined transform matrix, and the matrix V is a predetermined transform matrix^TIs a transpose of the predetermined transform matrix.

Specifically, the IDCT process is represented by the formula P ═ V^TU is a frequency domain transform coefficient matrix, V is a preset transform matrix, and V is^TIs a transposed matrix of the frequency domain transform matrix, and P is a result obtained after inverse transform of the frequency domain transform coefficient matrix.

Since the maximum ordinate value of the frequency domain transform coefficient matrix U is 5 and the maximum abscissa value is 3, the maximum ordinate value of the frequency domain transform coefficient matrix U is greater than the maximum abscissa value, and therefore, the one-dimensional horizontal IDCT is performed on the frequency domain transform coefficient matrix U first. That is, U × V is calculated to obtain an intermediate value matrix W, which is the same as the intermediate value matrix shown in fig. 4A. Recalculating V^TW, obtaining an inverse transformation result of the frequency domain transformation coefficient matrix U.

Referring to fig. 5, a schematic diagram of a second frequency domain transform coefficient matrix is provided.

Fig. 5 illustrates an example of the frequency domain transform coefficient matrix including 8 × 8 CGs, where a gray square is a non-zero CG and a white square is a full-zero CG. Here, each element in the matrix is represented by a square, and other shapes such as a rectangle may be used to represent each element in the matrix.

The maximum vertical coordinate values of the non-zero CG in each CG row of the frequency domain transform coefficient matrix shown in fig. 5 are 4, 2, 1, 0, and 0, respectively, so the maximum vertical coordinate value of the frequency domain transform coefficient matrix is 4, and the maximum horizontal coordinate values of the non-zero CG in each CG column of the frequency domain transform coefficient matrix are 4, 3, 2, 0, and 0, respectively, so the maximum horizontal coordinate value of the frequency domain transform coefficient matrix is 4, and the maximum vertical coordinate value is the same as the maximum horizontal coordinate value.

Referring to fig. 6, a schematic diagram of a second intermediate value matrix is provided.

Specifically, fig. 6 is a middle value matrix obtained by performing one-dimensional horizontal IDCT on the first coefficient in the frequency domain transform coefficient matrix shown in fig. 5, where a gray square is non-zero CG and a white square is zero CG. Here, each element in the matrix is represented by a square, and other shapes such as a rectangle may be used to represent each element in the matrix.

As can be seen from fig. 6, in the intermediate value matrix shown in the figure, the CGs included in the CG row whose abscissa value is equal to or less than 4 are non-zero CGs, and the CGs included in the other CG rows are all-zero CGs.

S103B: and performing one-dimensional vertical IDCT on coefficients contained in the CG with the abscissa value less than or equal to the maximum abscissa value in each CG row in the intermediate value matrix to obtain an inverse transformation result corresponding to each first coefficient.

Because only the CG in the CG row with the abscissa value less than or equal to the maximum abscissa value in the intermediate value matrix is non-zero CG, in the process of one-dimensional vertical IDCT, only the coefficients contained in the CG with the abscissa value less than or equal to the maximum abscissa value in each CG column in the intermediate value matrix are subjected to one-dimensional vertical IDCT, and the coefficients contained in other CGs are not subjected to one-dimensional vertical IDCT, so that the calculation amount in the IDCT process is reduced.

Specifically, the transposed matrix of the preset transformation matrix is multiplied by the intermediate value matrix, and the intermediate value matrix is subjected to one-dimensional vertical IDCT.

And for each CG row of the intermediate value matrix, matrix multiplication is respectively carried out on a coefficient contained in the CG of which the abscissa value is less than or equal to the maximum abscissa value in the CG row and an element contained in the CG of which the ordinate value is less than or equal to the maximum abscissa value in each CG row in the transposed matrix of the preset transformation matrix, so that an inverse transformation result of the coefficient contained in the CG of which the abscissa value is less than or equal to the maximum abscissa value in the CG row is obtained.

Since the coefficient included in the CG having the abscissa value equal to or less than the maximum abscissa value in the CG row is calculated from the first coefficient, the result of inverse transformation of the coefficient included in the CG having the abscissa value equal to or less than the maximum abscissa value in the CG row is the one-dimensional vertical inverse transformation result corresponding to the first coefficient.

