WO2018218950A1

WO2018218950A1 - Entropy decoding method, device and storage medium

Info

Publication number: WO2018218950A1
Application number: PCT/CN2017/119256
Authority: WO
Inventors: 黄勃; 吴桐庆
Original assignee: 深圳市中兴微电子技术有限公司
Priority date: 2017-05-27
Filing date: 2017-12-28
Publication date: 2018-12-06
Also published as: CN108965878B; CN108965878A

Abstract

The present invention provides an entropy decoding method and device, and a storage medium. The entropy decoding method comprises: performing binary arithmetic decoding of a binary symbol; for a predetermined syntactic element, predicting, when performing the binary arithmetic decoding, a syntactic element of the next decoding and a context model corresponding to the binary symbol when the binary symbol is 0 or 1; and according to the numerical value of the binary symbol obtained by performing the binary arithmetic decoding, selecting a corresponding context model from the predicted results. The present application can improve decoding throughput.

Description

Entropy decoding method, device and storage medium

Cross-reference to related applications

The present application is based on a Chinese patent application filed on Jan. 27, 2017, the entire disclosure of which is hereby incorporated by reference.

Technical field

The present invention relates to the field of video compression and codec, and in particular, to an entropy decoding method, apparatus, and storage medium.

Background technique

With the development of high-definition and ultra-high-definition video applications, video compression technology is facing more and more challenges. High efficiency video coding (HEVC) is a new generation video coding standard released by JCT-VC in 2013. Context-based Adaptive Binary Arithmetic Coding (CABAC) is an entropy coding method of the H.265/HEVC standard, which includes three basic steps: binarization, context model selection, and binary arithmetic coding.

On the HEVC decoder side, the entropy decoding process requires binary arithmetic decoding to perform the update operation of the relevant model, and then the next arithmetic decoding. The above two steps have sequential and data dependencies, and it is difficult to increase the throughput rate in parallel, which becomes a decoder performance bottleneck.

At present, an HEVC decoder can be implemented by using an ASIC. The existing entropy decoding technology can perform a binary symbol algorithm decoding and a context model update in one clock cycle by using a combination logic circuit. When the bit stream rate is high, if it is required to complete recovery of multiple binary symbols in one clock cycle, it is necessary to serially complete the arithmetic decoding of the binary symbol BIN0 in one clock cycle, the context model update, and the arithmetic of the binary symbol BIN1. Decoding, context model updates, can cause serious timing violations.

It can be seen that the prior art realizes binary arithmetic decoding and context model update in serial due to data dependency. The timing violation of the two-stage combined circuit is very large, and it is impossible to recover two binary symbols in one clock cycle, and high throughput cannot be realized. Rate of CABAC decoding.

Summary of the invention

In view of this, embodiments of the present invention are expected to provide an entropy decoding method, apparatus, and storage medium, which can improve decoding throughput.

The embodiments of the present application provide the following technical solutions.

An embodiment of the present invention provides an entropy decoding method, including:

Binary arithmetic decoding of binary symbols; for predetermined syntax elements, while performing binary arithmetic decoding, predicting a context model corresponding to the next decoded syntax element and binary symbol when the binary symbols are 0 and 1, respectively;

The corresponding context model is selected from the predicted results based on the values of the binary symbols obtained by binary arithmetic decoding.

In the above solution, the predetermined syntax element may include one or more of the following:

Residual coding syntax elements, motion vector difference syntax elements.

In the above solution, the binary arithmetic decoding of the binary symbols may include:

If the difference between the current interval and the offset value is less than or equal to the interval size of the minimum probity symbol (LPS), the current binary symbol is equal to the LPS; if the difference between the current interval R and the offset value Offset is greater than the interval size of the LPS, Then the current binary symbol is equal to the maximum probability symbol MPS.

In the above solution, for the two binary symbols parsed in one clock cycle, the first binary arithmetic decoding unit and the second binary arithmetic decoding unit may be used for binary arithmetic decoding, respectively, and the first context model selection unit may be adopted respectively. And the second context model selecting unit, when the predicted binary symbols are 0 and 1, respectively, the next decoded syntax element and the context model corresponding to the binary symbol.

