WO2020000487A1

WO2020000487A1 - Transformation method, inverse transformation method and apparatus

Info

Publication number: WO2020000487A1
Application number: PCT/CN2018/093923
Authority: WO
Inventors: 朱策; 曹小强; 林永兵; 郑建铧
Original assignee: 华为技术有限公司
Priority date: 2018-06-30
Filing date: 2018-06-30
Publication date: 2020-01-02

Abstract

Provided in the present application are an inverse transformation method and apparatus. The method comprises: parsing a code stream to acquire a residual coefficient of a current block, an intra-frame prediction mode, and a target transformation kernel pair determination parameter, wherein the target transformation kernel pair determination parameter comprises a first flag bit, or the target transformation kernel pair determination parameter comprises a first flag bit and a transformation index value; determining a target transformation kernel pair corresponding to the target transformation kernel pair determination parameter; and according to the target transformation kernel pair, carrying out inverse transformation on the residual coefficient to obtain a residual block. According to the present application, the target transformation kernel pair can be determined according to the target transformation kernel pair determination parameter, so that inverse transformation of the residual coefficient is realized.

Description

Transformation method, inverse transformation method and device

Technical field

The present application relates to the technical field of video encoding and decoding, and more particularly, to a transform method, an inverse transform method, and a device.

Background technique

Digital video capabilities can be incorporated into a wide variety of devices, including digital television, digital live broadcast systems, wireless broadcast systems, personal digital assistants (PDAs), laptop or desktop computers, tablet computers, electronics Book readers, digital cameras, digital recording devices, digital media players, video game devices, video game consoles, cellular or satellite radio phones (so-called "smart phones"), video teleconferencing devices, video streaming devices And similar. Digital video devices implement video compression technology, for example, in standards defined by MPEG-2, MPEG-4, ITU-T H.263, ITU-T H.264 / MPEG-4 Part 10 Advanced Video Coding (AVC), Video coding standard H.265 / High Efficiency Video Coding (HEVC) standard and the video compression technology described in the extension of such standards. Video devices can implement such video compression techniques to more efficiently transmit, receive, encode, decode, and / or store digital video information.

Video compression techniques perform spatial (intra-image) prediction and / or temporal (inter-image) prediction to reduce or remove redundancy inherent in video sequences. For block-based video coding, a video slice (ie, a video frame or a portion of a video frame) can be partitioned into image blocks, which can also be referred to as tree blocks, coding units (CU), and / Or coding nodes. Spatial predictions about reference samples in neighboring blocks in the same image are used to encode image blocks in the to-be-encoded (I) slice of the image. The image blocks in the to-be-intercoded (P or B) slice of the image may use spatial prediction relative to reference samples in neighboring blocks in the same image or temporal prediction relative to reference samples in other reference images. An image may be referred to as a frame, and a reference image may be referred to as a reference frame.

In the process of encoding an image block, it is necessary to perform prediction, transformation, quantization, and entropy encoding on the image block to obtain an encoded code stream. In the process of predicting an image block, an inter prediction mode or an intra prediction mode is often used for prediction. When the intra prediction mode is used for prediction, after obtaining the residual block of the image block, an optimal transform check is selected from a variety of transform checks based on the rate distortion, and then the residual block is performed according to the optimal transform check. Transform to get the transform coefficient.

Therefore, how to determine the target transformation check in the process of encoding and decoding the image block, and then realize the transformation and inverse transformation according to the target transformation check is a problem to be solved.

Summary of the invention

The present application provides a transform method, an inverse transform method, and a device to implement an inverse transform on a residual coefficient.

In a first aspect, an inverse transform method is provided. The method includes: parsing a bitstream to obtain a residual coefficient of a current block, an intra prediction mode, and a target transform check to determine parameters; a first transform check, a second transform check, And a third transformation kernel determines a target transformation check corresponding to the target transformation check determination parameter; wherein the second transformation check is a first candidate transformation check or a second transformation candidate transformation check, the first candidate transformation check and the second transformation candidate The transformation check includes the transformation kernel DST-VII, the first candidate transformation check and the second transformation candidate check correspond to the intra prediction mode corresponding to the current block; according to the target transformation check, the residual coefficient is inversely transformed to obtain the residual block. .

It should be understood that parsing the bitstream to obtain the residual coefficient of the current block may specifically refer to firstly parsing the quantized coefficient from the bitstream and then dequantizing the quantized coefficient to obtain the residual coefficient, which can be used for inverse transformation .

In this application, the target transformation check can be determined through the determined target transformation check determination parameters, and the inverse transformation of the residual coefficient can be realized according to the target transformation check.

Further, since the first candidate transformation check and the second transformation check both include a transformation kernel DST-VII, it is possible to increase the possibility of including the transformation kernel DST-VII in the final target transformation check, so that in the target transformation check, A better inverse transform effect may be obtained when inverse transforming the residual coefficients.

Specifically, because the intra prediction mode uses the top and right reference pixels of the current block to predict the current block, and the prediction error tends to gradually increase from top to bottom from left to right, the base of DST-VII The function is gradually increasing. Therefore, the DST-VII transformation kernel is suitable for this gradually increasing characteristic. Using DST-VII for transformation and inverse transformation can achieve better results.

Optionally, the target transformation check determination parameter includes a first flag bit, or the target transformation check determination parameter includes a first flag bit and a transformation index value.

Optionally, determining the target transformation check corresponding to the determination parameter corresponding to the target transformation check determination from the first transformation check, the second transformation check, and the third transformation check includes: when the first flag bit is the first value, The target transformation check corresponding to the target transformation check determination parameter is determined to be the first transformation check; when the first flag bit is the second value and the transformation index value is the first transformation index value, the target corresponding to the target transformation check determination parameter is determined The transformation check is the second transformation check; when the first flag bit is the second value and the transformation index value is the second transformation index value, the target transformation check corresponding to the target transformation check determination parameter is determined to be the third transformation check.

Optionally, the total number of types of the transformation kernels included in the first transformation check, the first candidate transformation check, the second candidate transformation check, and the third transformation check is less than 5.

In the present application, when the type of the transformation kernel is less than 5, the storage space occupied by storing the transformation kernel can be reduced.

Optionally, the transformation kernels included in the first transformation check, the first candidate transformation check, the second candidate transformation check, and the third transformation check include DCT-II, DCT-VIII, and DST-VII.

Optionally, the parameter value of the target transformation check determination parameter may also have a direct mapping relationship with the transformation check, so that the target transformation kernel may be directly determined according to the parameter value of the target transformation check determination parameter.

Optionally, determining a target transformation check corresponding to the target transformation check determination parameter from the first transformation check, the second transformation check, and the third transformation check includes: determining the parameter value of the parameter and the target transformation according to the target transformation check. The mapping relationship between the parameter value of the determined parameter and the transformation check is checked, and the target transformation check is determined.

Specifically, when the parameters determined according to the target transformation check are respectively the first parameter value to the third parameter value, the first transformation check to the third transformation check are respectively corresponding. Then, when the decoding end resolves that the target transformation check determination parameter is the first parameter value, it can directly determine that the target transformation check is the first transformation check.

Optionally, the first flag bit is an explicit multi-core transform EMT flag bit.

The above-mentioned EMT flag can also be called an EMT encoding flag.

Optionally, the first value is 0, and the second value is 1.

Specifically, when the value of the EMT flag is 0, it means that a single-core inverse transformation is used when inversely transforming the residual coefficient (correspondingly, the single-core transformation is also used when the encoding end is transformed); when the EMT flag is set, A value of 1 indicates that a multi-core inverse transform is used when inverse transforming the residual coefficient (correspondingly, the multi-core transform is also used when the encoding end performs the transform).

Optionally, the first transformation check is a transformation check composed of the same transformation kernel.

For example, the first transformation check is (DCT-II, DCT-II).

In some implementations of the first aspect, the first candidate transformation check is (DST-VII, DCT-VIII), and the second candidate transformation check is (DCT-VIII, DST-VII); within a frame corresponding to the current block When the mode number of the prediction mode is less than or equal to the preset threshold, the second transformation check is the first candidate transformation check; when the mode number of the intra prediction mode corresponding to the current block is greater than the preset threshold, the first transformation check is the second candidate. Transform check.

Optionally, the preset threshold is N. When the H.265 standard is adopted, N = 18, and when the H.266 standard is adopted, N = 34.

In some implementations of the first aspect, the third transformation check is (DST-VII, DST-VII).

In some implementations of the first aspect, the first transformation check is (DCT-II, DTC-II).

In some implementations of the first aspect, the first transformation index value is 0, the second transformation index value is 1, or the first transformation index value is 1 and the second transformation index value is 0.

In some implementations of the first aspect, the lengths of the codewords corresponding to the first transformation index value and the second transformation index value are both 1.

In this application, the lengths of the codewords corresponding to the transformation index value are all 1, which can reduce the number of occupied bits required to indicate the transformation index value.

Optionally, the codeword corresponding to the first transformation index value is 0, and the codeword corresponding to the second transformation index value is 1.

Optionally, the codeword corresponding to the first transformation index value is 1, and the codeword corresponding to the second transformation index value is 0.

In some implementation manners of the first aspect, before the target transformation check is determined, the above method further includes: determining an intra prediction mode corresponding to the current block as a directional prediction mode.

In some implementation manners of the first aspect, before determining the target transformation check, the above method further includes: determining that the width and height of the current block are both less than or equal to M.

In some implementations of the first aspect, when at least one of the width and height of the current block is greater than M, (DCT-II, DTC-II) is determined as the target transformation check.

Optionally, the value of M is 64.

In this application, when the size of the current block is large, (DCT-II, DTC-II) is directly used as the target transformation check to perform inverse transformation. When the size of the current block is small, the parameters are finally determined according to the target transformation check to determine the target. The transformation kernel can flexibly process image blocks of different sizes.

In some implementation manners of the first aspect, the above method further includes: when the intra prediction mode corresponding to the current block is a DC mode, determining transformation check (DCT-II, DTC-II) as a target transformation check.

In some implementation manners of the first aspect, the above method further includes: when the intra prediction mode corresponding to the current block is a planar mode, determining a transformation check (DST-VII, DST-VII) as a target transformation check.

In the present application, when the intra prediction mode is a DC mode or a planar mode, by using a preset transformation check directly to perform inverse transformation, the process of determining the target transformation kernel can be simplified and the time required for the inverse transformation can be reduced.

Optionally, when the intra prediction mode corresponding to the current block is a planar mode and the width and height of the current block are both less than or equal to N, a transformation check (DST-VII, DST-VII) is determined as the target transformation check.