Taking the median matrix shown in fig. 4A as an example, matrix multiplication calculation is performed on a coefficient included in a CG having an abscissa value equal to or less than 3 in the first CG row and an element included in a CG having an ordinate value equal to or less than 3 in each CG row of the transposed matrix of the preset transformation matrix, and matrix multiplication calculation is performed on a coefficient included in a CG having an abscissa value equal to or less than 3 in the second CG row and a coefficient included in a CG having an ordinate value equal to or less than 3 in each CG row of the transposed matrix of the preset transformation matrix, and so on.

In this embodiment, when the maximum ordinate value is greater than or equal to the maximum abscissa value, the one-dimensional horizontal IDCT is performed first and then the one-dimensional vertical IDCT is performed in the IDCT process, so that only the CGs included in the CG rows having the abscissa value less than or equal to the maximum abscissa value in the intermediate value matrix obtained after the one-dimensional horizontal IDCT is performed are non-zero CGs, and therefore, only the coefficients included in the CGs having the abscissa value less than or equal to the maximum abscissa value in each column in the intermediate value matrix are subjected to the second column-wise IDCT.

On the contrary, if the one-dimensional vertical IDCT is performed first and then the one-dimensional horizontal IDCT is performed during the IDCT, only the CGs included in the CG columns having the vertical coordinate value less than or equal to the maximum vertical coordinate value in the intermediate value matrix obtained after the one-dimensional vertical IDCT is non-zero CGs, and thus only the coefficients included in the CGs having the vertical coordinate value less than or equal to the maximum vertical coordinate value in each row in the intermediate value matrix are subjected to the one-dimensional horizontal IDCT.

However, since the maximum ordinate value is greater than or equal to the maximum abscissa value, compared with performing the one-dimensional vertical IDCT first and then performing the one-dimensional horizontal IDCT, the number of non-zero CGs in the intermediate value matrix is smaller, so that the number of coefficients to be calculated during performing the one-dimensional vertical IDCT is smaller, and therefore, when the maximum ordinate value is greater than or equal to the maximum abscissa value, performing the one-dimensional horizontal IDCT first and then performing the one-dimensional vertical IDCT calculation can reduce the calculation amount during the IDCT.

In addition, when the maximum abscissa value is equal to the maximum ordinate value, the number of coefficients to be calculated in the IDCT process is the same regardless of whether the one-dimensional horizontal IDCT is performed first and then the one-dimensional vertical IDCT is performed or the one-dimensional horizontal IDCT is performed first and then the one-dimensional vertical IDCT is performed. Therefore, when the maximum abscissa value is equal to the maximum ordinate value, the target inverse transformation method may be: in the characterization process, a one-dimensional vertical IDCT is firstly carried out, and then a one-dimensional horizontal IDCT is carried out.

As can be seen from the above, in both the one-dimensional horizontal IDCT and the one-dimensional vertical IDCT, only the coefficients included in the non-zero CG are calculated, and therefore the amount of calculation in the IDCT is low.

In the case where the maximum ordinate value is less than the maximum abscissa value, referring to fig. 7, a flow diagram of a third inverse transformation method is provided. In the above case, the target inverse transform method is: in the characterization process, a one-dimensional vertical IDCT is firstly carried out, and then a one-dimensional horizontal IDCT is carried out. Compared to the previous embodiment shown in FIG. 1, the above step S103 can be realized by the following steps S103C-S103D.

S103C: and performing one-dimensional vertical IDCT on the first coefficient in the CG row of which the target abscissa value is not 0 to obtain an intermediate value matrix.

Wherein, the target abscissa value is: the maximum abscissa value of a non-zero CG within the CG column.

The median matrix is: the number of CG rows included is the same as the frequency domain transform coefficient matrix, the number of CG columns included is the same as the frequency domain transform coefficient matrix, and the initial element value of each element is a matrix of 0, and the element value of the element corresponding to each first coefficient in the intermediate value matrix is: the one-dimensional vertical inverse transform result of each first coefficient.

In an embodiment of the invention, the transpose matrix of the preset transform matrix may be multiplied by the frequency domain transform coefficient matrix, so as to perform one-dimensional vertical IDCT on the first coefficient.

Specifically, for each CG row with a target coordinate value different from 0, matrix multiplication is performed on elements included in CGs with a vertical coordinate value less than or equal to a target horizontal coordinate value in each CG row of the transpose matrix of the preset transformation matrix and the first coefficients in the CG row, so as to obtain one-dimensional inverse vertical transformation results of the first coefficients in the CG row.

