CN113453002A - Quantization and entropy coding method and apparatus - Google Patents

Quantization and entropy coding method and apparatus Download PDF

Info

Publication number
CN113453002A
CN113453002A CN202010519720.4A CN202010519720A CN113453002A CN 113453002 A CN113453002 A CN 113453002A CN 202010519720 A CN202010519720 A CN 202010519720A CN 113453002 A CN113453002 A CN 113453002A
Authority
CN
China
Prior art keywords
dual
coefficient
reconstruction level
quantizer
state
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202010519720.4A
Other languages
Chinese (zh)
Other versions
CN113453002B (en
Inventor
虞露
章致好
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Zhejiang University ZJU
Original Assignee
Zhejiang University ZJU
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Zhejiang University ZJU filed Critical Zhejiang University ZJU
Publication of CN113453002A publication Critical patent/CN113453002A/en
Application granted granted Critical
Publication of CN113453002B publication Critical patent/CN113453002B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
    • H04N19/124—Quantisation
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
    • H04N19/129—Scanning of coding units, e.g. zig-zag scan of transform coefficients or flexible macroblock ordering [FMO]
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
    • H04N19/17—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object
    • H04N19/176—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object the region being a block, e.g. a macroblock
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/60—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/70—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals characterised by syntax aspects related to video coding, e.g. related to compression standards
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/90—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using coding techniques not provided for in groups H04N19/10-H04N19/85, e.g. fractals
    • H04N19/91—Entropy coding, e.g. variable length coding [VLC] or arithmetic coding

Abstract

The invention discloses a quantization and entropy coding method and device. The method is used in the multimedia field, for the image compression code stream, the values of a plurality of syntax elements in the code stream are analyzed to obtain the reconstruction level index value of the current coefficient, wherein the probability model used for analyzing at least one syntax element is selected according to the current state of a quantizer. And mapping the reconstructed level index value obtained by analyzing the code stream into a coefficient value by adopting an inverse quantization method corresponding to the state according to the current state of the quantizer. And updating the state of the quantizer according to the current state of the quantizer and the grouping of the selected reconstruction level value. The method provided by the invention can effectively improve the coding quality of the image under the condition of not improving the compression code rate of the image. The invention provides a method for quantization and entropy coding and a corresponding device.

Description

Quantization and entropy coding method and apparatus
Technical Field
The invention belongs to the field of image compression, and particularly relates to a method and a device for quantizing and entropy coding an image transformation block.
Background
Because the steps of transformation, quantization, entropy coding and the like are needed in the image compression process, and the main link of distortion generation in the image compression is quantization, the quantization is a key step in lossy compression, so that the better quantizer design can perform finer quantization on the image, thereby reducing information loss and effectively improving the compression performance. The quantized coefficients are subjected to binarization expression by the syntax elements, and a proper probability model is selected for the corresponding syntax elements, so that the data compression ratio can be improved.
The grid network quantization method can quantize an image more finely without increasing the compression code rate, is the application of a grid coding modulation technology in source coding, and has the core of signal expansion and set division, a plurality of groups of quantizers are used, and a better quantizer is selected for the current transform block coefficient in a self-adaptive manner through the conversion relation among the quantizers, so that the mean square error is reduced. The essence of this method is to make the distances between the reconstructed values more dense in the overall sense and the Euclidean distances of the reconstructed values from the original values smaller in the R-D (rate-distortion) sense.
Marcellin M.W. firstly proposes the idea of introducing a grid coding modulation technology into the information source coding, and uses a grid network quantization method for a memoryless information source and a Gaussian-Markov information source. Joshi proposes that image subbands are coded by using a grid network quantization method, quantization values are divided into four sets of D0, D1, D2 and D3, two codebooks a0 ═ D0 ═ D2 and a1 ═ D1 ═ D3 are established, the codebooks are selected according to network states, and arithmetic coding is used for quantization results. Aksu designs two-step grid network quantization on the basis, and performs finer quantization on a residual error, wherein the first step is to perform grid network quantization on an image prediction residual error, and the second step is to perform grid network quantization on the difference between a residual error reconstruction value and an original value, so that the image quantization performance is improved in a asymptotic mode.
Heiko Schwarz proposes an associated quantization technology based on a grid network quantizer, and firstly proposes the concept of using a double quantizer, wherein the structure of the double quantizer is shown in FIG. 1, two groups of quantizers Q0 and Q1 divide quantized values into four sets, V0, V1, V2 and V3, each coefficient has a separate state value, which quantizer is used during quantization is determined according to the state of the current coefficient, the state value of the current coefficient and the quantized value set to which the coefficient belongs jointly determine the state value of the next coefficient in a scanning order, and a specific conversion relationship is shown in FIG. 2. And, a new syntax element is designed for the quantization value, and a syntax element for representing the quantization value set to which the coefficient belongs is added. For the syntax element significant flag indicating whether the coefficient is non-zero, the model selection for this syntax element is divided into two groups using the Q0 or Q1 quantizer with the coefficient. However, for some transform blocks, the energy distribution of the coefficients is concentrated, resulting in fewer non-zero coefficients within the transform block, in which case the use of grid network quantization tends to reduce compression performance.
Disclosure of Invention
In order to solve the above technical problem, the present invention provides a method and an apparatus for dual quantization for a specific transform block. The purpose of the invention is realized by the following technical scheme: aiming at the range information of the nonzero coefficient of the change block, for the change block meeting the condition, two paired dual quantizers are used, one quantizer is adaptively selected to quantize the current coefficient according to the change condition of the size of the change coefficient on the scanning sequence of the change block, a more optimal model is selected for the corresponding syntax element, and the coding quality of the image is improved under the condition that the coding rate is not improved. Specifically, the method comprises the following steps:
a first aspect of the present invention provides a coefficient inverse quantization and entropy decoding method, including:
s1, analyzing the values of a plurality of syntax elements of the coefficients in the code stream, wherein for at least one syntax element S, a probability model is selected from a plurality of groups of probability models according to the current state of the dual inverse quantizer, and the selected probability model is used for analyzing;
s2, obtaining a reconstruction level index value of the current coefficient to be decoded based on the value combination of the plurality of syntax elements;
s3, according to the current state of the dual inverse quantizer, mapping the reconstruction level index value to a reconstruction level value by adopting a corresponding inverse quantization method, and outputting the reconstruction level value as the value of the coefficient to be decoded;
and S4, updating the state of the dual inverse quantizer according to the current state of the dual inverse quantizer and the reconstruction level index group to which the reconstruction level index value belongs.
Further, the syntax element S is an identifier indicating whether the coefficient to be decoded is non-zero;
there are four different states of the dual inverse quantizer;
the selecting and analyzing the probability model specifically comprises selecting a group of probability models from the four groups of probability models according to the current state of the dual inverse quantizer, and analyzing the syntax element S by using one probability model in the group of probability models.
Furthermore, in the code stream, all syntax elements of a previous coefficient of a current coefficient to be decoded are before the syntax element S of the current coefficient to be decoded.
Further, the method further comprises:
s0, before analyzing the values of multiple syntax elements of the coefficient in the code stream, judging whether the current transformation block meets the condition according to the range information of the nonzero coefficient of the current transformation block, if so, executing the method of the steps S1-S4.