The value of N can be 4, 8, and 16 and so on.

Since the residuals appearing in the planar mode tend to increase from left to right and from top to bottom, this type of signal is suitable for DST-VII. Therefore, (DST-VII, DST-VII) is used as the target in the planar mode. Transform check can improve the effect of inverse transform.

In a second aspect, an inverse transform method is provided. The method includes: parsing a bitstream to obtain a residual coefficient of a current block and a target transform check determination parameter; a first transform check, a second transform check, a third transform check, and A target transformation check corresponding to the target transformation check determination parameter is determined in the fourth transformation check; wherein the second transformation check, the third transformation check, and the fourth transformation check all include a transformation kernel DST-VII, so The second transformation check, at least one of the third transformation check and the fourth transformation check includes a transformation kernel DCT-VIII; according to the target transformation check, performing an inverse transformation on the residual coefficient to obtain a residual Bad block.

It should be understood that parsing the bitstream to obtain the residual coefficient of the current block may specifically refer to firstly parsing the quantized coefficient from the bitstream, and then dequantizing the quantized coefficient to obtain the residual coefficient, which can be used for inverse transformation .

Optionally, determining a target transformation check corresponding to the target transformation check determination parameter from the first transformation check, the second transformation check, the third transformation check, and the fourth transformation check includes: when the first flag bit is When it is the first value, it is determined that the target transformation check corresponding to the target transformation check determination parameter is the first transformation check; when the first flag bit is the second value and the transformation index value is the first transformation index value To determine a target transformation check corresponding to the target transformation check determination parameter as a second transformation check; when the first flag bit is a second value and the transformation index value is a second transformation index value, determine the target transformation The target transformation check corresponding to the transformation check determination parameter is a third transformation check; when the first flag bit has a second value and the transformation index value is a third transformation index value, determining a parameter corresponding to the target transformation check determination parameter The target transformation check is a fourth transformation check.

Optionally, the total number of types of transformation kernels included in the first transformation check, the second transformation check, the third transformation check, and the fourth transformation check is less than 5.

Optionally, determining, from the first transformation check, the second transformation check, the third transformation check, and the fourth transformation check, a target transformation check corresponding to the determined and target transformation check determination parameter includes: determining the parameter according to the target transformation check. The parameter value and the target transformation check determine the mapping relationship between the parameter value of the parameter and the transformation check to determine the target transformation check.

Specifically, when the target transformation check determination parameters are respectively the first parameter value to the fourth parameter value, they correspond to the first transformation check to the fourth transformation check, respectively. In this way, when the decoding end resolves that the target transformation check determination parameter is the first parameter value, it can directly determine that the target transformation check is the first transformation check.

Optionally, the first value is 0, and the second value is 1.

Specifically, when the value of the EMT flag is 0, it means that a single-core inverse transformation is used when the inverse transformation is performed on the residual coefficient (correspondingly, the single-core transformation is also used when the encoding end is transformed); A value of 1 indicates that a multi-core inverse transform is used when inverse transforming the residual coefficient (correspondingly, the multi-core transform is also used when the encoding end performs the transform).

When the first transform check is used to perform the inverse transform, a single-core inverse transform is performed.

In some implementations of the second aspect, the length of the codeword corresponding to the first transformation index value is 1, and the length of the codeword corresponding to the second transformation index value and the third transformation index value are both 2.

By setting the length of the codeword of the first transformation index value to 1, the number of bits required to indicate the transformation index value can be reduced.

Optionally, the first to third transformation index values are 0, 1, and 2, respectively.

Optionally, the codeword corresponding to the first transformation index value is 0, the codeword corresponding to the second transformation index value is 10, and the codeword corresponding to the third transformation index value is 11.

Optionally, the codewords corresponding to the second transformation index value and the third transformation index value are 10 and 01, respectively.

In some implementations of the second aspect, the first transformation check is (DCT-II, DTC-II), the second transformation check is (DST-VII, DST-VII), and the third transformation check is (DCT-VIII , DST-VII), and the fourth transformation check is (DST-VII, DCT-VIII).

In this application, there are four types of transformation kernels included in the first to fourth transformation verifications, which are smaller than five in the existing JEM (H.266 reference software model) standard, which can reduce the occupation of storing the transformation kernels. Of bits.

In some implementation manners of the second aspect, before determining the target transformation check, the above method further includes: determining an intra prediction mode corresponding to the current block as a directional prediction mode.

In some implementations of the second aspect, before the target transformation check is determined, the above method further includes: determining that a width and a height of the current block are both less than or equal to M.

In some implementations of the second aspect, when at least one of the width and height of the current block is greater than M, (DCT-II, DTC-II) is determined as the target transformation check.

Optionally, the value of M is 64.

When the size of the current block is large, (DCT-II, DTC-II) is directly used as the target transformation check to perform the transformation, and when the current block size is small, the target transformation kernel is finally determined according to the determination parameters of the target transformation check, which can be flexible Process image blocks of different sizes.

In some implementation manners of the second aspect, the above method further includes: when the intra prediction mode corresponding to the current block is a DC mode, determining a transformation check (DCT-II, DTC-II) as a target transformation check.

In some implementation manners of the second aspect, the above method further includes: when the intra prediction mode corresponding to the current block is a planar mode, determining a transformation check (DST-VII, DST-VII) as a target transformation check.

When the intra prediction mode is a DC mode or a plane mode pair, the inverse transformation can be simplified by directly using a preset transformation check, which can simplify the process of determining the target transformation kernel and reduce the time required for the inverse transformation.

The value of N can be 4, 8, and 16 and so on.

The residuals appearing in the flat mode show an increasing trend from left to right and from top to bottom. This type of signal is suitable for DST-VII. Therefore, (DST-VII, DST-VII) is used as the target in the flat mode. Transform check can improve the effect of inverse transform.

According to a third aspect, a transformation method is provided. The method includes: obtaining a residual block of an image block to be processed; obtaining a candidate transformation check of the residual block according to preset first mapping relationship information; and selecting from the candidate transformation check The transformation check with the lowest rate distortion is used as the target transformation check. According to the target transformation check, the residual block is transformed to obtain the transformation coefficient of the image block to be processed. The corresponding transformation index value of the target transformation check is written into the code stream.

The first mapping relationship information includes a mapping relationship of a first flag bit, a first transformation index value and a second transformation index value, and a variety of transformation checks. When the first flag bit is the first value, a single-core transformation is used. Multi-core transformation is used when the flag is the second value;

In addition, when the first flag bit is the first value, it corresponds to the first transformation check. When the first flag bit is the second value, the first transformation index value corresponds to the second transformation check. When the first flag bit is the second value, the first transformation bit is checked. The second transformation index value corresponds to the third transformation check, and the second transformation check is selected from the first candidate transformation check and the second candidate transformation check according to the intra prediction mode corresponding to the residual block, where the first candidate transformation check and The second transformation candidate transformation check all includes DST-VII.

In this application, because the first mapping relationship information includes only three transformation checkups, the number of traversals can be reduced in the process of selecting a target transformation checkup, thereby reducing the time required for encoding.

Optionally, the first value is 0, and the second value is 1.

Specifically, the first mapping relationship information includes an EMT flag bit. When the value of the EMT flag bit is 0, it means that single-core transformation is used when transforming the residual block; when the value of the EMT flag bit is 1, it means that Multi-core transform is used to transform the residual block.

For example, the first transformation check is (DCT-II, DCT-II).

Optionally, the first transformation index value is 0, and the second transformation index value is 1.

In some implementations of the third aspect, the first candidate transformation check is (DST-VII, DCT-VIII), and the second candidate transformation check is (DCT-VIII, DST-VII); when the frame corresponding to the residual block is When the mode number of the intra prediction mode is less than or equal to a preset threshold, the second transformation check is the first candidate transformation check; when the mode number of the intra prediction mode corresponding to the residual block is greater than the preset threshold, the first transformation check is the first Two candidate transformation checks.

Optionally, the preset threshold is N. When the H.265 standard is used for encoding, N = 18, and when the H.266 standard is used for encoding, N = 34.

In some implementations of the third aspect, the third transformation check is (DST-VII, DST-VII).

In some implementations of the third aspect, the first transformation index value is 0, the second transformation index value is 1, or the first transformation index value is 1 and the second transformation index value is 0.

In some implementations of the third aspect, the lengths of the codewords corresponding to the first transformation index value and the second transformation index value are both 1.

In some implementations of the third aspect, before obtaining candidate transformation check of the residual block according to the preset first mapping relationship information, the method further includes: determining an intra prediction mode corresponding to the residual block as a directional prediction mode.

In some implementations of the third aspect, before obtaining candidate transformation check of the residual block according to the preset first mapping relationship information, the method further includes: determining that the width and height of the residual block are both less than or equal to M.

In some implementations of the third aspect, when at least one of the width and height of the residual block is greater than M, (DCT-II, DTC-II) is determined as the target transformation check.

Optionally, the value of M is 64.

In this application, when the size of the residual block is large, (DCT-II, DTC-II) is directly used as the target transformation check to perform the transformation, and when the size of the residual block is small, the target transformation is finally determined according to the first mapping relationship information. The kernel can achieve better transformation of residual blocks of different sizes.

In some implementation manners of the third aspect, the above method further includes: when the intra prediction mode corresponding to the residual block is a DC mode, determining transformation check (DCT-II, DTC-II) as a target transformation check.

In some implementation manners of the third aspect, the above method further includes: when the intra prediction mode corresponding to the residual block is a planar mode, determining a transformation check (DST-VII, DST-VII) as a target transformation check.

In this application, when the DC mode and the plane mode are used to quantize the residual block, a preset transformation check can be directly used for quantization, which can save the traversal process and further reduce the time required for encoding.

Optionally, when the intra prediction mode corresponding to the residual block is a planar mode and the width and height of the residual block are both less than or equal to N, a transformation check (DST-VII, DST-VII) is determined as the target transformation check .

The value of N can be 4, 8, and 16 and so on.

According to a fourth aspect, a transformation method is provided. The method includes: obtaining a residual block of an image block to be processed; obtaining a candidate transformation check of the residual block according to preset first mapping relationship information; and selecting from the candidate transformation check The transformation check with the lowest rate distortion is used as the target transformation check. According to the target transformation check, the residual block is transformed to obtain the transformation coefficient of the image block to be processed. The corresponding transformation index value of the target transformation check is written into the code stream.