Referring to fig. 8, a schematic diagram of a third frequency-domain transform coefficient matrix is provided.

Fig. 8 illustrates an example of the frequency domain transform coefficient matrix including 8 × 8 CGs, where a gray square is a non-zero CG and a white square is a full-zero CG. Here, each element in the matrix is represented by a square, and other shapes such as a rectangle may be used to represent each element in the matrix.

The maximum vertical coordinate values of the non-zero CG in each row of the frequency domain transform coefficient matrix shown in fig. 8 are 3, 2, 1, 0, and 0, respectively, so that the maximum vertical coordinate value of the frequency domain transform coefficient matrix is 3, the maximum horizontal coordinate values of the non-zero CG in each column are 4, 3, 1, 0, and 0, respectively, and the maximum horizontal coordinate value of the frequency domain transform coefficient matrix is 4, that is, the maximum horizontal coordinate value is greater than the maximum vertical coordinate value.

Taking fig. 8 as an example, the matrix multiplication calculation may be performed on an element included in a CG whose vertical coordinate value is equal to or less than 4 in each CG row of the transpose matrix of the preset transformation matrix and a first coefficient included in a CG whose horizontal coordinate value is equal to or less than 4 in the first CG column of the frequency domain transformation coefficient matrix.

And carrying out matrix multiplication calculation on elements contained in the CG of which the vertical coordinate value is less than or equal to 3 in each CG row of the transposed matrix of the preset transformation matrix and a first coefficient contained in the CG of which the horizontal coordinate value is less than or equal to 3 in a second CG row of the frequency domain transformation coefficient matrix.

And carrying out matrix multiplication calculation on elements contained in the CG of which the vertical coordinate value is less than or equal to 1 in each CG row of the transposed matrix of the preset transformation matrix and a first coefficient contained in the CG of which the horizontal coordinate value is less than or equal to 1 in a third CG row of the frequency domain transformation coefficient matrix.

Because the first coefficient is positioned in the CG row of which the vertical coordinate value is less than or equal to the maximum vertical coordinate value in the frequency domain transformation coefficient matrix, according to the rule of a matrix multiplication calculation mode, one-dimensional vertical IDCT is carried out according to the first coefficient, and each CG in the CG row of which the vertical coordinate value is less than or equal to the maximum vertical coordinate value in the obtained intermediate value matrix is non-zero CG.

Referring to fig. 9A, a schematic diagram of a third intermediate value matrix is provided.

Specifically, fig. 9A is a middle value matrix obtained by performing one-dimensional vertical IDCT on the first coefficient in the frequency domain transform coefficient matrix shown in fig. 8, where a gray square is non-zero CG and a white square is all-zero CG. Here, each element in the matrix is represented by a square, and other shapes such as a rectangle may be used to represent each element in the matrix.

As can be seen from fig. 9A, in the intermediate value matrix shown in the figure, the CGs included in the CG columns whose ordinate values are equal to or less than 3 are non-zero CGs, and the CGs included in the other CG columns are all-zero CGs.

S103D: and performing one-dimensional horizontal IDCT on coefficients contained in the CG, of which the vertical coordinate value in each CG row is less than or equal to the maximum vertical coordinate value, in the intermediate value matrix to obtain an inverse transformation result corresponding to each first coefficient.

Because only the CG in the CG row with the vertical coordinate value less than or equal to the maximum vertical coordinate value in the intermediate value matrix is non-zero CG, in the process of one-dimensional horizontal IDCT, only the coefficients contained in the CG with the vertical coordinate value less than or equal to the maximum vertical coordinate value in each CG row in the intermediate value matrix are subjected to one-dimensional horizontal IDCT, and the coefficients contained in other CGs are not subjected to one-dimensional horizontal IDCT, so that the calculation amount in the IDCT process is reduced.

Specifically, the median matrix is multiplied by the preset transformation matrix, and a one-dimensional horizontal IDCT is performed on the median matrix.

And for each CG row of the intermediate value matrix, performing matrix multiplication calculation on a coefficient contained in the CG of which the vertical coordinate value is less than or equal to the maximum vertical coordinate value in the CG row and elements contained in the CG of which the horizontal coordinate value is less than or equal to the maximum vertical coordinate value in each CG column in the preset transformation matrix respectively to obtain an inverse transformation result of the coefficient contained in the CG of which the vertical coordinate value is less than or equal to the maximum vertical coordinate value in the CG row.