A second aspect of the present invention provides a coefficient quantization and entropy coding method, including:
according to the current state of the dual quantizer, mapping the current coefficient to be coded into a reconstruction level index value by adopting a corresponding quantization method;
describing the reconstruction level index value using values of a plurality of syntax elements;
entropy coding is carried out on the plurality of syntax elements and code streams are written in; selecting a probability model from a plurality of sets of probability models according to the current state of the dual quantizer for the value of at least one syntax element S in the plurality of syntax elements, and performing entropy coding by using the selected probability model;
and updating the state of the dual quantizer according to the current state of the dual quantizer and the reconstruction level index group to which the reconstruction level index value belongs.
Further, the syntax element S is an identifier indicating whether the coefficient to be coded is non-zero;
there are four different states of the dual quantizer;
the selecting the probability model and performing entropy coding specifically includes selecting a group of probability models from four groups of probability models according to the current state of the dual quantizer, and coding the syntax element S by using one probability model in the group of probability models.
A third aspect of the present invention provides a coefficient inverse quantization and entropy decoding apparatus, including:
a coefficient entropy decoding module: the input of the module is a code stream, the output of the module is a plurality of syntactic elements, the module analyzes the values of the syntactic elements in the code stream, wherein for at least one syntactic element S, a probability model is selected from a plurality of groups of probability models according to the current state of an inverse quantizer, and the selected probability model is used for analyzing the values of the syntactic elements;
the reconstruction level index value decoding module: the module combines the values of the plurality of syntax elements to obtain the reconstruction level index value of the current coefficient to be decoded;
a dual inverse quantization module: the module adopts a corresponding inverse quantization method to map the reconstruction level index value into the reconstruction level value according to the state of the dual inverse quantizer, and outputs the reconstruction level value as the value of the coefficient to be decoded;
a state updating module: the input of the dual inverse quantizer is the current state of the dual inverse quantizer and the reconstruction level group corresponding to the reconstruction level value, the output is the updated dual inverse quantizer state, and the module updates the state of the inverse quantizer according to the current state of the inverse quantizer and the reconstruction level index group to which the reconstruction level index value belongs.
Further, the syntax element S is an identifier indicating whether the coefficient to be decoded is non-zero;
the inverse quantizer exists in four different states;
the selecting and analyzing the probability model specifically comprises: and selecting one group of probability models from the four groups of probability models according to the current state of the dual inverse quantizer, and analyzing the syntax element S by using one probability model in the group of probability models.
A fourth aspect of the present invention provides an apparatus for coefficient quantization and entropy coding, comprising:
a dual quantization module: the input of the module is a coefficient to be coded and the current state of a dual quantizer, and the output is a reconstructed level index value, and the module adopts a corresponding quantization method to map the current coefficient to be coded into the reconstructed level index value;
the reconstruction level index value coding module: the module uses the values of the plurality of syntax elements to describe the reconstruction level index value;
a coefficient entropy coding module: the input of the module is a plurality of syntax elements, the output of the module is a code stream, and the module carries out entropy coding on the syntax elements and writes the syntax elements into the code stream; selecting a probability model from a plurality of sets of probability models according to the current state of the dual quantizer for the value of at least one syntax element S in the plurality of syntax elements, and performing entropy coding by using the selected probability model;
a state updating module: the module updates the state of the dual quantizer according to the current state of the dual quantizer and the reconstruction level index group to which the reconstruction level index value belongs.
Further, the syntax element S is an identifier indicating whether the coefficient to be coded is non-zero;
there are four different states of the quantizer;
the selecting and encoding of the probability model specifically includes: a set of probability models is selected from the four sets of probability models, based on the current state of the quantizer, and one of the set of probability models is used to encode the syntax element S.
Compared with the prior art, the invention has the advantages that:
1) when the values of a plurality of syntax elements in the code stream are analyzed, for at least one syntax element, a probability model is selected from a plurality of groups of probability models according to the current state of the dual inverse quantizer and is analyzed, so that the entropy coding performance is improved, and the coding efficiency is improved.
2) On one hand, the conversion block which is not suitable for using the dual quantization technology can be skipped, thereby reducing the encoding complexity and shortening the encoding time; on the other hand, performance degradation caused by transform blocks that are not suitable for using this technique can be avoided.
Drawings
Other features and advantages of the present invention will become apparent from the following description of the preferred embodiment, taken in conjunction with the accompanying drawings, which illustrate, by way of example, the principles of the invention.
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the invention and together with the description serve to explain the invention without limiting the invention. In the drawings:
fig. 1 is a structure of a dual quantizer.
Fig. 2 is a diagram of quantizer state transition relationships.
FIG. 3 is a schematic view of the apparatus of example 8.
FIG. 4 is a schematic view of the apparatus of example 10.
FIG. 5 is a schematic view of the apparatus of example 9.
FIG. 6 is a schematic view of the apparatus of example 11.
Detailed Description
For a further understanding of the invention, reference will now be made to the following examples describing preferred embodiments of the invention, but it is to be understood that the description is intended to illustrate further features and advantages of the invention and is not intended to limit the scope of the claims.
Example 1
A method for inverse quantization and entropy decoding, specifically comprising:
(1) analyzing values of a plurality of syntax elements in the code stream, wherein for at least one syntax element, a probability model is selected from a plurality of groups of probability models according to the current state of the dual inverse quantizer and is analyzed: for example, only the significant flag indicating whether the coefficient is non-zero is parsed by using the probability model selected from the plurality of sets of probability models according to the current state of the dual dequantizer.
(2) Combining values of a plurality of syntax elements to obtain a reconstruction level index value of the current coefficient to be decoded: specific implementations are various, for example, a combination of a flag indicating whether the coefficient is non-zero, a flag indicating whether the coefficient is greater than 1, a flag indicating whether the coefficient is greater than 2, and the remainder of the coefficient minus 2 is used, for example, a combination of a flag indicating whether the coefficient is non-zero, a flag indicating whether the coefficient is an even number, a flag indicating whether the coefficient is greater than 2, and the remainder of the coefficient minus 2 is used.
(3) The inverse quantization method comprises mapping the reconstruction level index value to a reconstruction level value by corresponding inverse quantization method according to the current state of the dual inverse quantizer, outputting the reconstruction level value as the value of the coefficient to be decoded, obtaining the reconstruction level value according to the current state of the dual inverse quantizer by using the dual inverse quantizer shown in FIG. 1, specifically performing inverse quantization operation as follows,
if the state is 0 or 1, using the quantization formula, coeff ═ x · 2 Δ,
if the state is 2 or 3, using the quantization formula,
Figure BDA0002531558580000051
where Δ is the quantization step. x is the reconstruction level index value and coeff is the reconstruction level value.
And mapping the reconstruction level index value into a reconstruction level value by adopting a corresponding inverse quantization method according to the current state of the dual inverse quantizer, wherein the reconstruction level value belongs to one of two reconstruction level value groups, the reconstruction level values in the two groups except the 0 value are distributed in an interlaced way, and the reconstruction level value is output as the value of the coefficient to be decoded, namely the structure of the dual quantizer.