The first mapping relationship information includes a mapping relationship of a first flag bit, a first transformation index value, a second transformation index value, a third transformation index value, and four transformation checklists. The first flag bit is a first value and a first mapping bit. Single-core transformation and multi-core transformation are used when the two values are taken. The first flag bit corresponds to the first transformation check when the first value is the first value. When the first flag bit is the second value, the first transformation index value and the second transformation index value are used. And the third transformation index value corresponds to the second transformation check, the third transformation check, and the fourth transformation check. Any one of the second transformation check to the fourth transformation check includes the transformation kernel DST-VII, and the second transformation check to At least one transformation check in the fourth transformation check includes a transformation kernel DCT-VIII.

In the present application, since the first mapping relationship information includes only four transformation checkups, the number of traversals can be reduced in the process of selecting a target transformation checkup, thereby reducing the time required for encoding.

Further, since the transformation kernel DST-VII is included in each transformation check corresponding to the first flag bit being the second value, and DST-VII has a higher usage rate in intra prediction transformation, therefore, according to the A mapping relationship information can select a transformation check with a higher usage rate for transformation, which can improve the transformation effect.

Optionally, the first flag bit is an explicit multiple core (EMT) flag.

Optionally, the first value is 0, and the second value is 1.

In some implementations of the fourth aspect, the length of the codeword corresponding to the first transformation index value is 1, and the length of the codeword corresponding to the second transformation index value and the third transformation index value are both 2.

In some implementations of the fourth aspect, the first transformation check is (DCT-II, DTC-II), the second transformation check is (DST-VII, DST-VII), and the third transformation check is (DCT-VIII , DST-VII), and the fourth transformation check is (DST-VII, DCT-VIII).

In some implementations of the fourth aspect, before obtaining candidate transformation check of the residual block according to the preset first mapping relationship information, the method further includes: determining an intra prediction mode corresponding to the residual block as a directional prediction mode.

In some implementations of the fourth aspect, before obtaining candidate transformation check of the residual block according to the preset first mapping relationship information, the method further includes: determining that the width and height of the residual block are both less than or equal to M.

In some implementations of the fourth aspect, when at least one of the width and height of the residual block is greater than M, (DCT-II, DTC-II) is determined as the target transformation check.

Optionally, the value of M is 64.

In some implementation manners of the fourth aspect, the above method further includes: when the intra prediction mode corresponding to the residual block is a DC mode, determining a transform check (DCT-II, DTC-II) as a target transform check.

In some implementation manners of the fourth aspect, the above method further includes: determining a transformation check (DST-VII, DST-VII) as a target transformation check when the intra prediction mode corresponding to the residual block is a plane mode.

The value of N can be 4, 8, and 16 and so on.

According to a fifth aspect, an inverse transform apparatus is provided, and the apparatus includes a module for performing a method in any one of the foregoing implementation manners of the first aspect or the second aspect.

According to a sixth aspect, a transformation apparatus is provided, and the apparatus includes a module for executing a method in any one of the foregoing third or fourth aspects.

According to a seventh aspect, a decoder is provided, including: a non-volatile memory and a processor coupled to each other, the processor calling program code stored in the memory to execute any of the first aspect or the second aspect Some or all steps of a method in one implementation.

According to an eighth aspect, an encoder is provided, including: a non-volatile memory and a processor coupled to each other, the processor invoking program code stored in the memory to execute any of the third aspect or the fourth aspect Some or all steps of a method in one implementation.

According to a ninth aspect, a computer-readable storage medium is provided. The computer-readable storage medium stores program code, where the program code includes a program for executing the first aspect, the second aspect, the third aspect, and the fourth aspect. An instruction in some or all of the steps of a method in any of the implementations.

According to a tenth aspect, a computer program product is provided, and when the computer program product runs on a computer, the computer is caused to execute any one of the first aspect, the second aspect, the third aspect, and the fourth aspect. Instructions in some or all steps of the method.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a schematic flowchart of an intra coding process;

FIG. 2 is a schematic flowchart of an intra decoding process;

3 is a schematic diagram of various intra prediction modes;

4 is a schematic flowchart of an inverse transform method according to an embodiment of the present application;

5 is a schematic flowchart of an inverse transform method according to an embodiment of the present application;

6 is a schematic flowchart of an inverse transform method according to an embodiment of the present application;

7 is a schematic flowchart of a conversion method according to an embodiment of the present application;

8 is a schematic flowchart of a conversion method according to an embodiment of the present application;

FIG. 9 is a schematic flowchart of a conversion method according to an embodiment of the present application; FIG.

10 is a schematic block diagram of an inverse transform device according to an embodiment of the present application;

11 is a schematic block diagram of a video decoder according to an embodiment of the present application;

FIG. 12 is a schematic block diagram of a video encoder according to an embodiment of the present application.

detailed description

The technical solutions in this application will be described below with reference to the drawings.

In order to better understand the embodiments of the present application, the main processes of intra-frame video coding and intra-frame video decoding are briefly introduced below with reference to FIG. 1 and FIG. 2. It should be understood that the intra video encoding here means that the prediction mode used in the process of encoding the video image is the intra prediction mode, and the intra video decoding means that the prediction mode used in the process of decoding the video image is the frame Intra prediction mode.

As shown in FIG. 1, by performing intra prediction on the image to be processed quickly, a prediction block of the image block to be processed can be obtained, and by making a difference between the image block to be processed and the prediction block (also called a prediction signal), a residual block can be obtained. (Can also be a residual signal) to remove redundancy between images. Next, transform / quantize the residual block to obtain a quantized coefficient, and perform entropy coding on the quantized coefficient to obtain a code stream. At the same time, in the encoding process, the quantization coefficients need to be inverse quantized / inverse transformed to obtain the residual block. Next, the residual block and the prediction block are added to obtain a reconstructed block (a loop can be performed before the reconstructed block is obtained Road filtering).

In the above process, intra-frame prediction can be performed from different directions using the encoded reconstruction block (reconstruction information) as reference information, and then the residual signal is transformed so that the energy of the residual signal is further concentrated, and then quantized Use a larger quantization step size for high-frequency signals and a smaller quantization step size for low-frequency signals to ensure that there is no loss of basic image information.

As shown in Figure 2, at the decoding end, encoding information such as quantization coefficients and coding modes is obtained by entropy decoding the code stream (binary code stream information), and the residual blocks are obtained after inverse quantization / inverse transformation of the quantized coefficients. The residual block is added to the intra prediction block and then loop filtered to obtain the final reconstruction speed. At the same time, the generated reconstruction block will also be used in the subsequent intra prediction process of other image blocks.

In the latest video coding standard being developed by the Joint Video Experts Group (JVET), a direction diagram is shown for intra prediction using 67 intra-angle modes, as shown in FIG. 3. More dense angles can bring better prediction results. The number of angle modes for intra prediction has increased from 33 to 65 in HEVC. When the mode number is 0, it indicates a planar mode; when the mode number is 1, , Indicates the DC mode; the plane mode and the DC mode do not involve the prediction direction, and are called non-directional prediction modes; when the mode number is 2-66, it represents the angle prediction mode (also known as the directional prediction mode), and different mode numbers represent Different prediction angles are shown. As shown in FIG. 3, the prediction angle corresponding to the prediction mode of the mode number 2-66 is indicated by a solid line or a dotted line.

The denser angular directions can more effectively capture the texture boundaries in any direction in natural video, and can use more reference information for rate distortion cost selection, so that more suitable intra prediction modes can be found to more accurately predict the current encoding. Block, which makes the predicted block closer to the original coded block, and fewer bits need to be transmitted to encode the current block, resulting in higher coding performance. When performing intra-frame coding, these 67 intra-modes are used on coding blocks of all sizes, including luminance components and chrominance components.

The embodiments of the present application will be described in detail below with reference to FIGS. 4 to 8.

FIG. 4 is a schematic flowchart of an inverse transform method according to an embodiment of the present application. The method shown in FIG. 4 may be executed by a decoding device. The method shown in FIG. 4 includes steps 110 to 130. The steps 110 to 130 are described in detail below.

110. Parse the bitstream to obtain the residual coefficient, the intra prediction mode, and the target transformation of the current block to determine the parameters.

After the code stream is obtained, operations such as entropy decoding on the code stream can be used to obtain the residual coefficient of the current block, the intra prediction mode, and the target transformation check and determination parameters.

The above-mentioned target transformation check determination parameter is information used to determine the target transformation check of the residual coefficient of the current block. The target transformation check determination parameter can directly indicate only the target transformation check through a single message, or the target transformation check determination parameter can also The multiple parameters are used to collectively indicate the target transformation check.

The above parsing the bitstream to obtain the residual coefficient of the current block may specifically refer to firstly parsing the quantized coefficient from the bitstream, and then dequantizing the quantized coefficient to obtain the residual coefficient, which can be used for inverse transformation.

120. Determine a target transformation check corresponding to the target transformation check determination parameter from the first transformation check, the second transformation check, and the third transformation check.

The second transformation check is the first candidate transformation check or the second change candidate transformation check. The first candidate transformation check and the second change candidate transformation check both include a transformation kernel DST-VII. The first candidate transformation check and the second transformation check. The candidate transformation check corresponds to the intra prediction mode corresponding to the current block.

Optionally, the first candidate transformation check is (DST-VII, DCT-VIII), and the second candidate transformation check is (DCT-VIII, DST-VII); when the mode of the intra prediction mode corresponding to the residual block is When the number is less than or equal to a preset threshold, the second transformation check is a first candidate transformation check; when the mode number of the intra prediction mode corresponding to the residual block is greater than the preset threshold, the first transformation The collation is a second candidate transformation collation.

When the above target transformation check determination parameter directly instructs the target transformation check through a single piece of information, the target transformation check may be determined through a mapping relationship or a corresponding relationship between the parameter value of the target transformation check determination parameter and the transformation check.

Specifically, when the target transformation check determination parameters are respectively the first parameter value to the third parameter value, they correspond to the first transformation check to the third transformation check, respectively. Then, when the decoding end resolves that the target transformation check determination parameter is the first parameter value, it can directly determine that the target transformation check is the first transformation check.

The corresponding relationship between the parameter values of the target transformation check determination parameters and the transformation check can be shown in Table 1. As shown in Table 1, after the decoder obtains the parameter values of the target transformation check determination parameters, it can directly map according to Table 1. Relationship determination target transformation check. The mapping relationship shown in Table 1 can be stored at the encoding end and the decoding end. After the encoding end determines the target transformation check, the parameter value of the target transformation check corresponding to the target transformation check can be written into the code stream. The decoding end can determine the target transformation check through the parsed target transformation check to determine the parameter values of the determined parameters and the mapping relationship shown in Table 1.