Since the coefficient included in the CG having the vertical coordinate value equal to or less than the maximum vertical coordinate value in the CG line is calculated from the first coefficient, the result of inverse transformation of the coefficient included in the CG having the vertical coordinate value equal to or less than the maximum vertical coordinate value in the CG line is the result of inverse transformation corresponding to the first coefficient.

Taking the median matrix shown in fig. 9A as an example, matrix multiplication calculation is performed on the coefficients included in the CGs whose vertical coordinate values are less than or equal to 3 in the first CG row and the elements included in the CGs whose horizontal coordinate values are less than or equal to 3 in each CG column of the preset transformation matrix, and matrix multiplication calculation is performed on the coefficients included in the CGs whose vertical coordinate values are less than or equal to 3 in the second CG row and the coefficients included in the CGs whose horizontal coordinate values are less than or equal to 3 in each CG column of the preset transformation matrix, and so on.

In this embodiment, when the maximum ordinate value is smaller than the maximum abscissa value, the priority of the one-dimensional vertical IDCT is performed first and then the priority of the one-dimensional horizontal IDCT is performed during the IDCT, so that only the CGs included in the CG columns having the ordinate values smaller than or equal to the maximum ordinate value in the intermediate value matrix obtained after the one-dimensional vertical IDCT is non-zero CGs, and thus only the coefficients included in the CGs having the ordinate values smaller than or equal to the maximum ordinate value in each row in the intermediate value matrix are subjected to the one-dimensional horizontal IDCT.

On the contrary, if the priority of the one-dimensional horizontal IDCT is performed first and then the priority of the one-dimensional vertical IDCT is performed during the IDCT, only the CGs included in the CG rows having the abscissa value less than or equal to the maximum abscissa value among the intermediate value matrix obtained after the one-dimensional horizontal IDCT is performed are non-zero CGs, and thus only the coefficients included in the CGs having the abscissa value less than or equal to the maximum abscissa value in each column in the intermediate value matrix are subjected to the one-dimensional vertical IDCT.

However, since the maximum ordinate value is smaller than the maximum abscissa value, compared with performing the one-dimensional horizontal IDCT first and then performing the one-dimensional vertical IDCT, the number of non-zero CGs in the intermediate value matrix is smaller, so that the number of coefficients to be calculated during the IDCT is smaller, and therefore, when the maximum ordinate value is smaller than the maximum abscissa value, the calculation amount during the IDCT can be reduced by performing the one-dimensional vertical IDCT first and then performing the one-dimensional horizontal IDCT.

As can be seen from the above, only the coefficients included in the non-zero CG are calculated in both the first-dimension IDCT and the second-dimension IDCT, and therefore the amount of calculation in the IDCT is made low.

Referring to fig. 9B, an embodiment of the present invention provides a schematic diagram of a matrix change process in a second inverse transform process.

Each set of squares represents a matrix, where matrix X is a frequency domain transform coefficient matrix, matrix X is the same as the frequency domain transform coefficient matrix shown in fig. 8, matrix Y is a predetermined transform matrix, and matrix Y is a predetermined transform matrix^TIs a transpose of the predetermined transform matrix.

Specifically, the process of IDCT is represented by the formula P ═ Y^TX and Y are realized, wherein X is a frequency domain transformation coefficient matrix, Y is a preset transformation matrix, and Y is^TAs a transpose of the frequency domain transform matrix, P is the frequency alignmentAnd performing inverse transformation on the domain transformation coefficient matrix to obtain a result.

Since the maximum ordinate value of the frequency domain transform coefficient matrix X is 3 and the maximum abscissa value is 4, the maximum abscissa value of the frequency domain transform coefficient matrix X is greater than the maximum ordinate value, and thus the one-dimensional vertical IDC and then the one-dimensional horizontal IDCT are performed on the frequency domain transform coefficient matrix X. I.e. calculate Y first^TX yields a median matrix Z, which is the same as the median matrix shown in fig. 9A. And then calculating Z X Y to obtain an inverse transformation result of the frequency domain transformation coefficient matrix X.

Referring to fig. 9C, a flowchart of a fourth inverse transformation method is provided in the embodiment of the present invention.