(4) And updating the state of the dual inverse quantizer according to the current state of the dual inverse quantizer and the reconstruction level group corresponding to the reconstruction level value. In the process of inverse quantization, after the current coefficient is inversely quantized, the state of the dual inverse quantizer is updated according to the state of the dual inverse quantizer and the parity of the coefficient, and the state determines the selection of the quantizer of the next coefficient.
As shown in fig. 2, the dual inverse quantizer state transition relationship is: when the dual dequantizer is in the state S00, if the selected reconstruction level value belongs to V00, the dual dequantizer is updated to S00, and if the selected reconstruction level value belongs to V01, the dual dequantizer is updated to S10; when the dual dequantizer is in the state S01, if the selected reconstruction level value belongs to V00, the dual dequantizer is updated to S10, and if the selected reconstruction level value belongs to V01, the dual dequantizer is updated to S00; when the dual dequantizer is in the state S10, if the selected reconstruction level value belongs to V10, the dual dequantizer is updated to S01, and if the selected reconstruction level value belongs to V11, the quantizer is updated to S11; when the dual inverse quantizer is in state S11, the dual inverse quantizer state is updated to S11 if the selected reconstruction level value belongs to V10, and the dual inverse quantizer state is updated to S01 if the selected reconstruction level value belongs to V11.
Example 2
A method for inverse quantization and entropy decoding, specifically comprising:
(1) and judging whether the current transformation block meets the condition according to the range information of the nonzero coefficient of the current transformation block, and if so, using a dual quantization technology. For how to judge whether the current transformation block meets the condition, different implementation methods can be provided, for example, according to x and y of the first non-zero coefficient in the scanning sequence in the transformation block, when x and y meet a certain size respectively, the transformation block meets the condition, and then a dual quantization technology is used; for example, according to the coordinates x and y of the lower right corner of the minimum rectangle enclosed by all non-zero coefficients in the transform block, when x and y satisfy a certain size respectively, the transform block meets the condition, and then the dual quantization technique is used.
(2) Analyzing values of a plurality of syntax elements in the code stream, wherein for at least one syntax element, a probability model selected from a plurality of sets of probability models according to the current state of the dual inverse quantizer is used and analyzed: for example, only the significant flag indicating whether the coefficient is non-zero is parsed by using the probability model selected from the plurality of sets of probability models according to the current state of the dual dequantizer.
(3) Combining values of a plurality of syntax elements to obtain a reconstruction level index value of the current coefficient to be decoded: specific implementations are various, for example, a combination of a flag indicating whether the coefficient is non-zero, a flag indicating whether the coefficient is greater than 1, a flag indicating whether the coefficient is greater than 2, and the remainder of the coefficient minus 2 is used, for example, a combination of a flag indicating whether the coefficient is non-zero, a flag indicating whether the coefficient is an even number, a flag indicating whether the coefficient is greater than 2, and the remainder of the coefficient minus 2 is used.
(4) The inverse quantization method comprises mapping the reconstruction level index value to a reconstruction level value by corresponding inverse quantization method according to the current state of the dual inverse quantizer, outputting the reconstruction level value as the value of the coefficient to be decoded, obtaining the reconstruction level value according to the current state of the dual inverse quantizer by using the dual inverse quantizer shown in FIG. 1, specifically performing inverse quantization operation as follows,
if the state is 0 or 1, using the quantization formula, coeff ═ x · 2 Δ,
if the state is 2 or 3, using the quantization formula,
Figure BDA0002531558580000061
where Δ is the quantization step. x is the reconstruction level index value and coeff is the reconstruction level value.
And mapping the reconstruction level index value into a reconstruction level value by adopting a corresponding inverse quantization method according to the current state of the dual inverse quantizer, wherein the reconstruction level value belongs to one of two reconstruction level value groups, the reconstruction level values in the two groups except the 0 value are distributed in an interlaced way, and the reconstruction level value is output as the value of the coefficient to be decoded, namely the structure of the dual quantizer.
(5) And updating the state of the dual inverse quantizer according to the current state of the dual inverse quantizer and the reconstruction level group corresponding to the reconstruction level value. In the process of inverse quantization, after the current coefficient is inversely quantized, the state of the dual inverse quantizer is updated according to the state of the dual inverse quantizer and the parity of the coefficient, and the state determines the selection of the quantizer of the next coefficient.
As shown in fig. 2, the dual inverse quantizer state transition relationship is: when the dual dequantizer is in the state S00, if the selected reconstruction level value belongs to V00, the dual dequantizer is updated to S00, and if the selected reconstruction level value belongs to V01, the dual dequantizer is updated to S10; when the dual dequantizer is in the state S01, if the selected reconstruction level value belongs to V00, the dual dequantizer is updated to S10, and if the selected reconstruction level value belongs to V01, the dual dequantizer is updated to S00; when the dual dequantizer is in the state S10, if the selected reconstruction level value belongs to V10, the dual dequantizer is updated to S01, and if the selected reconstruction level value belongs to V11, the quantizer is updated to S11; when the dual inverse quantizer is in state S11, the dual inverse quantizer state is updated to S11 if the selected reconstruction level value belongs to V10, and the dual inverse quantizer state is updated to S01 if the selected reconstruction level value belongs to V11.
Example 3
A method for inverse quantization and entropy decoding, specifically comprising:
(1) analyzing values of a plurality of syntax elements in the code stream, wherein for at least one syntax element, a probability model selected from a plurality of sets of probability models according to the current state of the dual inverse quantizer is used and analyzed: for example, only the significant flag indicating whether the coefficient is non-zero is analyzed by using the probability model selected from the multiple sets of probability models according to the current state of the dual dequantizer. In the code stream, all syntax elements of a previous coefficient of a current coefficient to be decoded should be before the syntax element S of the current coefficient to be decoded, for example, the syntax element S is a significant flag indicating whether the coefficient is nonzero, and after all syntax elements of the previous coefficient have to be completely analyzed, a probability model is selected from multiple groups of probability models for the syntax element S of the current coefficient according to the current state of the dual dequantizer and then analyzed.
(2) Combining values of a plurality of syntax elements to obtain a reconstruction level index value of the current coefficient to be decoded: specific implementations are various, for example, a combination of a flag indicating whether the coefficient is non-zero, a flag indicating whether the coefficient is greater than 1, a flag indicating whether the coefficient is greater than 2, and the remainder of the coefficient minus 2 is used, for example, a combination of a flag indicating whether the coefficient is non-zero, a flag indicating whether the coefficient is an even number, a flag indicating whether the coefficient is greater than 2, and the remainder of the coefficient minus 2 is used.
(3) The inverse quantization method is that according to the current state of the dual inverse quantizer, the reconstruction level index value is mapped into a reconstruction level value by adopting a corresponding inverse quantization method, the reconstruction level value is output as the value of the coefficient to be decoded, the dual inverse quantizer shown in figure 1 is adopted to divide the quantization value into four sets, V00, V01, V10 and V11, each coefficient has an individual state value, the reconstruction level value is obtained according to the state of the current dual inverse quantizer, the specific inverse quantization operation is as follows,
if the state is 0 or 1, using the quantization formula, coeff ═ x · 2 Δ,
if the state is 2 or 3, using the quantization formula,
Figure BDA0002531558580000071
where Δ is the quantization step. x is the reconstruction level index value and coeff is the reconstruction level value.