Table 1

目标变换核对确定参数的参数值The target transformation checks the parameter values of the determined parameters	目标变换核对Target transformation check
第一参数值First parameter value	第一变换核对First transformation check
第二参数值Second parameter value	第二变换核对Second transformation check
第三参数值Third parameter value	第三变换核对Third transformation check

Further, the correspondence between the parameter values of the target transformation check determination parameter and the transformation check can be shown in Table 2. As shown in Table 2, when the parameter values of the target transformation check determination parameter value are 0, 10 and 11, respectively, the corresponding target transformations are The checks are (DCT-II, DCT-II), (DST-VII, DCT-VIII), and (DST-VII, DST-VII).

Table 2

目标变换核对确定参数的参数值The target transformation checks the parameter values of the determined parameters	目标变换核对Target transformation check
00	DCT-II，DCT-IIDCT-II, DCT-II
1010	DST-VII，DCT-VIIIDST-VII, DCT-VIII
1111	DST-VII，DST-VIIDST-VII, DST-VII

It should be understood that the above Tables 1 and 2 are merely examples of a manifestation of the correspondence between the parameter values of the target transformation check determination parameters and the transformation check. The specific form is not limited, as long as there is a certain correspondence between the parameter values of the parameters determined by the target transformation check and the transformation check.

Optionally, determining, from the first transformation check, the second transformation check, and the third transformation check, the target transformation check corresponding to the determination parameter corresponding to the target transformation check, specifically includes the following process:

(1) When the first flag bit is the first value, determine that the target transformation check corresponding to the target transformation check determination parameter is the first transformation check;

(2) when the first flag bit has a second value and the transformation index value is the first transformation index value, determining a target transformation check corresponding to the target transformation check determination parameter as a second transformation check;

(3) When the first flag bit has a second value and the transformation index value is a second transformation index value, it is determined that the target transformation check corresponding to the target transformation check determination parameter is a third transformation check.

Optionally, the first flag bit is an explicit multiple core transform (EMT) flag bit.

Optionally, the first value is 0, and the second value is 1.

Optionally, the types of the transformation kernels included in the first transformation check, the first candidate transformation check, the second candidate transformation check, and the third transformation check may be a part of the five transformation kernels defined in the JEM. The group of transform kernels are specifically DCT-II, DCT-V, DCT-VIII, DST-1 and DST-VII. The basic information of the five sets of transformation kernels is shown in Table 3.

table 3

There may be a certain correspondence relationship (mapping relationship) between the first flag bit, the transformation index value, and the target transformation check, and the specific form of the correspondence relationship may be a table.

For example, the correspondence between the first flag bit, the transformation index value, and the target transformation check may be shown in Table 4.

Table 4

It can be known from Table 4 that when the first flag bit is the first value, there is only a corresponding relationship between the target transformation check and the first flag bit, and the target transformation check is the first transformation check. When the first flag bit is the second value, there is a corresponding relationship between the target transformation check and the first flag bit and the transformation index value. When the first flag bit is the second value and the transformation index value is the first The target transformation check corresponding to each of the transformation index value and the second transformation index value are a second transformation check and a third transformation check.

Correspondence between the first flag bit, the transformation index value, and the target transformation check when the first transformation index value and the second transformation index value are 0 and 1, respectively, and the first and second values are 0 and 1, respectively. The relationship can be shown in Table 5.

table 5

It should be understood that the foregoing first transformation index value and the second transformation index value may also be 1 and 0, respectively, and the first and second values may also be 1 and 0, respectively.

Optionally, the length of the codeword corresponding to the first transformation index value and the second transformation index value is both 1.

Table 6 is obtained by adding the codewords corresponding to the first transformation index value and the second transformation index value to Table 5.

Table 6

第一标志位First flag	目标变换核对Target transformation check	变换索引值Transform index	码字numbers
00	第一变换核对First transformation check	N/AN / A	N/AN / A
11	第二变换核对Second transformation check	00	00

Zh

Third transformation check

1

As can be seen from Table 6, the lengths of the codewords corresponding to the transformation index value are all 1, which can reduce the number of occupied bits required to indicate the transformation index value.

It should be understood that the codewords corresponding to the first transformation index value and the second transformation index value may also be 1 and 0, respectively.

Optionally, the third transformation check is (DST-VII, DST-VII), and the first transformation check is (DCT-II, DTC-II).

Fill in the specific transformation check indicated by the first transformation check and the third transformation check into Table 6, and adjust the order of the second transformation check and the third transformation check to obtain Table 7.

Table 7

As shown in Table 7, the second transformation check may be determined according to the intra prediction mode corresponding to the current block. Specifically, when the mode number of the intra prediction mode is less than or equal to 34, the second transformation check in Table 7 is (DST-VII, DCT-VIII); when the mode number of the intra prediction mode is greater than 34, Table 7 The second transformation check in (DST-VII, DCT-VIII).

In addition, the second transformation check corresponding to each intra prediction mode can also be determined by looking up the table. For example, after the intra prediction mode corresponding to the current block is known, the foregoing second transformation check may be determined according to the correspondence between the prediction mode number of the intra prediction mode and the index number of the transformation check in Table 8 below.

Table 8

预测模式号Prediction mode number	00	11	22	33	44	55	66	77	88	99	1010	1111	1212
变换核对 Transformation check	22	22	00	11	00	11	00	11	00	11	00	11	00
预测模式号 Prediction mode number	1313	1414	1515	1616	1717	1818	1919	2020	21twenty one	22twenty two	23twenty three	24twenty four	2525
变换核对Transformation check	11	00	11	00	11	00	11	00	11	00	11	00	11
预测模式号 Prediction mode number	2626	2727	2828	2929	3030	3131	3232	3333	3434	Zh	Zh	Zh	Zh
变换核对Transformation check	00	11	00	11	00	11	00	11	00	Zh	Zh	Zh	Zh

Table 8 contains a total of 35 intra prediction modes. The mode number of the intra prediction mode is 0-34. Table 7 contains three kinds of transformation checks. These three kinds of transformation checks are represented by the numbers 0-2, which are represented by 0-2. The transformation check is as follows:

0: transform check 0, (DST-VII, DCT-VIII);

1: transformation check 1, ie (DCT-VIII, DST-VII);

2: Transform check 2, ie (DCT-VIII, DCT-VIII).

In the above-mentioned transformation kernel 0 to transformation kernel 2, the former transformation kernel represents a horizontal transformation kernel, and the latter transformation kernel represents a vertical transformation kernel. Of course, the former may represent a vertical transformation kernel, and the latter may represent a horizontal transformation kernel.

Assume that the first transformation check and the third transformation check are shown in Table 7. The second transformation check in Table 7 can be determined according to the relationship in Table 8. Then, although the first transformation check to the third transformation check use 4 Kinds of transformation check (including DCT-II, DCT-II), but for each prediction mode, there are only 3 transformation check options available, and only 3 RDOs are required when encoding at the encoding end, thereby reducing encoding time Overhead. Here, because more 4 kinds of transformation check are used, and the mapping relationship between the transformation check and the prediction mode is more flexible (as can be seen from Table 7, for the angle prediction mode, 0 and 1 alternately appear), you can Achieve better performance with the same encoding time overhead.

In addition, for the specific transformation kernel selection in Table 8, the mirror DST-VII can also be used to replace DCT-VIII in this scheme. In terms of the entire transformation scheme, the types of transformation kernels used have been reduced from three to two, namely DCT-II and DST-VII. When the types of transform kernels are reduced, the memory occupied by the encoding kernel and the decoder to store the transform kernels can be reduced.

130: Perform inverse transformation on the residual coefficient according to the target transformation check to obtain a residual block.

Optionally, as an embodiment, before step 120, the method shown in FIG. 4 further includes: determining an intra prediction mode corresponding to the residual block as a directional prediction mode.

That is, in the method shown in FIG. 4, when the intra prediction mode corresponding to the residual block is a directional prediction mode, the candidate transformation check of the residual block is obtained according to step 120.

Optionally, as an embodiment, before step 120, the method shown in FIG. 4 further includes: determining that the width and height of the residual block are both less than or equal to M.

The value of M can be set according to the situation during the transformation, for example, M can be 64.

By determining the candidate transformation check of the residual block according to the first mapping relationship information when the residual block is smaller than a certain size, the transformation effect can be improved.

Optionally, when at least one of the width and height of the residual block is greater than M, (DCT-II, DTC-II) is determined as the target transformation check.

It should be understood that the method shown in FIG. 4 may be applicable to any intra prediction mode, that is, in the directional prediction mode or the non-directional prediction mode, the candidate transformation check of the residual block may be directly obtained according to step 120.

Alternatively, the method of determining the candidate transformation check of the residual block in step 120 may only be applicable to the directional prediction mode, and for the non-directional prediction mode, a preset transformation check may be directly adopted as the target transformation check.

Optionally, the method shown in FIG. 4 further includes: when the intra prediction mode corresponding to the residual block is a DC mode, determining a transformation check (DCT-II, DTC-II) as a target transformation check.

Optionally, the method shown in FIG. 4 further includes: when the intra prediction mode corresponding to the residual block is a planar mode, determining a transformation check (DST-VII, DST-VII) as a target transformation check.

The value of N can be 4, 8, and 16, etc. Further, the value of N can be a multiple of 4.

That is, when the intra prediction mode is a planar mode, if at least one of the width and height of the current block is greater than N, other preset transformation checks can be used as the target transformation check, for example, the transformation check (DCT- II, DTC-II) is determined as the target transformation check.

FIG. 5 is a schematic flowchart of an inverse transform method according to an embodiment of the present application. The method shown in FIG. 5 may be executed by a decoding device. The method shown in FIG. 5 includes steps 210 to 230. The steps 210 to 230 are described in detail below.

210: Parse the bitstream to obtain the residual coefficient of the current block and the target transformation to check and determine the parameters.

After the code stream is obtained, information such as the residual coefficient of the current block and the target transformation check and determination parameters can be obtained by performing entropy decoding on the code stream.

220. Determine a target transformation check corresponding to the target transformation check determination parameter from the first transformation check, the second transformation check, the third transformation check, and the fourth transformation check.

The second transformation check, the third transformation check, and the fourth transformation check each include a transformation kernel DST-VII, the second transformation check, the third transformation check, and the fourth transformation check. A transformation check includes the transformation kernel DCT-VIII.

Optionally, determining a target transformation check corresponding to the target transformation check determination parameter from the first transformation check, the second transformation check, the third transformation check, and the fourth transformation check includes: determining the parameter according to the target transformation check. The parameter value and the target transformation check determine the mapping relationship between the parameter value of the parameter and the transformation check to determine the target transformation check.