S901: and determining the maximum ordinate value of the non-zero CG contained in each CG row in the frequency domain transformation coefficient matrix as each target ordinate value, and determining the maximum value of each target ordinate value as the maximum ordinate value of the frequency domain transformation coefficient matrix.

S902: and determining the maximum abscissa value of the non-zero CG contained in each CG row in the frequency domain transformation coefficient matrix as each target abscissa value, and determining the maximum value of each target abscissa value as the maximum abscissa value of the frequency domain transformation coefficient matrix.

S903: and under the condition that the maximum longitudinal coordinate value is larger than or equal to the maximum horizontal coordinate value, performing one-dimensional horizontal IDCT on the non-zero CG in each CG row according to the target longitudinal coordinate value of each CG row to obtain a middle value matrix, and performing one-dimensional vertical IDCT on the CG columns of which the longitudinal coordinate values are smaller than or equal to the maximum longitudinal coordinate value in the middle value matrix.

S904: and under the condition that the maximum longitudinal coordinate value is smaller than the maximum horizontal coordinate value, performing one-dimensional vertical IDCT on the non-zero CG in each CG row according to the target horizontal coordinate value of each CG row to obtain a middle value matrix, and performing one-dimensional horizontal IDCT on the CG rows of which the horizontal coordinate value is smaller than or equal to the maximum horizontal coordinate value in the middle value matrix.

Corresponding to the foregoing inverse transform method, an embodiment of the present invention further provides an inverse transform apparatus.

Referring to fig. 10, an embodiment of the present invention provides a schematic structural diagram of a first inverse transform apparatus, where the apparatus includes:

a coordinate value determining module 1001, configured to determine a maximum ordinate value and a maximum abscissa value of a non-zero coefficient block CG in a frequency domain transform coefficient matrix of an image block in a video frame;

a manner determining module 1002, configured to determine a target inverse transformation manner according to a magnitude relationship between the maximum ordinate value and the maximum abscissa value, where the target inverse transformation manner is: representing an inverse transformation mode of a one-dimensional horizontal IDCT and a one-dimensional vertical IDCT execution sequence in an IDCT process;

a result obtaining module 1003, configured to perform IDCT on each first coefficient in a target inverse transformation unit with a target coordinate value different from 0 according to the target inverse transformation manner, to obtain an inverse transformation result corresponding to each first coefficient, where the target inverse transformation unit is: the target inverse transformation mode represents a CG row or a CG column which is subjected to IDCT preferentially, and when a target inverse transformation unit is a CG row, the target coordinate value is as follows: the maximum ordinate value of the non-zero CG in the target inverse transformation unit, the first coefficient being: the longitudinal coordinate value in the target inverse transformation unit is less than or equal to a coefficient included in the CG of the target coordinate value, and when the target inverse transformation unit is a CG column, the target coordinate value is: the maximum abscissa value of the non-zero CG in the target inverse transformation unit, the first coefficient is: a coefficient included in CG, whose abscissa value is equal to or less than the target coordinate value, in the target inverse transformation unit;

a result setting module 1004, configured to set an inverse transform result corresponding to a second coefficient to 0, where the second coefficient is: and transforming the coefficients except the first coefficient in the coefficient matrix of the frequency domain.

In one embodiment of the present invention, when the maximum ordinate value is equal to or greater than the maximum abscissa value, the target inverse transformation method is: in the process of representing the IDCT, an inverse transform mode of performing a one-dimensional horizontal IDCT first and then performing a one-dimensional vertical IDCT is performed, and referring to fig. 11, a schematic structural diagram of a second inverse transform device is provided.

Compared with the foregoing embodiment shown in fig. 10, the result obtaining module 1003 includes:

a first matrix obtaining submodule 1003A, configured to perform one-dimensional horizontal IDCT on a first coefficient in a CG row where each target ordinate value is not 0, to obtain an intermediate value matrix, where the target ordinate value is: the maximum longitudinal coordinate value of the non-zero CG in the CG line, and the intermediate value matrix is as follows: the number of included CG rows is the same as the frequency domain transform coefficient matrix, the number of included CG columns is the same as the frequency domain transform coefficient matrix, the initial element value of each element is a 0 matrix, and the element value of the element corresponding to each first coefficient in the intermediate value matrix is: one-dimensional horizontal inverse transformation results of the respective first coefficients;

a first result obtaining sub-module 1003B, configured to perform one-dimensional vertical IDCT on coefficients included in the CGs with abscissa values less than or equal to the maximum abscissa value in each CG column in the intermediate value matrix, so as to obtain inverse transform results corresponding to each first coefficient.