And mapping the reconstruction level index value into a reconstruction level value by adopting a corresponding inverse quantization method according to the current state of the dual inverse quantizer, wherein the reconstruction level value belongs to one of two reconstruction level value groups, the reconstruction level values in the two groups except the 0 value are distributed in an interlaced way, and the reconstruction level value is output as the value of the coefficient to be decoded, namely the structure of the dual quantizer.
(4) And updating the state of the dual inverse quantizer according to the current state of the dual inverse quantizer and the reconstruction level group corresponding to the reconstruction level value. In the process of inverse quantization, after the current coefficient is inversely quantized, the state of the dual inverse quantizer is updated according to the state of the dual inverse quantizer and the parity of the coefficient, and the state determines the selection of the quantizer of the next coefficient.
As shown in fig. 2, the dual inverse quantizer state transition relationship is: when the dual dequantizer is in the state S00, if the selected reconstruction level value belongs to V00, the dual dequantizer is updated to S00, and if the selected reconstruction level value belongs to V01, the dual dequantizer is updated to S10; when the dual dequantizer is in the state S01, if the selected reconstruction level value belongs to V00, the dual dequantizer is updated to S10, and if the selected reconstruction level value belongs to V01, the dual dequantizer is updated to S00; when the dual dequantizer is in the state S10, if the selected reconstruction level value belongs to V10, the dual dequantizer is updated to S01, and if the selected reconstruction level value belongs to V11, the quantizer is updated to S11; when the dual inverse quantizer is in state S11, the dual inverse quantizer state is updated to S11 if the selected reconstruction level value belongs to V10, and the dual inverse quantizer state is updated to S01 if the selected reconstruction level value belongs to V11.
Example 4
A method for inverse quantization and entropy decoding, specifically comprising:
(1) analyzing values of a plurality of syntax elements in the code stream, wherein for at least one syntax element, a probability model is selected from a plurality of groups of probability models according to the current state of the dual inverse quantizer and is analyzed: for example, only the significant flag indicating whether the coefficient is non-zero is parsed by using the probability model selected from the plurality of sets of probability models according to the current state of the dual dequantizer.
(2) Combining values of a plurality of syntax elements to obtain a reconstruction level index value of the current coefficient to be decoded: specific implementations are various, for example, a combination of a flag indicating whether the coefficient is non-zero, a flag indicating whether the coefficient is greater than 1, a flag indicating whether the coefficient is greater than 2, and the remainder of the coefficient minus 2 is used, for example, a combination of a flag indicating whether the coefficient is non-zero, a flag indicating whether the coefficient is an even number, a flag indicating whether the coefficient is greater than 2, and the remainder of the coefficient minus 2 is used.
(3) The inverse quantization method comprises mapping the reconstruction level index value to a reconstruction level value by corresponding inverse quantization method according to the current state of the dual inverse quantizer, outputting the reconstruction level value as the value of the coefficient to be decoded, obtaining the reconstruction level value by using a dual inverse quantizer different from that shown in FIG. 1 according to the current state of the dual inverse quantizer, specifically performing inverse quantization operation as follows,
if the state is 0 or 1, using the quantization formula, coeff ═ x · 2 Δ,
if the state is 2 or 3, using the quantization formula,
Figure BDA0002531558580000081
where Δ is the quantization step. x is the reconstruction level index value and coeff is the reconstruction level value.
According to the current state of the dual dequantizer, the reconstruction level index value is mapped into a reconstruction level value by adopting a corresponding dequantization method, the reconstruction level value belongs to one of two reconstruction level value groups, the reconstruction level values in the two groups are distributed in a staggered mode, and the reconstruction level value is output as the value of a coefficient to be decoded, so that the dual dequantizer is in the structure, and is different from a common dequantizer in that more candidate reconstruction level values are provided, and therefore the coding efficiency is improved.
(4) And updating the state of the dual inverse quantizer according to the current state of the dual inverse quantizer and the reconstruction level group corresponding to the reconstruction level value. In the process of inverse quantization, after the current coefficient is inversely quantized, the state of the dual inverse quantizer is updated according to the state of the dual inverse quantizer and the parity of the coefficient, and the state determines the selection of the quantizer of the next coefficient.
As shown in fig. 2, the dual inverse quantizer state transition relationship is: when the dual dequantizer is in the state S00, if the selected reconstruction level value belongs to V00, the dual dequantizer is updated to S00, and if the selected reconstruction level value belongs to V01, the dual dequantizer is updated to S10; when the dual dequantizer is in the state S01, if the selected reconstruction level value belongs to V00, the dual dequantizer is updated to S10, and if the selected reconstruction level value belongs to V01, the dual dequantizer is updated to S00; when the dual dequantizer is in the state S10, if the selected reconstruction level value belongs to V10, the dual dequantizer is updated to S01, and if the selected reconstruction level value belongs to V11, the quantizer is updated to S11; when the dual inverse quantizer is in state S11, the dual inverse quantizer state is updated to S11 if the selected reconstruction level value belongs to V10, and the dual inverse quantizer state is updated to S01 if the selected reconstruction level value belongs to V11.
Example 5
A method for inverse quantization and entropy decoding, specifically comprising:
(1) and judging whether the current transformation block meets the condition according to the range information of the nonzero coefficient of the current transformation block, and if so, using a dual quantization technology. For how to judge whether the current transformation block meets the condition, different implementation methods can be provided, for example, according to x and y of the first non-zero coefficient in the scanning sequence in the transformation block, when x and y meet a certain size respectively, the transformation block meets the condition, and then a dual quantization technology is used; for example, according to the coordinates x and y of the lower right corner of the minimum rectangle enclosed by all non-zero coefficients in the transform block, when x and y satisfy a certain size respectively, the transform block meets the condition, and then the dual quantization technique is used.
(2) Analyzing values of a plurality of syntax elements in the code stream, wherein for at least one syntax element, a probability model selected from a plurality of sets of probability models according to the current state of the dual inverse quantizer is used and analyzed: for example, only the significant flag indicating whether the coefficient is non-zero is parsed by using the probability model selected from the plurality of sets of probability models according to the current state of the dual dequantizer.
(3) Combining values of a plurality of syntax elements to obtain a reconstruction level index value of the current coefficient to be decoded: specific implementations are various, for example, a combination of a flag indicating whether the coefficient is non-zero, a flag indicating whether the coefficient is greater than 1, a flag indicating whether the coefficient is greater than 2, and the remainder of the coefficient minus 2 is used, for example, a combination of a flag indicating whether the coefficient is non-zero, a flag indicating whether the coefficient is an even number, a flag indicating whether the coefficient is greater than 2, and the remainder of the coefficient minus 2 is used.