In the above solution, the selecting the corresponding context model from the predicted result according to the value of the binary symbol obtained by the binary arithmetic decoding may include:

When the binary symbols are predicted to be 0 and 1, respectively, the next decoded syntax element and the context model corresponding to the binary symbol are respectively used as input signals of the selector; the numerical value of the binary symbol obtained according to binary arithmetic decoding is used as the The control signal of the selector; the output signal of the selector is the selected context model.

An embodiment of the present invention further provides an entropy decoding apparatus, including:

a parsing module configured to perform binary arithmetic decoding on the binary symbol; for a predetermined syntax element, while performing binary arithmetic decoding, predicting that the binary symbol is 0 and 1, respectively, the next decoded syntax element and the binary symbol corresponding to Context model

And a selection module configured to select a corresponding context model from the predicted results according to the value of the binary symbol obtained by binary arithmetic decoding.

Residual coding syntax elements, motion vector difference syntax elements.

If the difference between the current interval and the offset value is less than or equal to the interval size of the LPS, the current binary symbol is equal to the LPS; if the difference between the current interval R and the offset value Offset is greater than the interval size of the LPS, the current binary symbol is equal to the maximum probability symbol MPS .

In the above solution, the parsing module may include:

a first binary arithmetic decoding unit, a first context model selecting unit, a second binary arithmetic decoding unit, and a second context model selecting unit;

The first binary arithmetic decoding unit and the second binary arithmetic decoding unit may be respectively used for binary arithmetic decoding of two binary symbols parsed in one clock cycle;

The first context model selection unit and the second context model selection unit may be respectively used for two binary symbols parsed in one clock cycle, and when the prediction is 0 and 1, respectively, the next decoded syntax element and the binary symbol correspond to Context model.

In the above solution, the selecting module may include:

a first selector, a second selector;

The input signal of the first selector may be a context model corresponding to the next decoded syntax element and the binary symbol when the binary symbols predicted by the first context model selection unit are 0 and 1, respectively; the control signal may be The value of the binary symbol obtained by the first binary arithmetic decoding unit; the output signal may be the selected first context model, and input to the second binary arithmetic decoding unit;

The input signal of the second selector may be a context model corresponding to the next decoded syntax element and the binary symbol when the binary symbols predicted by the second context model selection unit are 0 and 1, respectively; the control signal may be The value of the binary symbol obtained by the second binary arithmetic decoding unit; the output signal may be the selected second context model.

An embodiment of the present invention further provides an entropy decoding apparatus, including: a processor and a memory;

The memory is configured to save a program for performing entropy decoding;

The processor is configured to execute the program for performing entropy decoding, and implement the following operations:

Residual coding syntax elements, motion vector difference syntax elements.

If the difference between the current interval and the offset value is less than or equal to the interval size of the minimum probability symbol LPS, the current binary symbol is equal to the LPS; if the difference between the current interval R and the offset value Offset is greater than the interval size of the LPS, the current binary symbol is equal to the maximum Probability symbol MPS.

The embodiment of the present application can reduce the timing path of the circuit, and can greatly improve the throughput rate of the HEVC standard entropy decoding circuit.

Other features and advantages of the present application will be set forth in the description which follows. The objectives and other advantages of the present invention can be realized and obtained by the structure of the invention.

DRAWINGS

The drawings are used to provide a further understanding of the technical solutions of the present application, and constitute a part of the specification, which is used together with the embodiments of the present application to explain the technical solutions of the present application, and does not constitute a limitation of the technical solutions of the present application.

1 is a schematic flowchart of an entropy decoding method according to an embodiment of the present invention;

2 is a schematic flowchart of CABAC decoding according to an embodiment of the present invention;

3 is a schematic diagram of the operation of a binary arithmetic decoding unit according to an embodiment of the present invention;

4 is a schematic diagram of calculating a binary symbol using an offset value and a minimum probability symbol according to an embodiment of the present invention;

FIG. 5 is a schematic diagram of an entropy decoding apparatus according to an embodiment of the present invention.

detailed description

In order to make the objects, technical solutions and advantages of the present application more clear, the embodiments of the present application will be described in detail below with reference to the accompanying drawings. It should be noted that, in the case of no conflict, the features in the embodiments and the embodiments in the present application may be arbitrarily combined with each other.