In an embodiment of the present invention, when the maximum ordinate value is smaller than the maximum abscissa value, the target inverse transformation method is: in the characterization process, an inverse transform mode of performing one-dimensional vertical IDCT first and then performing one-dimensional horizontal IDCT is performed, and referring to fig. 12, a schematic structural diagram of a third inverse transform apparatus is provided.

a second matrix obtaining submodule 1003C, configured to perform one-dimensional vertical IDCT on a first coefficient in the CG column where each target abscissa value is not 0, to obtain an intermediate value matrix, where the target abscissa value is: the maximum abscissa value of non-zero CG in the CG row, and the intermediate value matrix is as follows: the number of included CG rows is the same as the frequency domain transform coefficient matrix, the number of included CG columns is the same as the frequency domain transform coefficient matrix, the initial element value of each element is a 0 matrix, and the element value of the element corresponding to each first coefficient in the intermediate value matrix is: a one-dimensional vertical inverse transform result of each first coefficient;

a second result obtaining submodule 1003D is configured to perform one-dimensional horizontal IDCT on coefficients included in the CG where the ordinate in each CG row is less than or equal to the maximum ordinate in the intermediate value matrix, so as to obtain an inverse transform result corresponding to each first coefficient.

In an embodiment of the present invention, the CG is one of the following sizes:

1×1、2×2、4×4、8×8、16×16、32×32、64×64。

an embodiment of the present invention further provides an electronic device, as shown in fig. 13, including a processor 1301, a communication interface 1302, a memory 1303, and a communication bus 1304, where the processor 1301, the communication interface 1302, and the memory 1303 complete mutual communication through the communication bus 1304,

a memory 1303 for storing a computer program;

the processor 1301 is configured to implement the method steps according to any one of the embodiments of the inverse transform method when executing the program stored in the memory 1303.

When the electronic device provided by the embodiment of the invention is applied to inverse transformation, the IDCT is carried out according to the coefficient contained in the non-zero CG in the target inverse transformation unit with each target coordinate value not being 0, but not according to the coefficient contained in the CG in the fixed area. The coefficient actually used for IDCT is changed according to the specific position of the non-zero CG in the coefficient matrix of different frequency domain transformation, so that only the coefficient contained in each non-zero CG is accurately subjected to IDCT, and the inverse transformation results corresponding to the coefficients contained in other all-zero CGs are directly set to be 0, thereby improving the accuracy of the non-zero CG determined in the IDCT process and further reducing the calculation amount in the IDCT process.

The communication bus mentioned in the above terminal may be a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, or the like. The communication bus may be divided into an address bus, a data bus, a control bus, etc. For ease of illustration, only one thick line is shown, but this does not mean that there is only one bus or one type of bus.

The communication interface is used for communication between the terminal and other equipment.

The Memory may include a Random Access Memory (RAM) or a non-volatile Memory (non-volatile Memory), such as at least one disk Memory. Optionally, the memory may also be at least one memory device located remotely from the processor.

The Processor may be a general-purpose Processor, and includes a Central Processing Unit (CPU), a Network Processor (NP), and the like; the Integrated Circuit may also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA) or other Programmable logic device, a discrete Gate or transistor logic device, or a discrete hardware component.

In yet another embodiment of the present invention, a computer-readable storage medium is further provided, in which a computer program is stored, and the computer program, when executed by a processor, implements the inverse transform method in any of the above embodiments.

When inverse transformation is performed by applying the computer program stored in the computer-readable storage medium provided in the present embodiment, IDCT is performed based on coefficients included in non-zero CGs in the target inverse transformation unit for which respective target coordinate values are not 0, instead of IDCT performed on coefficients included in CGs within a fixed area. The coefficient actually used for IDCT is changed according to the specific position of the non-zero CG in the coefficient matrix of different frequency domain transformation, so that only the coefficient contained in each non-zero CG is accurately subjected to IDCT, and the inverse transformation results corresponding to the coefficients contained in other all-zero CGs are directly set to be 0, thereby improving the accuracy of the non-zero CG determined in the IDCT process and further reducing the calculation amount in the IDCT process.