Specifically, when the target transformation check determination parameters are respectively the first parameter value to the fourth parameter value, they correspond to the first transformation check to the fourth transformation check, respectively. Then, when the decoding end resolves that the target transformation check determination parameter is the second parameter value, it can directly determine that the target transformation check is the second transformation check.

The corresponding relationship between the parameter values of the target transformation check determination parameters and the transformation check can be shown in Table 9. The mapping relationship shown in Table 9 can be stored at the encoding end and the decoding end. After the encoding end determines the target transformation check, the parameter value of the target transformation check corresponding to the target transformation check can be written into the code stream. The decoding end can determine the target transformation check through the parsed target transformation check to determine the parameter values of the determined parameters and the mapping relationship shown in Table 9.

Table 9

目标变换核对确定参数的参数值The target transformation checks the parameter values of the determined parameters	目标变换核对Target transformation check
第一参数值First parameter value	第一变换核对First transformation check
第二参数值Second parameter value	第二变换核对Second transformation check
第三参数值Third parameter value	第三变换核对Third transformation check
第四参数值Fourth parameter value	第四变换核对Fourth transformation check

Optionally, the first transformation check is (DCT-II, DTC-II), the second transformation check is (DST-VII, DST-VII), and the third transformation check is (DCT-VIII, DST-VII). The four transform checks are (DST-VII, DCT-VIII).

Adding the first transformation check to the fourth transformation check and adding the specific transformation check to Table 9 can obtain Table 10. In Table 10, when the first parameter value is 0, 01, 10, and 11, the corresponding target transformation check is (DCT). -II, DTC-II), (DST-VII, DST-VII), (DCT-VIII, DST-VII), (DST-VII, DCT-VIII).

Table 10

目标变换核对确定参数的参数值The target transformation checks the parameter values of the determined parameters	目标变换核对Target transformation check
00	DCT-II，DTC-IIDCT-II, DTC-II
0101	DST-VII，DST-VIIDST-VII, DST-VII
1010	DCT-VIII，DST-VIIDCT-VIII, DST-VII
1111	DST-VII，DCT-VIIIDST-VII, DCT-VIII

It should be understood that the above-mentioned Tables 9 and 10 are merely examples of a manifestation of the correspondence between the parameter values of the target transformation check determination parameters and the transformation check. The specific form is not limited, as long as there is a certain correspondence between the parameter values of the parameters determined by the target transformation check and the transformation check.

When the target transformation check determination parameter collectively indicates the target transformation check through multiple parameters, the multiple parameters in the target transformation check determination parameter may be obtained first, and then the target transformation is determined according to the correspondence between the multiple parameters and the target transformation check. Check.

Optionally, determining a target transformation check corresponding to the target transformation check determination parameter from the first transformation check, the second transformation check, the third transformation check, and the fourth transformation check specifically includes the following process:

(1) when the first flag bit has a first value, determining a target transformation check corresponding to the target transformation check determination parameter as a first transformation check;

(3) when the first flag bit has a second value and the transformation index value is a second transformation index value, determining a target transformation check corresponding to the target transformation check determination parameter as a third transformation check;

(4) When the first flag bit has a second value and the transformation index value is a third transformation index value, determining a target transformation check corresponding to the target transformation check determination parameter as a fourth transformation check.

For example, the correspondence between the first flag bit, the transformation index value, and the target transformation check may be shown in Table 11.

Table 11

It can be known from Table 11 that when the first flag bit is the first value, the target transformation check only has a corresponding relationship with the first flag bit, and the target transformation check is the first transformation check. When the first flag bit is the second value, there is a corresponding relationship between the target transformation check and the first flag bit and the transformation index value. When the first flag bit is the second value and the transformation index value is the first The target transformation check corresponding to the transformation index value, the second transformation index value, and the third transformation index value are respectively a second transformation check, a third transformation check, and a fourth transformation check.

When the first transformation index value, the second transformation index value, and the third transformation index value are 0, 1, and 2, respectively, and the first and second values are 0 and 1, respectively, the first flag bit and the transformation index The corresponding relationship between the value and the target transformation check can be specifically shown in Table 12.

Table 12

It should be understood that the first value and the second value may also be 1 and 0, respectively, and the first transformation index value to the third transformation index value may also be other values.

Optionally, the first flag is an EMT flag.

Optionally, the length of the codeword corresponding to the first transformation index value is 1, and the length of the codeword corresponding to the second transformation index value and the third transformation index value are both 2.

When the first flag bit is the EMT flag bit, and the first to third transformation index values are 0, 1, 2, and the codewords of the three index values are 0, 10, and 11, respectively, the first flag bit is The correspondence between the transformation index value and the target transformation check can be shown in Table 13.

Table 13

It should be understood that the codewords corresponding to the second transformation index value and the third transformation index value may also be 10 and 01, respectively.

In the present application, there are four types of transformation kernels included in the first transformation verification to the fourth transformation verification, which are less than five in the existing JEM standard, which can reduce the bits occupied when storing the transformation kernel.

Table 14 can be obtained by filling the first transformation check and the fourth transformation check into Table 13.

Table 14

230: Perform inverse transformation on the residual coefficient according to the target transformation check to obtain a residual block.

Optionally, as an embodiment, before the target transformation check is determined, the method shown in FIG. 5 further includes: determining an intra prediction mode corresponding to the current block as a directional prediction mode.

Optionally, as an embodiment, before determining the target transformation check, the method shown in FIG. 5 further includes: determining that the width and height of the residual block are both less than or equal to M.

Optionally, as an embodiment, when at least one of the width and height of the residual block is greater than M, (DCT-II, DTC-II) is determined as the target transformation check.

Optionally, the value of M is 64.

In this application, when the size of the current block is large, (DCT-II, DTC-II) is directly used as the target transformation check to perform the transformation, and when the size of the current block is small, the determination parameters are finally determined according to the target transformation check to determine the target transformation Core, capable of flexibly processing image blocks of different sizes.

Optionally, as an embodiment, the method shown in FIG. 5 further includes: when the intra prediction mode corresponding to the current block is a DC mode, determining a transform check (DCT-II, DTC-II) as the target transform Check.

Optionally, as an embodiment, the above method further includes: determining a transformation check (DST-VII, DST-VII) as the target transformation check if the intra prediction mode corresponding to the current fast mode is a plane mode.

In the present application, when the intra prediction mode is a DC mode or a plane mode pair, the inverse transformation can be simplified by directly using a preset transformation check, which can simplify the process of determining the target transformation kernel and reduce the time required for the inverse transformation.

Optionally, when the intra prediction mode corresponding to the current block is a planar mode and the width and height of the residual block are both less than or equal to N, a transformation check (DST-VII, DST-VII) is determined as the target transformation Check.

The value of N can be 4, 8, and 16 and so on.

The corresponding conditions of the EMT flag bits, candidate transformation kernels, transformation index values, and codewords in the existing JEM standard are shown in Table 15.

Table 15

In the existing scheme, when the EMT flag bit is 0 core 1 and corresponding to five kinds of transformation check, it is necessary to perform five traversals in the process of rate distortion optimization, which may lead to an increase in the final encoding time.

According to the statistics of the transformation check and transformation check (DCT-II, DCT-II) corresponding to various transformation index values, it is found that the proportion of different transformation checks varies greatly. As shown in Table 16, the transformation check (DCT-II II, DCT-II) accounted for the highest proportion, reaching 43.8%, and the index value 3 accounted for the smallest proportion, only 7.2%. Through analysis, it is found that in the multi-core transformation of all intra prediction modes, the proportion of the transformation check corresponding to the index value 0 to the index value 3 sequentially decreases.

Among them, the horizontal and vertical transformation kernels corresponding to the index values 0,1,2 both use at least one DST-VII transformation kernel, and the index value 3 indicates that neither the horizontal or vertical transformation kernels use DST-VII transformation kernels. The index value 0 corresponds to the (DST-VII, DST-VII) transform kernel, and its proportion is obviously higher than the index values 1,2,3. This is because intra prediction uses the top and left reference pixels to predict the current block. The prediction error tends to gradually increase from top to bottom from left to right. The DST-VII transform kernel is suitable for this type of residual signal. .

Table 16

变换核对Transformation check	DCT-II，DCT-IIDCT-II, DCT-II	索引值0Index value 0	索引值1Index value 1	索引值2 Index value 2	索引值3Index value 3
平均占比Average share	43.8％43.8%	23.2％23.2%	12.8％12.8%	13.0％13.0%	7.2％7.2%

Therefore, this application provides a new mapping relationship. In this mapping relationship, the candidate transformation kernel corresponding to the EMT flag of 0 is still a transformation check (DCT-II, DCT-II), and a new transformation check is added. The number of types is reduced to three, as shown in Table 17. In Table 17, for the newly added transform kernels, at least one DST-VII transform kernel is used for the horizontal and vertical transform kernels according to the proportion of the transform kernels. In addition, in Table 17, when the transformation index value is 0, the codeword is 0, and the index values of 1 and 2 correspond to the codewords of 10 and 11, respectively.

For an index value of 0, the horizontal and vertical transformation kernels are both DST-VII transformation kernels; for an index value of 1, the horizontal transformation kernel is DST-VII, and the vertical transformation kernel is not a DST-VII transformation kernel; for an index value of 2, horizontal transformation The kernel is not a DST-VII transformation kernel, and the vertical transformation kernel is DST-VII; for a combination of transformation kernels with an index value of 3, the traversal process with an index value of 3 is skipped because it has the lowest ratio.

Therefore, in the transformation process at the encoding end, the number of traversals can be reduced and the coding time can be reduced by using the table shown in Table 17. In addition, some transformation index values can correspond to one codeword, which can save the occupation of the transformation index value. Bits.

Table 17

The newly added transformation kernel in Table 17 above can be selected by querying the relationship between the residual block and the corresponding candidate transformation kernel.

For example, the above-mentioned newly added transform kernels can be selected according to Tables 18 and 19.

Table 18

变换候选集Transform candidate set	候选变换核Candidate transformation kernel
00	DST-VII,DCT-VIIIDST-VII, DCT-VIII
11	DST-VII,DST-IDST-VII, DST-I
22	DST-VII,DCT-VDST-VII, DCT-V

Table 19

For example, when the mode number of the intra prediction mode corresponding to the residual block is 51, it is determined according to Table 19 that the transformation candidate set corresponding to the horizontal direction is 2 and the transformation candidate set corresponding to the vertical direction is 0. As shown in Table 18, transformation candidate set 0 contains candidate transformation kernels DST-VII, DCT-VIII, and transformation candidate set 2 contains candidate transformation kernels DST-VII, DCT-V. Then, transformations can be selected from transformation candidate set 0 In the kernel DST-VII, a transformation kernel DCT-V is selected from the transformation candidate set 2 to form a new transformation kernel. There are three combinations of these two kinds of new transformation kernels, and these three cases correspond to index values 0 to 2 respectively.