The steps illustrated in the flowchart of the figures may be executed in a computer system such as a set of computer executable instructions. Also, although logical sequences are shown in the flowcharts, in some cases the steps shown or described may be performed in a different order than the ones described herein.

An embodiment of the present invention provides an entropy decoding method, as shown in FIG. 1 , and relates to S110 to S120, and then separately describes:

S110: Perform binary arithmetic decoding on the binary symbol; for the predetermined syntax element, while performing binary arithmetic decoding, predict the context model corresponding to the next decoded syntax element and the binary symbol when the binary symbols are 0 and 1, respectively.

S120. Select a corresponding context model from the predicted results according to the value of the binary symbol obtained by binary arithmetic decoding.

In this embodiment, it is possible to parse two binary symbols in one clock cycle, shorten the timing path of the circuit, and improve the CABAC throughput rate of the HEVC decoder.

In this embodiment, the prediction at the same time of decoding means that the prediction and the decoding are performed synchronously, which is a parallel process; and the process of decoding and predicting a binary symbol can be started, but not limited to, at the same time.

In this embodiment, the predicted result includes a prediction result of a branch in which the binary symbol is 0, and a prediction result of a branch in which the binary symbol is 1. When the value of the binary symbol is decoded, the corresponding prediction result can be directly selected according to the value. .

In an embodiment, the predetermined syntax element may include, but is not limited to, one or more of the following:

Residual coding syntax elements, motion vector difference syntax elements.

In an embodiment, the binary arithmetic decoding of the binary symbols may include:

If the difference between the current interval R and the offset value Offset is less than or equal to the interval size of the LPS, the current binary symbol is equal to the LPS; if the difference between the current interval R and the offset value Offset is greater than the interval size of the LPS, the current binary symbol is equal to the MPS.

In an embodiment, for two binary symbols parsed in one clock cycle, binary arithmetic decoding may be performed by using the first binary arithmetic decoding unit and the second binary arithmetic decoding unit, respectively, and the first context model may be adopted respectively. The selection unit and the second context model selection unit, when the predicted binary symbols are 0 and 1, respectively, the next decoded syntax element and the context model corresponding to the binary symbol.

In an embodiment, the selecting the corresponding context model from the predicted results according to the binary symbol obtained by the binary arithmetic decoding may include:

Next, CABAC decoding will be described as an example.

This example applies two-part syntax elements in the H.265/HEVC standard, namely: residual coding syntax elements and motion vector difference syntax elements. In the conventional decoding process, the context model is determined in advance by means of prediction; in the binary arithmetic decoding process In the method of comparing the offset value with the interval size of the Least Probable Symbol (LPS), the two binary symbols are recovered in one clock cycle, and the timing path of the circuit is shortened, thereby improving the HEC decoder CABAC throughput. rate.

In actual implementation, the entropy decoding function in the HEVC standard can be implemented in a manner that software and hardware cooperate. The software completes the resolution of the video layer parameter set (VPS, Video Parameter Set), sequence parameter set (SPS, Sequence Paramater Set), and image parameter set (PPS, Picture Paramater Set). The hardware circuit completes the parsing of the slice header and the following syntax elements. The entropy decoding hardware may include a state machine control unit, two sets of binary arithmetic decoding units, and two sets of context model selection units.

This example works as follows, as shown in FIG. 2, including the following steps 201-207:

Step 201: Read the bit stream, parse the VPS, SPS, and PPS parameter sets, and perform hardware configuration.

In actual implementation, the software completes the analysis of the VPS, SPS, and PPS parameter sets, configures the hardware through the Advanced Peripheral Bus (APB), and writes the decoding parameters to the hardware registers.

Step 202: Perform binary arithmetic decoder initialization, and initialize the context probability model.

When parsing a new slice, a new tile or a new coding tree unit (CTU), binary arithmetic decoder initialization, context probability model initialization is performed as described in the HEVC standard. .

For binary arithmetic decoders, initialize as follows:

(1) The current interval (currange) is set to decimal 510;

(2) The read stream starts with 9 bits as the current interval offset value (Offset).