In a further embodiment provided by the present invention, there is also provided a computer program product comprising instructions which, when run on a computer, cause the computer to perform the inverse transform method as described in any of the above embodiments.

When the inverse transform is performed by the computer program product provided in the present embodiment, IDCT is performed based on the coefficient included in the non-zero CG in the target inverse transform unit for which each target coordinate value is not 0, instead of performing IDCT on the coefficient included in the CG within the fixed region. The coefficient actually used for IDCT is changed according to the specific position of the non-zero CG in the coefficient matrix of different frequency domain transformation, so that only the coefficient contained in each non-zero CG is accurately subjected to IDCT, and the inverse transformation results corresponding to the coefficients contained in other all-zero CGs are directly set to be 0, thereby improving the accuracy of the non-zero CG determined in the IDCT process and further reducing the calculation amount in the IDCT process.

In the above embodiments, the implementation may be wholly or partially realized by software, hardware, firmware, or any combination thereof. When implemented in software, may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When loaded and executed on a computer, cause the processes or functions described in accordance with the embodiments of the invention to occur, in whole or in part. The computer may be a general purpose computer, a special purpose computer, a network of computers, or other programmable device. The computer instructions may be stored in a computer readable storage medium or transmitted from one computer readable storage medium to another, for example, from one website site, computer, server, or data center to another website site, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, Digital Subscriber Line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device, such as a server, a data center, etc., that incorporates one or more of the available media. The usable medium may be a magnetic medium (e.g., floppy Disk, hard Disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., Solid State Disk (SSD)), among others.

It is noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.

All the embodiments in the present specification are described in a related manner, and the same and similar parts among the embodiments may be referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, for the apparatus, the electronic device, the computer-readable storage medium and the computer program product, since they are substantially similar to the method embodiments, the description is relatively simple, and in relation to what is described in the partial description of the method embodiments.

The above description is only for the preferred embodiment of the present invention, and is not intended to limit the scope of the present invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention shall fall within the protection scope of the present invention.

Claims

1. An inverse transform method, the method comprising:

2. The method according to claim 1, wherein when the maximum ordinate value is equal to or greater than the maximum abscissa value, the target inverse transformation method is: in the characterization process, a one-dimensional horizontal IDCT is firstly carried out, and then an inverse transformation mode of a one-dimensional vertical IDCT is carried out;

3. The method according to claim 1, wherein in the case where the maximum ordinate value is smaller than the maximum abscissa value, the target inverse transformation manner is: in the characterization process, a one-dimensional vertical IDCT is firstly carried out, and then a one-dimensional horizontal IDCT inverse transformation mode is carried out;

4. The method of any of claims 1-3, wherein the size of the CG is one of the following sizes:

1×1、2×2、4×4、8×8、16×16、32×32、64×64。

5. an inverse transform apparatus, the apparatus comprising:

6. The apparatus according to claim 5, wherein when the maximum ordinate value is equal to or greater than the maximum abscissa value, the target inverse transformation method is: in the characterization process, a one-dimensional horizontal IDCT is firstly carried out, and then an inverse transformation mode of a one-dimensional vertical IDCT is carried out;

the result obtaining module comprises:

and the first result obtaining submodule is used for carrying out one-dimensional vertical IDCT on coefficients contained in the CG of which the abscissa value in each CG row is less than or equal to the maximum abscissa value in the intermediate value matrix to obtain an inverse transformation result corresponding to each first coefficient.

7. The apparatus according to claim 5, wherein, in the case where the maximum ordinate value is smaller than the maximum abscissa value, the target inverse transformation manner is: in the characterization process, a one-dimensional vertical IDCT is firstly carried out, and then a one-dimensional horizontal IDCT inverse transformation mode is carried out;

the result obtaining module comprises:

8. The apparatus of any of claims 5-7, wherein the size of the CG is one of the following:

1×1、2×2、4×4、8×8、16×16、32×32、64×64。

9. an electronic device is characterized by comprising a processor, a communication interface, a memory and a communication bus, wherein the processor and the communication interface are used for realizing mutual communication by the memory through the communication bus;

a memory for storing a computer program;

a processor for implementing the method steps of any of claims 1 to 4 when executing a program stored in the memory.

10. A computer-readable storage medium, characterized in that a computer program is stored in the computer-readable storage medium, which computer program, when being executed by a processor, carries out the method steps of any one of claims 1 to 4.