When performing transformations, when there are many types of transformation kernels, a certain memory overhead is required to store these transformation kernels. In order to reduce the memory overhead, the storage overhead of the transformation kernel and the complexity of the transformation can be reduced by reducing the types of transformation kernels.

Optionally, there may be three types of transform kernels, namely, DCT-II, DCT-VIII, and DST-VII. In this case, the use of the transform kernel can only depend on the transform index value, and has nothing to do with the intra prediction mode, as shown in Table 20.

Table 20

It should be noted that the DCT-VIII in the above table 20 can also be replaced with other transformation kernels with similar characteristics, such as DCT-V, or a variant based on DST-VII, such as DST-VII after flipping. (This is because the characteristics of the DST-VII transformation kernel after mirror transformation are similar to those of DCT-VIII. It should be understood that the mirror transformation here can be used to do left-right mirroring or up-down mirroring.

Further, the types of transformation kernels can also be reduced to two. The two transformation kernels are DCT-II and DST-VII, as shown in Table 21.

Table 21

The inverse transform method according to the embodiment of the present application is described in detail with reference to FIG. 4 and FIG. 5 above, and the process of performing inverse transform at the decoder is described in detail with reference to FIG. 6.

FIG. 6 is a schematic flowchart of an inverse transform method according to an embodiment of the present application. The method shown in FIG. 6 can be executed by a decoding device. The method shown in FIG. 6 includes steps 1001 to 1010. Steps 1001 to 1010 are described in detail below.

1001, start.

Step 1001 indicates that after the residual coefficients of the current block are obtained, the target transformation check corresponding to the residual coefficients is started, so as to perform inverse transformation on the residual coefficients according to the target transformation check.

It should be understood that the residual coefficient of the current block can be obtained by analyzing the code stream.

1002. Determine whether the width (W) and height (H) of the current block are both less than or equal to N.

The value of N can be set to 64 or other values.

When at least one of W and H of the residual block is greater than N, step 1003 is performed, and when both of W and H of the residual block are less than or equal to N, step 1004 is performed.

1003. Determine (DCT-II, DCT-II) as the target transformation check.

1004. Determine whether the intra prediction mode is a directional prediction mode.

It should be understood that the intra prediction mode here is the intra prediction mode corresponding to the current block, and specifically refers to the intra prediction mode corresponding to the encoding end when performing intra prediction.

Step 1005 is performed when the intra prediction mode is a directional prediction mode; step 1007 is performed when the intra prediction mode is a non-directional prediction mode.

1005. Determine whether the EMT flag (or the EMT encoding flag) is 1.

When the EMT flag is not 1 (for example, when it is 0), step 1003 is performed to determine (DCT-II, DCT-II) as the target conversion check; when the EMT flag is 1, step 1006 is performed.

1006. Obtain a transformation index value, and determine a transformation check corresponding to the transformation index value as a target transformation check.

It should be understood that in step 1006, the target transformation check may be specifically determined according to

steps

120 and 220 in the method shown in FIG. 4 and FIG. 5, and the specific process is not described repeatedly.

1007. Determine whether the intra prediction mode is a DC mode.

Step 1008 is performed when the intra prediction mode is a DC mode, and step 1009 is performed when the intra prediction mode is not a DC mode (the intra prediction mode is a planar mode at this time).

1008. Determine (DCT-II, DTC-II) as the target transformation check.

1009. Determine (DST-VII, DST-VII) as the target transformation check.

Since the prediction value of the coded block is obtained by averaging the reference pixel values, the DC mode is suitable for a large area with a flat area, and the DCT-II transform kernel is suitable for such residual signals. Therefore, when the intra prediction mode is the DC mode , Determining (DCT-II, DTC-II) as the target transformation check can improve the transformation effect, and can reduce the number of bits required to indicate the target transformation check in the DC mode.

1010. Perform inverse transformation on the residual coefficients according to the target transformation check to obtain residual blocks.

It should be understood that in the process of determining the target transformation check in FIG. 6, the number of non-zero coefficients may also be comprehensively considered to determine the target transformation check.

In addition, the process shown in FIG. 6 may only be an example of a process of inverse transform performed by a decoder, and the process shown in FIG. 6 may include only a part of a process performed by an encoder.

The inverse transform method of the embodiment of the present application is described in detail above with reference to FIGS. 4 to 6. The transform method of the embodiment of the present application is described below with reference to FIGS. 7 to 9. It should be understood that the The transformation method corresponds to the inverse transformation method shown in FIGS. 4 to 6. The transformation method according to the embodiment of the present application will be described in detail below with reference to FIGS. 7 to 9.

FIG. 7 is a schematic flowchart of a conversion method according to an embodiment of the present application. The method shown in FIG. 7 may be executed by a coding end device. The method shown in FIG. 7 includes steps 310 to 350. The steps 310 to 350 are described in detail below.

310: Obtain a residual block of an image block to be processed.

The above residual block may be obtained by making a difference between the current block (current image block) and the prediction block of the current block according to the process shown in FIG. 1. In the process of obtaining the prediction block of the current block, intra prediction is used. mode. Specifically, when the prediction block of the current block is obtained, and then the residual block of the current block is obtained, various intra prediction modes as shown in FIG. 3 may be adopted.

320: Obtain a candidate transformation check of the residual block according to preset first mapping relationship information.

Optionally, the transformation check in the first mapping relationship information may include at least a part of five sets of transformation kernels. The five sets of transformation kernels are DCT-II, DCT-V, DCT-VIII, and DST-1. And DST-VII.

Optionally, the first value is 0, and the second value is 1.

Optionally, the specific expression form of the first mapping relationship information is a table.

For example, the foregoing first mapping relationship information may be shown in Table 22.

Table 22

As can be seen from Table 22, when the first flag bit is the first value, the first flag bit corresponds to the first transformation check, and when the first flag bit is the second value, the first flag bit corresponds to the two transformation check, and These two transformation checks correspond to the first transformation index value and the second transformation index value, respectively.

It should be understood that the first transformation check may include only one kind of transformation kernel, for example, the first transformation check is (DCT-II, DCT-II).

Optionally, the third transformation check is (DST-VII, DST-VII).

Optionally, the first transformation index value is 0, the second transformation index value is 1, or the first transformation index value is 1 and the second transformation index value is 0.

The foregoing first mapping relationship information may also be specifically shown in Table 23.

Table 23

Taking the coding standard H.266 as an example, when the mode number of the intra prediction mode is less than or equal to 34, the second transformation check in Table 2 is (DST-VII, DCT-VIII); when the mode number of the intra prediction mode is When it is greater than 34, the second transformation check in Table 2 is (DST-VII, DCT-VIII).

In the method shown in FIG. 7, the second transformation check needs to be selected according to the intra prediction mode corresponding to the residual block.

In the H.265 standard, the number of intra prediction modes is 35, including 2 non-directional prediction modes (that is, planar mode and DC mode) and 33 directional prediction modes.

For the H.266 standard being developed, the number of intra prediction modes may be 67, including 2 non-directional prediction modes (planar mode and DC mode) and 65 directional prediction modes. The intra prediction modes can be divided into two or three groups, and each group of intra prediction modes corresponds to a different transformation check.

For example, one possible scenario is as follows:

Transform check 0: (DST-VII, DCT-VIII): when the intra prediction mode is less than or equal to N;

Transform check 1: (DCT-VIII, DST-VII): When the intra prediction mode is greater than N.

The value of N is related to the number of intra prediction modes. For the H265 standard, N may be 2+ (34-2) / 2 = 18, and for the H266 standard under development, N may be 2+ (66-2) / 2 = 34.

The above-mentioned transformation check 0 and transformation check 1 correspond to the first candidate transformation check and the second candidate transformation check respectively.

It should be understood that when the intra prediction mode is divided, the intra prediction mode may also be divided into three or more groups, each group corresponding to a kind of transformation check, and the number of the divided groups of the intra prediction mode in this application. And the correspondence relationship with the conversion check is not limited.

330. Select a transformation check with the lowest rate distortion from the candidate transformation checks as a target transformation check.

It should be understood that each transformation kernel may be used to transform the residual block in turn and perform rate-distortion optimization (RDO) to obtain the rate distortion corresponding to each transformation check for transformation, and then check the transformation with the smallest rate distortion RDO Determined as the target transformation check.

340: Transform the residual block according to the target transformation check to obtain a transformation coefficient of the image block to be processed.

350. Write the corresponding flag information corresponding to the target transformation check into the code stream.

It should be understood that the flag information corresponding to the target transformation check may include a first flag bit and a transformation index value corresponding to the target transformation check. Specifically, when the target transformation check is the first transformation check, the flag information corresponding to the target transformation check includes only the first flag bit, and the value of the first flag bit is the first value; when the target transformation check is the second transformation check Or in the third transformation check, the corresponding flag information of the target transformation check includes only the first flag bit and the transformation index value. When the target transformation check is the second transformation check, the value of the first flag bit is the second value and the transformation is performed. The index value is the first transformation index value. When the target transformation check is the third transformation check, the value of the first flag bit is the second value and the transformation index value is the second transformation index value.

Optionally, as an embodiment, before step 320, the method shown in FIG. 7 further includes: determining an intra prediction mode corresponding to the residual block as a directional prediction mode.

That is, in the method shown in FIG. 7, when the intra prediction mode corresponding to the residual block is a directional prediction mode, the candidate transformation check of the residual block is obtained according to step 320.

Optionally, as an embodiment, before step 320, the method shown in FIG. 7 further includes: determining that the width and height of the residual block are both less than or equal to M.

It should be understood that the method shown in FIG. 7 may be applicable to any intra prediction mode, that is, in the directional prediction mode or the non-directional prediction mode, the candidate transformation check of the residual block may be directly obtained according to step 320.

Alternatively, the method for determining candidate transformation check of the residual block in step 320 may only be applicable to the directional prediction mode, and for the non-directional prediction mode, a preset transformation check may be directly adopted as the target transformation check.

Optionally, the method shown in FIG. 7 further includes: when the intra prediction mode corresponding to the residual block is a DC mode, determining a transformation check (DCT-II, DTC-II) as the target transformation check .