For the probability model, each probability model is initialized as described in the HEVC standard, including:

pStateIdx and valMps.

Step 203: Transfer the entropy decoding process of the corresponding syntax element, and select a probability model.

The state machine control unit selects the probability model according to the HEVC syntax element, transfers to the entropy decoding process of the corresponding syntax element, and selects the probability model. For example, when parsing a CTU, the first syntax element to be decoded is sample adaptive offset.

In this example, branch prediction is performed on two parts of the residual and motion vector difference, and the throughput is improved; that is, for the residual coefficient syntax element and the motion vector difference syntax element, The operations of steps 204 to 207 are employed. For other syntax elements, a binary symbol is parsed in one clock cycle.

Step 204: Parse the first binary symbol BIN0, and start the context model selection unit CM1 to perform branch prediction.

The startup binary arithmetic decoding unit BAD1 parses the first binary symbol BIN0. At the same time as starting BAD1, the startup context model selection unit CM1 predicts the context model corresponding to the next decoded syntax element and binary symbol when BIN0 is 0 and 1, respectively. In the binary arithmetic decoding process, the method of comparing the offset value with the interval size of the minimum probability symbol (LPS) is adopted. If the new interval obtained after the interval is less than 256, the renormalization process is required.

In practical applications, when entering the residual syntax element (the case when entering the motion vector syntax element can be analogized, and will not be described here), the binary arithmetic decoding unit is shown in FIG. The context model CM_0 of the first binary symbol BIN0 to be decoded is selected at the clock period clk0. If the decoding mode of BIN0 is conventional decoding, BAD1 is started to parse BIN0 in the next clock cycle clk1. When BIN0 is parsed, when the CM1 prediction BIN0 is 0 and 1, respectively, the context model corresponding to the next decoded syntax element and binary symbol is denoted as CM_1_if=0 and CM_1_if=1, respectively.

In the BAD1 parsing process, as shown in FIG. 4, the manner in which the offset value is compared with the interval size of the LPS is adopted. First calculate the index value of the range (ρ):

ρ=(R>>6)&3;

Where R is the size of the current interval.

The size of the subinterval corresponding to the LPS (ie, the interval size of the LPS) R _{Lps is} obtained by looking up the table:

R _Lps =rangeTablps[σ][ρ]

Where σ is the probability state index, different σ values correspond to different LPS probabilities, and different probability state conversion results: if the current binary symbol to be coded is MPS, the probability state will be converted to transIdxMps[σ], if current If the binary symbol to be coded is LPS, the probability state will be converted to transIdxLps[σ]; for example, when σ=0, the probability of LPS is 0.5, transIdxLps[σ] is 0, and transIdxMps[σ] is 1.

Get the size of the subinterval of the current maximum probability symbol (MPS, Most Probable Symbol) R _Mps :

R _Mps =R-R _Lps ;

The decision is made using the difference between the current interval R and the offset value Offset. If the difference R-Offset is less than or equal to R _Lps , the current binary symbol Bin is equal to LPS; if the difference R-Offset is greater than R _Lps , the current binary symbol Bin is equal to MPS.

If the current binary symbol Bin is equal to MPS, the encoding interval R _{Mps of} the MPS is taken as the encoding interval R of the next binary symbol (ie: R=R _Mps ), the lower limit L of the interval is unchanged; σ=transIdxMps[σ].

If the current binary symbol Bin is equal to LPS, the interval R _{Lps of} the LPS is _taken as the encoding interval R of the next binary symbol (ie: R = R _Lps ), and the lower limit of the interval is increased _by the length of R _Mps . Offset is updated to: R _Lps - (R-Offset).

It is judged whether σ is equal to 0; if σ=0, that is, the probability of LPS is the same as the probability of MPS, and since LPS is present again, MPS and LPS need to be interchanged, that is, MPS=1-MPS; then σ=transIdxLps[σ]. If σ is not equal to 0, then σ=transIdxLps[σ].

The context model is then updated based on the current Bin value.

If the new interval obtained after the interval is less than 256, the renormalization process is required.