(4) The inverse quantization method comprises mapping the reconstruction level index value to a reconstruction level value by corresponding inverse quantization method according to the current state of the dual inverse quantizer, outputting the reconstruction level value as the value of the coefficient to be decoded, obtaining the reconstruction level value by using a dual inverse quantizer different from that shown in FIG. 1 according to the current state of the dual inverse quantizer, specifically performing inverse quantization operation as follows,
if the state is 0 or 1, using the quantization formula, coeff ═ x · 2 Δ,
if the state is 2 or 3, using the quantization formula,
Figure BDA0002531558580000091
where Δ is the quantization step. x is the reconstruction level index value and coeff is the reconstruction level value.
According to the current state of the dual dequantizer, the reconstruction level index value is mapped into a reconstruction level value by adopting a corresponding dequantization method, the reconstruction level value belongs to one of two reconstruction level value groups, the reconstruction level values in the two groups are distributed in a staggered mode, and the reconstruction level value is output as the value of a coefficient to be decoded, so that the dual dequantizer is in the structure, and is different from a common dequantizer in that more candidate reconstruction level values are provided, and therefore the coding efficiency is improved.
(5) And updating the state of the dual inverse quantizer according to the current state of the dual inverse quantizer and the reconstruction level group corresponding to the reconstruction level value. In the process of inverse quantization, after the current coefficient is inversely quantized, the state of the dual inverse quantizer is updated according to the state of the dual inverse quantizer and the parity of the coefficient, and the state determines the selection of the quantizer of the next coefficient.
As shown in fig. 2, the dual inverse quantizer state transition relationship is: when the dual dequantizer is in the state S00, if the selected reconstruction level value belongs to V00, the dual dequantizer is updated to S00, and if the selected reconstruction level value belongs to V01, the dual dequantizer is updated to S10; when the dual dequantizer is in the state S01, if the selected reconstruction level value belongs to V00, the dual dequantizer is updated to S10, and if the selected reconstruction level value belongs to V01, the dual dequantizer is updated to S00; when the dual dequantizer is in the state S10, if the selected reconstruction level value belongs to V10, the dual dequantizer is updated to S01, and if the selected reconstruction level value belongs to V11, the quantizer is updated to S11; when the dual inverse quantizer is in state S11, the dual inverse quantizer state is updated to S11 if the selected reconstruction level value belongs to V10, and the dual inverse quantizer state is updated to S01 if the selected reconstruction level value belongs to V11.
Example 6
A method for inverse quantization and entropy decoding, specifically comprising:
(1) analyzing values of a plurality of syntax elements in the code stream, wherein for at least one syntax element, a probability model selected from a plurality of sets of probability models according to the current state of the dual inverse quantizer is used and analyzed: for example, only the significant flag indicating whether the coefficient is non-zero is analyzed by using the probability model selected from the multiple sets of probability models according to the current state of the dual dequantizer. In the code stream, all syntax elements of a previous coefficient of a current coefficient to be decoded should be before the syntax element S of the current coefficient to be decoded, for example, the syntax element S is a significant flag indicating whether the coefficient is nonzero, and after all syntax elements of the previous coefficient have to be completely analyzed, a probability model is selected from multiple groups of probability models for the syntax element S of the current coefficient according to the current state of the dual dequantizer and then analyzed.
(2) Combining values of a plurality of syntax elements to obtain a reconstruction level index value of the current coefficient to be decoded: specific implementations are various, for example, a combination of a flag indicating whether the coefficient is non-zero, a flag indicating whether the coefficient is greater than 1, a flag indicating whether the coefficient is greater than 2, and the remainder of the coefficient minus 2 is used, for example, a combination of a flag indicating whether the coefficient is non-zero, a flag indicating whether the coefficient is an even number, a flag indicating whether the coefficient is greater than 2, and the remainder of the coefficient minus 2 is used.
(3) The inverse quantization method is that according to the current state of the dual inverse quantizer, the reconstruction level index value is mapped into a reconstruction level value by adopting a corresponding inverse quantization method, the reconstruction level value is output as the value of the coefficient to be decoded, the dual inverse quantizer different from the one shown in figure 1 is adopted to divide the quantization value into four sets, V00, V01, V10 and V11, each coefficient has an individual state value, the reconstruction level value is obtained according to the state of the current dual inverse quantizer, the specific inverse quantization operation is as follows,
if the state is 0 or 1, using the quantization formula, coeff ═ x · 2 Δ,
if the state is 2 or 3, using the quantization formula,
Figure BDA0002531558580000101
where Δ is the quantization step. x is the reconstruction level index value and coeff is the reconstruction level value.
According to the current state of the dual dequantizer, the reconstruction level index value is mapped into a reconstruction level value by adopting a corresponding dequantization method, the reconstruction level value belongs to one of two reconstruction level value groups, the reconstruction level values in the two groups are distributed in a staggered mode, and the reconstruction level value is output as the value of a coefficient to be decoded, so that the dual dequantizer is in the structure, and is different from a common dequantizer in that more candidate reconstruction level values are provided, and therefore the coding efficiency is improved.
(4) And updating the state of the dual inverse quantizer according to the current state of the dual inverse quantizer and the reconstruction level group corresponding to the reconstruction level value. In the process of inverse quantization, after the current coefficient is inversely quantized, the state of the dual inverse quantizer is updated according to the state of the dual inverse quantizer and the parity of the coefficient, and the state determines the selection of the quantizer of the next coefficient.
As shown in fig. 2, the dual inverse quantizer state transition relationship is: when the dual dequantizer is in the state S00, if the selected reconstruction level value belongs to V00, the dual dequantizer is updated to S00, and if the selected reconstruction level value belongs to V01, the dual dequantizer is updated to S10; when the dual dequantizer is in the state S01, if the selected reconstruction level value belongs to V00, the dual dequantizer is updated to S10, and if the selected reconstruction level value belongs to V01, the dual dequantizer is updated to S00; when the dual dequantizer is in the state S10, if the selected reconstruction level value belongs to V10, the dual dequantizer is updated to S01, and if the selected reconstruction level value belongs to V11, the quantizer is updated to S11; when the dual inverse quantizer is in state S11, the dual inverse quantizer state is updated to S11 if the selected reconstruction level value belongs to V10, and the dual inverse quantizer state is updated to S01 if the selected reconstruction level value belongs to V11.
Example 7
A method for inverse quantization and entropy decoding, specifically comprising:
(1) analyzing values of a plurality of syntax elements in the code stream, wherein for at least one syntax element, a probability model selected from a plurality of sets of probability models according to the current state of the dual inverse quantizer is used and analyzed: in the code stream, the decoding sequence of the coefficient to be decoded of the transform block is shown by the following pseudo code:
Figure BDA0002531558580000112
(2) combining values of a plurality of syntax elements to obtain a reconstruction level index value of the current coefficient to be decoded: specific implementations are various, for example, a combination of a flag indicating whether the coefficient is non-zero, a flag indicating whether the coefficient is greater than 1, a flag indicating whether the coefficient is greater than 2, and the remainder of the coefficient minus 2 is used, for example, a combination of a flag indicating whether the coefficient is non-zero, a flag indicating whether the coefficient is an even number, a flag indicating whether the coefficient is greater than 2, and the remainder of the coefficient minus 2 is used.