Optionally, the method shown in FIG. 7 further includes: when the intra prediction mode corresponding to the residual block is a planar mode, determining a transformation check (DST-VII, DST-VII) as the target transformation check .

Optionally, when the intra prediction mode corresponding to the residual block is a planar mode and the width and height of the residual block are both less than or equal to N, a transformation check (DST-VII, DST-VII) is determined as The target transformation check is described.

The value of N can be 4, 8, and 16 and so on.

That is, when the intra prediction mode is a planar mode, if at least one of the width and height of the residual block is greater than N, other preset transformation check may be used as the target transformation check. For example, a DCT -II, DTC-II) is determined as the target transformation check.

FIG. 8 is a schematic flowchart of a conversion method according to an embodiment of the present application. The method shown in FIG. 8 may be executed by a coding-end device. The method shown in FIG. 8 includes steps 410 to 450. Steps 410 to 450 are described in detail below.

410: Obtain a residual block of an image block to be processed.

420. Acquire a candidate transformation check of the residual block according to preset first mapping relationship information.

The first mapping relationship information includes a mapping relationship of a first flag bit, a first transformation index value, a second transformation index value, a third transformation index value, and four transformation checklists. The first flag bit is a first value. Single-core transformation and multi-core transformation are respectively used for the second value and the second value. The first flag bit corresponds to the first transformation check when the first value is the first value, and the first transformation index value is the first flag bit when the second value is the second value. , The second transformation index value and the third transformation index value correspond to the second transformation check, the third transformation check, and the fourth transformation check, respectively, and any one of the second to fourth transformation check includes a transformation Kernel DST-VII, at least one of the second to fourth transformation checkups includes a transformation kernel DCT-VIII.

Optionally, the first value is 0, and the second value is 1.

430. Select a transformation check with the lowest rate distortion from the candidate transformation checks as the target transformation check.

440: Transform the residual block according to the target transformation check to obtain a transform coefficient of the image block to be processed.

450. Write the corresponding flag information corresponding to the target transformation check into the code stream.

In the present application, since the first mapping relationship information includes only four transformation checkups, the number of traversals can be reduced in the process of selecting a target transformation checkup, thereby reducing the time required for encoding. Further, since the transformation kernel DST-VII is included in each transformation check corresponding to the first flag bit being the second value, and DST-VII has a higher usage rate in intra prediction transformation, therefore, according to the A mapping relationship information can select a transformation check with a higher usage rate for transformation, which can improve the transformation effect.

Further, because the intra-prediction mode uses the top and right reference pixels of the current block to predict the current block, and the prediction error gradually increases from top to bottom from left to right, the base of DST-VII The function is gradually increasing. Therefore, the DST-VII transformation kernel is suitable for this gradually increasing characteristic. The transformation using DST-VII can achieve better results.

Optionally, the first value is 0, and the second value is 1.

Optionally, as an embodiment, the length of the codeword corresponding to the first transformation index value is 1, and the length of the codeword corresponding to the second transformation index value and the third transformation index value are both 2.

Optionally, the first transformation check is (DCT-II, DTC-II), the second transformation check is (DST-VII, DST-VII), and the third transformation check is (DCT-VIII, DST-VII), and the fourth transformation check is (DST-VII, DCT-VIII).

Optionally, as an embodiment, before obtaining the candidate transformation check of the residual block according to preset first mapping relationship information, the method further includes: determining an intra prediction mode corresponding to the residual block as Direction prediction mode.

Optionally, as an embodiment, before obtaining candidate transformation check of the residual block according to preset first mapping relationship information, the method further includes: determining that the width and height of the residual block are both smaller than or Is equal to M.

Optionally, the value of M is 64.

Optionally, as an embodiment, the method further includes: in a case where an intra prediction mode corresponding to the residual block is a DC mode, determining a transform check (DCT-II, DTC-II) as the target transform Check.

Optionally, as an embodiment, the above method further includes: determining a transformation check (DST-VII, DST-VII) as the target transformation when the intra prediction mode corresponding to the residual block is a planar mode. Check.

The value of N can be 4, 8, and 16 and so on.

The transformation method of the embodiment of the present application is described in detail with reference to FIG. 7 and FIG. 8 respectively above, and the process of performing transformation at the encoding end is described in detail with reference to FIG. 9.

FIG. 9 is a schematic flowchart of a conversion method according to an embodiment of the present application. The method shown in FIG. 9 can be executed by a coding end device. The method shown in FIG. 9 includes steps 2001 to 2009, and steps 2001 to 2009 are described in detail below.

The transformation method in the embodiment of the present application is described in detail below with reference to FIG. 9. The method shown in FIG. 9 includes:

2001, the beginning.

Step 2001 represents the start of performing a transform operation on the residual block. Before performing the transform operation, a prediction block of the current block may be obtained through intra prediction, and then a residual block of the current block may be obtained according to the current block and the prediction block of the current block. 2002. Determine whether the width (W) and height (H) of the residual block are both less than or equal to N.

The value of N can be set to 64 or other values.

When at least one of W and H of the residual block is greater than N, step 2003 is performed, and when both of W and H of the residual block are less than or equal to N, step 2004 is performed.

In 2003, (DCT-II, DCT-II) was determined as the target transformation check.

When the size of the residual block is large, (DCT-II, DCT-II) can be used to transform the residual block to obtain a better transform effect.

2004. Determine whether the intra prediction mode corresponding to the residual block is a directional prediction mode.

It should be understood that the intra prediction mode corresponding to the residual block may be the intra prediction mode used in the process of obtaining the residual block. Specifically, the intra prediction mode corresponding to the residual block is to obtain the predicted block of the current block. The intra prediction mode used in the process.

It should be understood that the EMT flag can be determined before step 2004. If the EMT flag is 0, then (DCT-II, DCT-II) is directly determined as the target transformation check; when the EMT flag is 1, Go to step 2004.

In 2005, it traversed a variety of transformation checks, and determined the transformation check with the lowest rate distortion as the target transformation check.

In step 2005, candidate transformation checks can be obtained according to the first mapping relationship information in the methods shown in FIG. 7 and FIG. 8, and then these candidate transformation checks are traversed in order to determine the transformation check with the lowest rate distortion as the target transformation Check.

2006. Determine whether the intra prediction mode is a DC mode.

Step 2007 is performed when the intra prediction mode is the DC mode, and step 2008 is performed when the intra prediction mode is not the DC mode (the intra prediction mode is the planar mode at this time).

2007. Determine (DCT-II, DTC-II) as the target transformation check.

In 2008, (DST-VII, DST-VII) was determined as the target transformation check.

2009. Transform the residual block according to the target transformation check to obtain the transformation coefficient.

It should be understood that, in the process shown in FIG. 9, when determining the target transformation check, the number of non-zero coefficients may be comprehensively considered to determine the target transformation check.

It should be understood that the process shown in FIG. 9 may only be an example of a process of transforming the encoding end, and the process shown in FIG. 9 may only include a part of the process of transforming the encoding end.

The inverse transform method and transform method of the embodiment of the present application are described in detail above with reference to FIGS. 1 to 9. The inverse transform device of the embodiment of the present application is described below with reference to FIG. 10. The inverse transform device shown in FIG. For each step in the inverse transform method of the embodiment of the application, the limitation on the inverse transform method of the embodiment of the present application also applies to the inverse transform device shown in FIG. 10. In order to avoid unnecessary repetition, the following describes the The inverse conversion device will be described.

FIG. 10 is a schematic block diagram of an inverse transform apparatus according to an embodiment of the present application. The inverse transforming apparatus 500 shown in FIG. 10 includes:

An obtaining module 510 is configured to parse a code stream to obtain a residual coefficient, an intra prediction mode, and a target transformation check determination parameter of a current block, where the target transformation check determination parameter includes a first flag bit or the target transformation check Determining parameters include a first flag bit and a transformation index value;

A determining module 520, the determining module 520 is configured to:

When the first flag bit has a first value, determining a target transformation check corresponding to the target transformation check determination parameter as a first transformation check;

When the first flag bit has a second value and the transformation index value is the first transformation index value, determining a target transformation check corresponding to the target transformation check determination parameter as a second transformation check, wherein the second transformation check The transformation check is a first candidate transformation check or a second transformation candidate transformation check, and the first candidate transformation check and the second transformation candidate check include a transformation kernel DST-VII, and the first candidate transformation check and the A second change candidate transformation check corresponding to an intra prediction mode corresponding to the current block;

When the first flag bit has a second value and the transformation index value is a second transformation index value, determining a target transformation check corresponding to the target transformation check determination parameter as a third transformation check;

An inverse transform module 530 is configured to perform inverse transform on the residual coefficient according to the target transform check to obtain a residual block.

That is, the inverse transforming apparatus 500 may be configured to perform steps 110 to 130 in the method shown in FIG. 4.

Optionally, the inverse transforming apparatus 500 may also be used to execute steps 210 to 230 in the method shown in FIG. 5.

Specifically, when the inverse transforming apparatus 500 performs steps 210 to 230 in the method shown in FIG. 5, the functions of the modules in the inverse transforming apparatus 500 are as follows:

An obtaining module 510 is configured to parse a bitstream to obtain a residual coefficient of a current block and a target transformation check determination parameter, wherein the target transformation check determination parameter includes a first flag bit, or the target transformation check determination parameter includes a first flag Flag bit and transform index value;

A determining module 520, the determining module 520 is configured to:

When the first flag bit has a second value and the transformation index value is the first transformation index value, determining a target transformation check corresponding to the target transformation check determination parameter as a second transformation check;

When the first flag bit has a second value and the transformation index value is a third transformation index value, determining a target transformation check corresponding to the target transformation check determination parameter as a fourth transformation check;

The second transformation check, the third transformation check, and the fourth transformation check each include a transformation kernel DST-VII, the second transformation check, the third transformation check, and the fourth transformation check. A transformation check includes the transformation kernel DCT-VIII;

FIG. 11 is a schematic block diagram of a video decoder according to an embodiment of the present application.

The video decoder 600 shown in FIG. 11 includes: an entropy decoding module 610, an inverse transform and inverse quantization module 620, a decoding end prediction module 630, a decoding reconstruction module 640, and a decoding end filtering module 650.

Optionally, the inverse transforming device 500 corresponds to the video decoder 600, and the inverse transforming device 500 may specifically be a decoder. The corresponding relationship between the inverse transforming device 500 and each module in the video decoder 600 is as follows:

The obtaining module 510 is equivalent to the entropy decoding module 610, and is configured to obtain information such as a residual coefficient and a first flag bit;

The determination module 520 and the inverse transform module 530 are equivalent to the inverse transform and inverse quantization module 620, and are configured to perform inverse transform on the residual coefficients to obtain residual blocks.