Step 205: The value of the parsed BIN0 selects a corresponding context model.

The value of BIN0 parsed according to the binary arithmetic decoding unit BAD1 is 0 or 1, and the corresponding context model CM_1 is selected from the prediction results of CM1.

As shown in Figure 3, the value of BIN0 can be used as the control signal of the first selector. The two inputs of the first selector are CM_1_if=0 and CM_1_if=1, respectively, and the output is connected to the input of BAD2; if the value of BIN0 is 0, select CM_1_if=0 as the CM_1 output; if the value of BIN0 is 1, select CM_1_if=1 as the CM_1 output.

Step 206: Parse the next binary symbol BIN1, and start the context model selection unit CM2 to perform branch prediction.

The binary arithmetic decoding unit BAD2 is started to parse the next binary symbol BIN1. When the BAD2 is started, the context model selection unit CM2 is also started to predict that the BIN1 is 0 and 1, respectively, and the context model corresponding to the next decoded syntax element and the binary symbol is denoted as CM_2_if=0 and CM_2_if=1, respectively. If the new interval obtained after the interval is less than 256, the renormalization process is required.

Reverse binarization of BIN0 and BIN1 to obtain the value of the corresponding syntax element. The number of syntax elements is large and requires a large amount of storage space. In this example, the syntax elements are stored in a double rate synchronous dynamic random access memory (DDR) through an Advanced EXtensible Interface (AXI) bus.

The value of BIN1 parsed according to BAD2 is 0 or 1, and the corresponding context model CM_2 is selected from the results of CM2. As shown in FIG. 3, the value of BIN1 can be used as the control signal of the second selector, and the two inputs of the second selector are CM_2_if=0 and CM_2_if=1, respectively, and the selected context model is output; if BIN1 When the value is 0, CM_2_if=0 is selected as CM_2; if the value of BIN1 is 1, CM_2_if=1 is selected as CM_2.

Step 207: The next clock cycle, the operations of steps 204 to 206 are repeated until the residual or motion vector difference syntax element entropy decoding is completed.

There are many H.265/HEVC syntax elements. If branch prediction is performed for all syntax elements, the complexity of the entropy decoding part will be greatly increased. For the case where the residual and motion vectors in the bit stream are relatively large, in this example, only the residual coding syntax element and the motion vector difference syntax element are predicted, which not only improves the throughput of CABAC, but also reduces the hardware design. Difficulty. This is one of the improvement points of this application. For other syntax elements, it can be restored in the traditional way.

The parsing of two binary symbols BIN0 and BIN1 is completed in one clock cycle. The usual step is to update the context model after BAD1 and BAD2 decoding, which can cause serious timing violations when implemented using combinatorial logic. In the present application, the branch prediction method is adopted, and when the BAD1 is started to parse the BIN0, the first context model selection unit CM1 is started to predict the context model corresponding to the next decoded syntax element and the binary symbol when the BIN0 is 0 and 1, respectively. The same operation is performed for BAD2 and CM2. As shown in Fig. 3, in the present application, while the binary arithmetic decoding unit parses the binary symbol, the context model selection unit that performs branch prediction performs the determination of the context model in advance, which is one of the improvement points of the present application.

When performing binary arithmetic decoding, the present application does not adopt a method in which the offset value is compared with the maximum probability symbol (MPS) in the HEVC standard, but a method in which the offset value is compared with the interval size of the minimum probability symbol (LPS). The timing path of the hardware circuit is reduced, so that the resolution of two binary symbols can be realized in one clock cycle. As shown in Figure 4.

An embodiment of the present invention further provides an entropy decoding device, as shown in FIG. 5, including:

The parsing module 21 is configured to perform binary arithmetic decoding on the binary symbols; for the predetermined syntax element, when the binary arithmetic decoding is performed, when the binary symbols are predicted to be 0 and 1, respectively, the next decoded syntax element and the binary symbol correspond to Context model

The selecting module 22 is configured to select a corresponding context model from the predicted results according to the value of the binary symbol obtained by binary arithmetic decoding.

In an embodiment, the predetermined syntax element may include one or more of the following:

Residual coding syntax elements, motion vector difference syntax elements.