(3) The inverse quantization method is that according to the current state of the dual inverse quantizer, the reconstruction level index value is mapped into a reconstruction level value by adopting a corresponding inverse quantization method, the reconstruction level value is output as the value of the coefficient to be decoded, the dual inverse quantizer shown in figure 1 is adopted to divide the quantization value into four sets, V00, V01, V10 and V11, each coefficient has an individual state value, the reconstruction level value is obtained according to the state of the current dual inverse quantizer, the specific inverse quantization operation is as follows,
if the state is 0 or 1, using the quantization formula, coeff ═ x · 2 Δ,
if the state is 2 or 3, using the quantization formula,
Figure BDA0002531558580000111
where Δ is the quantization step. x is the reconstruction level index value and coeff is the reconstruction level value.
And mapping the reconstruction level index value into a reconstruction level value by adopting a corresponding inverse quantization method according to the current state of the dual inverse quantizer, wherein the reconstruction level value belongs to one of two reconstruction level value groups, the reconstruction level values in the two groups except the 0 value are distributed in an interlaced way, and the reconstruction level value is output as the value of the coefficient to be decoded, namely the structure of the dual quantizer.
(4) And updating the state of the dual inverse quantizer according to the current state of the dual inverse quantizer and the reconstruction level group corresponding to the reconstruction level value. In the process of inverse quantization, after the current coefficient is inversely quantized, the state of the dual inverse quantizer is updated according to the state of the dual inverse quantizer and the parity of the coefficient, and the state determines the selection of the quantizer of the next coefficient.
As shown in fig. 2, the dual inverse quantizer state transition relationship is: when the dual dequantizer is in the state S00, if the selected reconstruction level value belongs to V00, the dual dequantizer is updated to S00, and if the selected reconstruction level value belongs to V01, the dual dequantizer is updated to S10; when the dual dequantizer is in the state S01, if the selected reconstruction level value belongs to V00, the dual dequantizer is updated to S10, and if the selected reconstruction level value belongs to V01, the dual dequantizer is updated to S00; when the dual dequantizer is in the state S10, if the selected reconstruction level value belongs to V10, the dual dequantizer is updated to S01, and if the selected reconstruction level value belongs to V11, the quantizer is updated to S11; when the dual inverse quantizer is in state S11, the dual inverse quantizer state is updated to S11 if the selected reconstruction level value belongs to V10, and the dual inverse quantizer state is updated to S01 if the selected reconstruction level value belongs to V11.
Example 8
A method for quantization and entropy coding, specifically comprising:
(1) and mapping the current coefficient to be coded into a reconstruction level index value by adopting a corresponding quantization method according to the current state of the dual quantizer. The quantization method comprises obtaining reconstruction level value according to the state of the dual quantizer shown in FIG. 1, specifically performing the following quantization operations,
if the state is 0 or 1, using the quantization formula, coeff ═ x · 2 Δ,
if the state is 2 or 3, using the quantization formula,
Figure BDA0002531558580000121
where Δ is the quantization step. x is the reconstruction level index value and coeff is the reconstruction level value.
(2) Describing the reconstruction level index value using values of a plurality of syntax elements: specific implementations are various, for example, a combination of a flag indicating whether the coefficient is non-zero, a flag indicating whether the coefficient is greater than 1, a flag indicating whether the coefficient is greater than 2, and the remainder of the coefficient minus 2 is used, for example, a combination of a flag indicating whether the coefficient is non-zero, a flag indicating whether the coefficient is an even number, a flag indicating whether the coefficient is greater than 2, and the remainder of the coefficient minus 2 is used.
(3) Entropy coding the plurality of syntax elements and writing the syntax elements into a code stream, wherein for the value of at least one syntax element S in the plurality of syntax elements, the entropy coding is carried out by using a probability model selected from a plurality of groups of probability models according to the current state of a dual quantizer: for example, only the significant flag indicating whether the coefficient is non-zero is parsed by using the probability model selected from the plurality of sets of probability models according to the current state of the dual quantizer.
(4) And updating the state of the dual quantizer according to the current state of the dual quantizer and the reconstruction level index group to which the reconstruction level index value belongs. In the process of quantization, after the current coefficient is quantized, the state of the dual quantizer is updated according to the state of the dual quantizer and the parity of the coefficient, and the state determines the selection of the dual quantizer of the next coefficient.
As shown in fig. 2, the dual quantizer state transition relationship is: when the dual quantizer is in state S00, if the selected reconstruction level value belongs to V00, the dual quantizer state is updated to S00, and if the selected reconstruction level value belongs to V01, the dual quantizer state is updated to S10; when the dual quantizer is in state S01, if the selected reconstruction level value belongs to V00, the dual quantizer state is updated to S10, and if the selected reconstruction level value belongs to V01, the dual quantizer state is updated to S00; when the dual quantizer is in state S10, updating the dual quantizer state to S01 if the selected reconstruction level value belongs to V10, and updating the quantizer state to S11 if the selected reconstruction level value belongs to V11; when the dual quantizer is in state S11, the dual quantizer state is updated to S11 if the selected reconstruction level value belongs to V10, and the dual quantizer state is updated to S01 if the selected reconstruction level value belongs to V11.
Example 9
A method for quantization and entropy coding, specifically comprising:
(1) judging whether the current transformation block meets the condition according to the range information of the nonzero coefficient of the current transformation block,
(2) if so, a dual quantization technique is used. For how to judge whether the current transformation block meets the condition, different implementation methods can be provided, for example, according to x and y of the first non-zero coefficient in the scanning sequence in the transformation block, whether the range information of the non-zero coefficient of the current transformation block meets the condition is judged, when x and y respectively meet a certain size, the transformation block meets the condition, and then a dual quantization technology is used; for example, according to the coordinates x and y of the lower right corner of a minimum rectangle surrounded by all non-zero coefficients in a transform block, whether the range information of the non-zero coefficients of the current transform block meets the condition is judged, and when x and y respectively meet a certain size, the transform block meets the condition, and then a dual quantization technology is used.
(2) And mapping the current coefficient to be coded into a reconstruction level index value by adopting a corresponding quantization method according to the current state of the dual quantizer. The quantization method comprises obtaining reconstruction level value according to the state of the dual quantizer shown in FIG. 1, specifically performing the following quantization operations,
if the state is 0 or 1, using the quantization formula, coeff ═ x · 2 Δ,
if the state is 2 or 3, using the quantization formula,
Figure BDA0002531558580000131
where Δ is the quantization step. x is the reconstruction level index value and coeff is the reconstruction level value.
(3) Describing the reconstruction level index value using values of a plurality of syntax elements: specific implementations are various, for example, a combination of a flag indicating whether the coefficient is non-zero, a flag indicating whether the coefficient is greater than 1, a flag indicating whether the coefficient is greater than 2, and the remainder of the coefficient minus 2 is used, for example, a combination of a flag indicating whether the coefficient is non-zero, a flag indicating whether the coefficient is an even number, a flag indicating whether the coefficient is greater than 2, and the remainder of the coefficient minus 2 is used.
(4) Entropy coding the plurality of syntax elements and writing the syntax elements into a code stream, wherein for the value of at least one syntax element S in the plurality of syntax elements, entropy coding is carried out by using a probability model selected from a plurality of groups of probability models according to the current state of a dual quantizer: for example, only the significant flag indicating whether the coefficient is non-zero is parsed by using the probability model selected from the plurality of sets of probability models according to the current state of the dual quantizer.