The video encoder 700 shown in FIG. 12 includes an encoding end prediction module 710, a transform quantization module 720, an entropy encoding module 730, an encoding reconstruction module 740, and an encoding end filtering module 750. The video encoder shown in FIG. 12 can execute the transformation method in the embodiment of the present application.

Those of ordinary skill in the art may realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. A professional technician can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments, and are not repeated here.

In the several embodiments provided in this application, it should be understood that the disclosed systems, devices, and methods may be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the unit is only a logical function division. In actual implementation, there may be another division manner. For example, multiple units or components may be combined or Can be integrated into another system, or some features can be ignored or not implemented. In addition, the displayed or discussed mutual coupling or direct coupling or communication connection may be indirect coupling or communication connection through some interfaces, devices or units, which may be electrical, mechanical or other forms.

The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, may be located in one place, or may be distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the objective of the solution of this embodiment.

In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each of the units may exist separately physically, or two or more units may be integrated into one unit.

When the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially a part that contributes to the existing technology or a part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including Several instructions are used to cause a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in the embodiments of the present application. The aforementioned storage media include: U disks, mobile hard disks, read-only memories (ROM), random access memories (RAM), magnetic disks or optical disks, and other media that can store program codes .

The above is only a specific implementation of this application, but the scope of protection of this application is not limited to this. Any person skilled in the art can easily think of changes or replacements within the technical scope disclosed in this application. It should be covered by the protection scope of this application. Therefore, the protection scope of this application shall be subject to the protection scope of the claims.

Claims

An inverse transform method, comprising:

Parse the bitstream to obtain the residual coefficient, the intra prediction mode, and the target transformation check determination parameter of the current block, wherein the target transformation check determination parameter includes a first flag bit, or the target transformation check determination parameter includes a first flag Bit and transform index values;

When the first flag bit has a first value, determining a target transformation check corresponding to the target transformation check determination parameter as a first transformation check;

When the first flag bit has a second value and the transformation index value is the first transformation index value, determining a target transformation check corresponding to the target transformation check determination parameter as a second transformation check, wherein the second transformation check The transformation check is a first candidate transformation check or a second transformation candidate transformation check, and the first candidate transformation check and the second transformation candidate check include a transformation kernel DST-VII, and the first candidate transformation check and the A second change candidate transformation check corresponding to an intra prediction mode corresponding to the current block;

When the first flag bit has a second value and the transformation index value is a second transformation index value, determining a target transformation check corresponding to the target transformation check determination parameter as a third transformation check;

According to the target transformation check, the residual coefficient is inversely transformed to obtain a residual block.
The method of claim 1, wherein the first candidate transformation check is (DST-VII, DCT-VIII), and the second candidate transformation check is (DCT-VIII, DST-VII);

When the mode number of the intra prediction mode corresponding to the current block is less than or equal to a preset threshold, the second transformation check is the first candidate transformation check;

When the mode number of the intra prediction mode corresponding to the current block is greater than the preset threshold, the second transformation check is the second candidate transformation check.
The method according to claim 1 or 2, wherein the third transformation check is (DST-VII, DST-VII).
The method according to any one of claims 1-3, wherein a length of a codeword corresponding to the first transformation index value and the second transformation index value is both 1.
The method according to any one of claims 1-4, wherein the first transformation check is (DCT-II, DTC-II).
The method according to any one of claims 1-5, wherein before determining the target transformation check, the method further comprises:

It is determined that the intra prediction mode corresponding to the current block is a directional prediction mode.
The method of claim 1, further comprising:

When the intra prediction mode corresponding to the current block is a DC mode, a transformation check (DCT-II, DTC-II) is determined as the target transformation check.
The method according to claim 1, further comprising:

When the intra prediction mode corresponding to the current block is a planar mode, a transformation check (DST-VII, DST-VII) is determined as the target transformation check.
An inverse transform method, comprising:

Parse the bitstream to obtain the residual coefficient of the current block and the target transformation check determination parameter, wherein the target transformation check determination parameter includes a first flag bit, or the target transformation check determination parameter includes a first flag bit and a transformation index value ;

When the first flag bit has a first value, determining a target transformation check corresponding to the target transformation check determination parameter as a first transformation check;

When the first flag bit has a second value and the transformation index value is the first transformation index value, determining a target transformation check corresponding to the target transformation check determination parameter as a second transformation check;

When the first flag bit has a second value and the transformation index value is a second transformation index value, determining a target transformation check corresponding to the target transformation check determination parameter as a third transformation check;

When the first flag bit has a second value and the transformation index value is a third transformation index value, determining a target transformation check corresponding to the target transformation check determination parameter as a fourth transformation check;

The second transformation check, the third transformation check, and the fourth transformation check each include a transformation kernel DST-VII, the second transformation check, the third transformation check, and the fourth transformation check. A transformation check includes the transformation kernel DCT-VIII;

According to the target transformation check, the residual coefficient is inversely transformed to obtain a residual block.
The method according to claim 9, wherein the length of the codeword corresponding to the first transformation index value is 1, and the length of the codeword corresponding to the second transformation index value and the third transformation index value are both 2.
The method according to claim 9 or 10, wherein the first transformation check is (DCT-II, DTC-II), and the second transformation check is (DST-VII, DST-VII), so The third transformation check is (DCT-VIII, DST-VII), and the fourth transformation check is (DST-VII, DCT-VIII).
The method according to any one of claims 9-11, wherein before determining the target transformation check, the method further comprises:

It is determined that the intra prediction mode corresponding to the current block is a directional prediction mode.
The method of claim 9, further comprising:

When the intra prediction mode corresponding to the current block is a DC mode, a transformation check (DCT-II, DTC-II) is determined as the target transformation check.
The method of claim 9, further comprising:

When the intra prediction mode corresponding to the current block is a planar mode, a transformation check (DST-VII, DST-VII) is determined as the target transformation check.
An inverse transforming device, comprising:

An obtaining module, configured to parse a bitstream to obtain a residual coefficient, an intra prediction mode, and a target transformation check determination parameter of a current block, wherein the target transformation check determination parameter includes a first flag bit or the target transformation check determination The parameters include a first flag bit and a transformation index value.

A determining module for:

When the first flag bit has a first value, determining a target transformation check corresponding to the target transformation check determination parameter as a first transformation check;

When the first flag bit has a second value and the transformation index value is the first transformation index value, determining a target transformation check corresponding to the target transformation check determination parameter as a second transformation check, wherein the second transformation check The transformation check is a first candidate transformation check or a second transformation candidate transformation check, and the first candidate transformation check and the second transformation candidate check include a transformation kernel DST-VII, and the first candidate transformation check and the A second change candidate transformation check corresponding to an intra prediction mode corresponding to the current block;

When the first flag bit has a second value and the transformation index value is a second transformation index value, determining a target transformation check corresponding to the target transformation check determination parameter as a third transformation check;

An inverse transform module is configured to perform inverse transform on the residual coefficient according to the target transform check to obtain a residual block.
The apparatus according to claim 15, wherein the first candidate transformation check is (DST-VII, DCT-VIII), and the second candidate transformation check is (DCT-VIII, DST-VII);

When the mode number of the intra prediction mode corresponding to the current block is less than or equal to a preset threshold, the second transformation check is the first candidate transformation check;

When the mode number of the intra prediction mode corresponding to the current block is greater than the preset threshold, the second transformation check is the second candidate transformation check.
The device according to claim 15 or 16, wherein the third transformation check is (DST-VII, DST-VII).
The device according to any one of claims 15 to 17, wherein a length of a codeword corresponding to the first transformation index value and the second transformation index value is both 1.
The device according to any one of claims 15 to 18, wherein the first transformation check is (DCT-II, DTC-II).
The apparatus according to any one of claims 15 to 19, wherein before determining the target transformation check, the determining module is further configured to:

It is determined that the intra prediction mode corresponding to the current block is a directional prediction mode.
The apparatus according to claim 15, wherein the determining module is further configured to:

When the intra prediction mode corresponding to the current block is a DC mode, a transformation check (DCT-II, DTC-II) is determined as the target transformation check.
The apparatus according to claim 15, wherein the determining module is further configured to:

When the intra prediction mode corresponding to the current block is a planar mode, a transformation check (DST-VII, DST-VII) is determined as the target transformation check.
An inverse transforming device, comprising:

An obtaining module, configured to parse a bitstream to obtain a residual coefficient of a current block and a target transformation check determination parameter, wherein the target transformation check determination parameter includes a first flag bit, or the target transformation check determination parameter includes a first flag Bit and transform index values;

A determining module for:

When the first flag bit has a first value, determining a target transformation check corresponding to the target transformation check determination parameter as a first transformation check;

When the first flag bit has a second value and the transformation index value is the first transformation index value, determining a target transformation check corresponding to the target transformation check determination parameter as a second transformation check;

When the first flag bit has a second value and the transformation index value is a second transformation index value, determining a target transformation check corresponding to the target transformation check determination parameter as a third transformation check;

When the first flag bit has a second value and the transformation index value is a third transformation index value, determining a target transformation check corresponding to the target transformation check determination parameter as a fourth transformation check;

The second transformation check, the third transformation check, and the fourth transformation check each include a transformation kernel DST-VII, the second transformation check, the third transformation check, and the fourth transformation check. A transformation check includes the transformation kernel DCT-VIII;

An inverse transform module is configured to perform inverse transform on the residual coefficient according to the target transform check to obtain a residual block.
The apparatus according to claim 23, wherein the length of the codeword corresponding to the first transformation index value is 1, and the length of the codeword corresponding to the second transformation index value and the third transformation index value are both 2.
The device according to claim 22 or 23, wherein the first transformation check is (DCT-II, DTC-II), and the second transformation check is (DST-VII, DST-VII), so The third transformation check is (DCT-VIII, DST-VII), and the fourth transformation check is (DST-VII, DCT-VIII).
The apparatus according to any one of claims 23 to 25, wherein before determining the target transformation check, the determining module is further configured to:

It is determined that the intra prediction mode corresponding to the current block is a directional prediction mode.
The apparatus according to claim 23, wherein the determining module is further configured to:

When the intra prediction mode corresponding to the current block is a DC mode, a transformation check (DCT-II, DTC-II) is determined as the target transformation check.
The apparatus according to claim 23, wherein the determining module is further configured to:

When the intra prediction mode corresponding to the current block is a planar mode, a transformation check (DST-VII, DST-VII) is determined as the target transformation check.