In an embodiment, the parsing module may include:

The first binary arithmetic decoding unit and the second binary arithmetic decoding unit are respectively configured to perform binary arithmetic decoding on two binary symbols parsed in one clock cycle;

The first context model selection unit and the second context model selection unit are respectively configured to use two binary symbols parsed in one clock cycle, and when the prediction is 0 and 1, respectively, the next decoded syntax element and the binary symbol correspond to Context model.

In an embodiment, the selecting module may include:

a first selector, a second selector;

The input signal of the first selector is a context model corresponding to a next decoded syntax element and a binary symbol when the binary symbols predicted by the first context model selection unit are 0 and 1, respectively; the control signal is the a binary arithmetic decoding unit obtains a value of a binary symbol; the output signal is a selected first context model, and is input to the second binary arithmetic decoding unit;

The input signal of the second selector is a context model corresponding to the next decoded syntax element and the binary symbol when the binary symbols predicted by the second context model selection unit are 0 and 1, respectively; the control signal is the The value of the binary symbol obtained by the binary arithmetic decoding unit; the output signal is the selected second context model.

The memory is configured to save a program for performing entropy decoding;

Residual coding syntax elements, motion vector difference syntax elements.

It should be noted that the above description of the entropy decoding device is similar to the above description of the method, and the beneficial effects of the same method are described without further description. For technical details not disclosed in the embodiment of the entropy decoding device of the present invention, please refer to the description of the method embodiment of the present invention.

Those skilled in the art will appreciate that embodiments of the present invention can be provided as a method, system, or computer program product. Accordingly, the present invention can take the form of a hardware embodiment, a software embodiment, or a combination of software and hardware. Moreover, the invention can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage and optical storage, etc.) including computer usable program code.

The present invention has been described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (system), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or FIG. These computer program instructions can be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing device to produce a machine for the execution of instructions for execution by a processor of a computer or other programmable data processing device. Means for implementing the functions specified in one or more of the flow or in a block or blocks of the flow chart.

The computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing device to operate in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture comprising the instruction device. The apparatus implements the functions specified in one or more blocks of a flow or a flow and/or block diagram of the flowchart.

These computer program instructions can also be loaded onto a computer or other programmable data processing device such that a series of operational steps are performed on a computer or other programmable device to produce computer-implemented processing for execution on a computer or other programmable device. The instructions provide steps for implementing the functions specified in one or more of the flow or in a block or blocks of a flow diagram.

The above is only the preferred embodiment of the present invention and is not intended to limit the scope of the present invention.

Industrial applicability

In the embodiment of the present invention, binary arithmetic decoding is performed on binary symbols; for predetermined syntax elements, when binary arithmetic decoding is performed, when the binary symbols are predicted to be 0 and 1, respectively, the next decoded syntax element and the binary symbol correspond to a context model; selecting a corresponding context model from the predicted results based on the values of the binary symbols obtained by binary arithmetic decoding. In this way, the timing path of the circuit can be reduced, and the throughput of the HEVC standard entropy decoding circuit can be greatly improved.

Claims

An entropy decoding method, comprising:

Binary arithmetic decoding of binary symbols; for predetermined syntax elements, while performing binary arithmetic decoding, predicting a context model corresponding to the next decoded syntax element and binary symbol when the binary symbols are 0 and 1, respectively;

The corresponding context model is selected from the predicted results based on the values of the binary symbols obtained by binary arithmetic decoding.
The entropy decoding method of claim 1, wherein the predetermined syntax element comprises one or more of the following:

Residual coding syntax elements, motion vector difference syntax elements.
The entropy decoding method according to claim 1, wherein said binary arithmetic decoding of the binary symbols comprises:

If the difference between the current interval and the offset value is less than or equal to the interval size of the minimum probability symbol LPS, the current binary symbol is equal to the LPS; if the difference between the current interval R and the offset value Offset is greater than the interval size of the LPS, the current binary symbol is equal to the maximum Probability symbol MPS.
The entropy decoding method of claim 1 wherein:

For the two binary symbols parsed in one clock cycle, the first binary arithmetic decoding unit and the second binary arithmetic decoding unit are respectively used for binary arithmetic decoding, and the first context model selection unit and the second context model are respectively selected. The unit, when the predicted binary symbols are 0 and 1, respectively, the next decoded syntax element and the context model corresponding to the binary symbol.
The entropy decoding method according to claim 1, wherein the selecting the corresponding context model from the predicted results according to the value of the binary symbol obtained by the binary arithmetic decoding comprises:

When the binary symbols are predicted to be 0 and 1, respectively, the next decoded syntax element and the context model corresponding to the binary symbol are respectively used as input signals of the selector; the numerical value of the binary symbol obtained according to binary arithmetic decoding is used as the The control signal of the selector; the output signal of the selector is the selected context model.
An entropy decoding device includes:

a parsing module configured to perform binary arithmetic decoding on the binary symbol; for a predetermined syntax element, while performing binary arithmetic decoding, predicting that the binary symbol is 0 and 1, respectively, the next decoded syntax element and the binary symbol corresponding to Context model

And a selection module configured to select a corresponding context model from the predicted results according to the value of the binary symbol obtained by binary arithmetic decoding.
The entropy decoding apparatus according to claim 6, wherein said predetermined syntax element comprises one or more of the following:

Residual coding syntax elements, motion vector difference syntax elements.
The entropy decoding apparatus according to claim 6, wherein said binary arithmetic decoding of the binary symbols comprises:

If the difference between the current interval and the offset value is less than or equal to the interval size of the minimum probability symbol LPS, the current binary symbol is equal to the LPS; if the difference between the current interval R and the offset value Offset is greater than the interval size of the LPS, the current binary symbol is equal to the maximum Probability symbol MPS.
The entropy decoding apparatus of claim 6, wherein the parsing module comprises:

a first binary arithmetic decoding unit, a first context model selecting unit, a second binary arithmetic decoding unit, and a second context model selecting unit;

The first binary arithmetic decoding unit and the second binary arithmetic decoding unit are respectively configured to perform binary arithmetic decoding on two binary symbols parsed in one clock cycle;

The first context model selection unit and the second context model selection unit are respectively configured to respectively perform two binary symbols parsed in one clock cycle, and when the predictions are 0 and 1, respectively, the next decoded syntax element and the binary symbol correspond to Context model.
The entropy decoding apparatus according to claim 9, wherein said selection module comprises:

a first selector, a second selector;

The input signal of the first selector is a context model corresponding to a next decoded syntax element and a binary symbol when the binary symbols predicted by the first context model selection unit are 0 and 1, respectively; the control signal is the a binary arithmetic decoding unit obtains a value of a binary symbol; the output signal is a selected first context model, and is input to the second binary arithmetic decoding unit;

The input signal of the second selector is a context model corresponding to the next decoded syntax element and the binary symbol when the binary symbols predicted by the second context model selection unit are 0 and 1, respectively; the control signal is the The value of the binary symbol obtained by the binary arithmetic decoding unit; the output signal is the selected second context model.
An entropy decoding apparatus includes: a processor and a memory; wherein:

The memory is configured to save a program for performing entropy decoding;

The processor is configured to execute the program for performing entropy decoding, and implement the following operations:

Binary arithmetic decoding of binary symbols; for predetermined syntax elements, while performing binary arithmetic decoding, predicting a context model corresponding to the next decoded syntax element and binary symbol when the binary symbols are 0 and 1, respectively;

The corresponding context model is selected from the predicted results based on the values of the binary symbols obtained by binary arithmetic decoding.
The entropy decoding apparatus according to claim 11, wherein said predetermined syntax element comprises one or more of the following:

Residual coding syntax elements, motion vector difference syntax elements.
The entropy decoding apparatus according to claim 11, wherein said binary arithmetic decoding of the binary symbols comprises:

If the difference between the current interval and the offset value is less than or equal to the interval size of the minimum probability symbol LPS, the current binary symbol is equal to the LPS; if the difference between the current interval R and the offset value Offset is greater than the interval size of the LPS, the current binary symbol is equal to the maximum Probability symbol MPS.
A computer storage medium having stored therein computer executable instructions for performing the entropy decoding method of any one of claims 1 to 5.