(5) And updating the state of the dual quantizer according to the current state of the dual quantizer and the reconstruction level index group to which the reconstruction level index value belongs. In the process of quantization, after the current coefficient is quantized, the state of the dual quantizer is updated according to the state of the dual quantizer and the parity of the coefficient, and the state determines the selection of the dual quantizer of the next coefficient.
As shown in fig. 2, the dual quantizer state transition relationship is: when the dual quantizer is in state S00, if the selected reconstruction level value belongs to V00, the dual quantizer state is updated to S00, and if the selected reconstruction level value belongs to V01, the dual quantizer state is updated to S10; when the dual quantizer is in state S01, if the selected reconstruction level value belongs to V00, the dual quantizer state is updated to S10, and if the selected reconstruction level value belongs to V01, the dual quantizer state is updated to S00; when the dual quantizer is in state S10, updating the dual quantizer state to S01 if the selected reconstruction level value belongs to V10, and updating the quantizer state to S11 if the selected reconstruction level value belongs to V11; when the dual quantizer is in state S11, the dual quantizer state is updated to S11 if the selected reconstruction level value belongs to V10, and the dual quantizer state is updated to S01 if the selected reconstruction level value belongs to V11.
Example 10
A method for quantization and entropy coding, specifically comprising:
(1) and mapping the current coefficient to be coded into a reconstruction level index value by adopting a corresponding quantization method according to the current state of the dual quantizer. The quantization method comprises obtaining reconstruction level value according to the state of the dual quantizer shown in FIG. 1, specifically performing the following quantization operations,
if the state is 0 or 1, using the quantization formula, coeff ═ x · 2 Δ,
if the state is 2 or 3, using the quantization formula,
Figure BDA0002531558580000141
where Δ is the quantization step. x is the reconstruction level index value and coeff is the reconstruction level value.
(2) Describing the reconstruction level index value using values of a plurality of syntax elements: specific implementations are various, for example, a combination of a flag indicating whether the coefficient is non-zero, a flag indicating whether the coefficient is greater than 1, a flag indicating whether the coefficient is greater than 2, and the remainder of the coefficient minus 2 is used, for example, a combination of a flag indicating whether the coefficient is non-zero, a flag indicating whether the coefficient is an even number, a flag indicating whether the coefficient is greater than 2, and the remainder of the coefficient minus 2 is used.
(3) And entropy coding the plurality of syntax elements and writing the syntax elements into a code stream, wherein, for a significant flag which is used for representing whether the coefficient is nonzero in the plurality of syntax elements, entropy coding is carried out by using a probability model selected from the plurality of groups of probability models according to the current state of the dual quantizer.
(4) And updating the state of the dual quantizer according to the current state of the dual quantizer and the reconstruction level index group to which the reconstruction level index value belongs. In the process of quantization, after the current coefficient is quantized, the state of the dual quantizer is updated according to the state of the dual quantizer and the parity of the coefficient, and the state determines the selection of the dual quantizer of the next coefficient.
As shown in fig. 2, the dual quantizer state transition relationship is: when the dual quantizer is in state S00, if the selected reconstruction level value belongs to V00, the dual quantizer state is updated to S00, and if the selected reconstruction level value belongs to V01, the dual quantizer state is updated to S10; when the dual quantizer is in state S01, if the selected reconstruction level value belongs to V00, the dual quantizer state is updated to S10, and if the selected reconstruction level value belongs to V01, the dual quantizer state is updated to S00; when the dual quantizer is in state S10, updating the dual quantizer state to S01 if the selected reconstruction level value belongs to V10, and updating the quantizer state to S11 if the selected reconstruction level value belongs to V11; when the dual quantizer is in state S11, the dual quantizer state is updated to S11 if the selected reconstruction level value belongs to V10, and the dual quantizer state is updated to S01 if the selected reconstruction level value belongs to V11.
Example 11
An apparatus for inverse quantization and entropy decoding, comprising:
(1) a coefficient entropy decoding module: the input of the module is a code stream, the output of the module is a plurality of syntactic elements, the module analyzes the values of the syntactic elements in the code stream, wherein for at least one syntactic element S, a probability model is selected from a plurality of groups of probability models according to the current state of the inverse quantizer, and the selected probability model is used for analyzing the values of the syntactic elements.
(2) The reconstruction level index value decoding module: the module combines values of the plurality of syntax elements to obtain a reconstruction level index value of the current coefficient to be decoded.
(3) A dual inverse quantization module: the module maps the reconstruction level index value to the reconstruction level value by adopting a corresponding inverse quantization method according to the state of the dual inverse quantizer, and outputs the reconstruction level value as the value of the coefficient to be decoded.
(4) A state updating module: the input of the dual inverse quantizer is the current state of the dual inverse quantizer and the reconstruction level group corresponding to the reconstruction level value, the output is the updated dual inverse quantizer state, and the module updates the state of the inverse quantizer according to the current state of the inverse quantizer and the reconstruction level index group to which the reconstruction level index value belongs.
Example 12
An apparatus for inverse quantization and entropy decoding, comprising:
(1) a coefficient entropy decoding module: the input of the module is a code stream, the output of the module is a plurality of syntactic elements, the module analyzes the values of the syntactic elements in the code stream, wherein for at least one syntactic element S, a probability model is selected from a plurality of groups of probability models according to the current state of the inverse quantizer, and the selected probability model is used for analyzing the values of the syntactic elements.
(2) The reconstruction level index value decoding module: the module combines values of the plurality of syntax elements to obtain a reconstruction level index value of the current coefficient to be decoded.
(3) A dual inverse quantization module: the module maps the reconstruction level index value to the reconstruction level value by adopting a corresponding inverse quantization method according to the state of the dual inverse quantizer, and outputs the reconstruction level value as the value of the coefficient to be decoded.
(4) A state updating module: the input of the dual inverse quantizer is the current state of the dual inverse quantizer and the reconstruction level group corresponding to the reconstruction level value, the output is the updated dual inverse quantizer state, and the module updates the state of the inverse quantizer according to the current state of the inverse quantizer and the reconstruction level index group to which the reconstruction level index value belongs.
(5) A transform block selection module: the input is the area of the non-zero coefficient area of the transformation block, the selection result of the transformation block is output, and the module determines whether the transformation block is selected to use the dual quantization technology according to the area of the non-zero coefficient area of the transformation block.
Example 13
An apparatus for inverse quantization and entropy decoding, comprising:
(1) a coefficient entropy decoding module: the input of the module is a code stream, the output of the module is a plurality of syntactic elements, the module analyzes the values of the syntactic elements in the code stream, wherein for at least one syntactic element S, a probability model is selected from a plurality of groups of probability models according to the current state of the inverse quantizer, and the selected probability model is used for analyzing the values of the syntactic elements.
Model storage submodule: the sub-module stores a probability model of the candidate.
Model selection submodule: the sub-module selects a probability model from the candidate probability models.
Model parsing submodule: the sub-module parses the syntax element using the selected probabilistic model.
(2) The reconstruction level index value decoding module: the module combines values of the plurality of syntax elements to obtain a reconstruction level index value of the current coefficient to be decoded.
(3) A dual inverse quantization module: the module maps the reconstruction level index value to the reconstruction level value by adopting a corresponding inverse quantization method according to the state of the dual inverse quantizer, and outputs the reconstruction level value as the value of the coefficient to be decoded.
(4) A state updating module: the input of the dual inverse quantizer is the current state of the dual inverse quantizer and the reconstruction level group corresponding to the reconstruction level value, the output is the updated dual inverse quantizer state, and the module updates the state of the inverse quantizer according to the current state of the inverse quantizer and the reconstruction level index group to which the reconstruction level index value belongs.
Example 14
An apparatus for coefficient quantization and entropy coding, comprising:
(1) a dual quantization module: the input of the module is the coefficient to be coded and the current state of the dual quantizer, and the output is the index value of the reconstruction level.
(2) The reconstruction level index value coding module: the module uses values of the plurality of syntax elements to describe the reconstruction level index value.
(3) A coefficient entropy coding module: the input of the module is a plurality of syntax elements, the output of the module is a code stream, and the module carries out entropy coding on the syntax elements and writes the syntax elements into the code stream; wherein for a value of at least one syntax element S of the plurality of syntax elements, entropy coding is performed using a probability model selected from a plurality of sets of probability models in dependence on a current state of a dual quantizer.
(4) A state updating module: the module updates the state of the dual quantizer according to the current state of the dual quantizer and the reconstruction level index group to which the reconstruction level index value belongs.

Claims (10)

1. A method of inverse coefficient quantization and entropy decoding, comprising:
s1, analyzing the values of a plurality of syntax elements of the coefficients in the code stream, wherein for at least one syntax element S, a probability model is selected from a plurality of groups of probability models according to the current state of the dual inverse quantizer, and the selected probability model is used for analyzing;
s2, obtaining a reconstruction level index value of the current coefficient to be decoded based on the value combination of the plurality of syntax elements;
s3, according to the current state of the dual inverse quantizer, mapping the reconstruction level index value to a reconstruction level value by adopting a corresponding inverse quantization method, and outputting the reconstruction level value as the value of the coefficient to be decoded;
and S4, updating the state of the dual inverse quantizer according to the current state of the dual inverse quantizer and the reconstruction level index group to which the reconstruction level index value belongs.
2. A method of inverse coefficient quantization and entropy decoding as claimed in claim 1, characterized by:
the syntax element S is an identifier for indicating whether the coefficient to be decoded is nonzero;
there are four different states of the dual inverse quantizer;
the selecting and analyzing the probability model specifically comprises selecting a group of probability models from the four groups of probability models according to the current state of the dual inverse quantizer, and analyzing the syntax element S by using one probability model in the group of probability models.
3. A method of inverse coefficient quantization and entropy decoding as claimed in claim 1, characterized by:
in the code stream, all syntax elements of a previous coefficient of a current coefficient to be decoded are before the syntax element S of the current coefficient to be decoded.
4. A method of inverse coefficient quantization and entropy decoding as claimed in claim 1, further comprising:
s0, before analyzing the values of multiple syntax elements of the coefficient in the code stream, judging whether the current transformation block meets the condition according to the range information of the nonzero coefficient of the current transformation block, if so, executing the method of the steps S1-S4.
5. A method of coefficient quantization and entropy coding, comprising:
according to the current state of the dual quantizer, mapping the current coefficient to be coded into a reconstruction level index value by adopting a corresponding quantization method;
describing the reconstruction level index value using values of a plurality of syntax elements;
entropy coding is carried out on the plurality of syntax elements and code streams are written in; selecting a probability model from a plurality of sets of probability models according to the current state of the dual quantizer for the value of at least one syntax element S in the plurality of syntax elements, and performing entropy coding by using the selected probability model;
and updating the state of the dual quantizer according to the current state of the dual quantizer and the reconstruction level index group to which the reconstruction level index value belongs.
6. A method of coefficient quantization and entropy coding as claimed in claim 5, characterized by:
the syntax element S is a mark for indicating whether the coefficient to be coded is nonzero or not;
there are four different states of the dual quantizer;
the selecting the probability model and performing entropy coding specifically includes selecting a group of probability models from four groups of probability models according to the current state of the dual quantizer, and coding the syntax element S by using one probability model in the group of probability models.
7. A coefficient inverse quantization and entropy decoding apparatus, characterized by comprising:
a coefficient entropy decoding module: the input of the module is a code stream, the output of the module is a plurality of syntactic elements, the module analyzes the values of the syntactic elements in the code stream, wherein for at least one syntactic element S, a probability model is selected from a plurality of groups of probability models according to the current state of an inverse quantizer, and the selected probability model is used for analyzing the values of the syntactic elements;
the reconstruction level index value decoding module: the module combines the values of the plurality of syntax elements to obtain the reconstruction level index value of the current coefficient to be decoded;
a dual inverse quantization module: the module adopts a corresponding inverse quantization method to map the reconstruction level index value into the reconstruction level value according to the state of the dual inverse quantizer, and outputs the reconstruction level value as the value of the coefficient to be decoded;
a state updating module: the input of the dual inverse quantizer is the current state of the dual inverse quantizer and the reconstruction level group corresponding to the reconstruction level value, the output is the updated dual inverse quantizer state, and the module updates the state of the inverse quantizer according to the current state of the inverse quantizer and the reconstruction level index group to which the reconstruction level index value belongs.
8. The apparatus for inverse coefficient quantization and entropy decoding of claim 7, wherein:
the syntax element S is an identifier for indicating whether the coefficient to be decoded is nonzero;
the inverse quantizer exists in four different states;
the selecting and analyzing the probability model specifically comprises: and selecting one group of probability models from the four groups of probability models according to the current state of the dual inverse quantizer, and analyzing the syntax element S by using one probability model in the group of probability models.
9. An apparatus for coefficient quantization and entropy coding, comprising:
a dual quantization module: the input of the module is a coefficient to be coded and the current state of a dual quantizer, and the output is a reconstructed level index value, and the module adopts a corresponding quantization method to map the current coefficient to be coded into the reconstructed level index value;
the reconstruction level index value coding module: the module uses the values of the plurality of syntax elements to describe the reconstruction level index value;
a coefficient entropy coding module: the input of the module is a plurality of syntax elements, the output of the module is a code stream, and the module carries out entropy coding on the syntax elements and writes the syntax elements into the code stream; selecting a probability model from a plurality of sets of probability models according to the current state of the dual quantizer for the value of at least one syntax element S in the plurality of syntax elements, and performing entropy coding by using the selected probability model;
a state updating module: the module updates the state of the dual quantizer according to the current state of the dual quantizer and the reconstruction level index group to which the reconstruction level index value belongs.
10. An apparatus for coefficient quantization and entropy coding as defined in claim 9, wherein:
the syntax element S is a mark for indicating whether the coefficient to be coded is nonzero or not;
there are four different states of the quantizer;
the selecting and encoding of the probability model specifically includes: a set of probability models is selected from the four sets of probability models, based on the current state of the quantizer, and one of the set of probability models is used to encode the syntax element S.
CN202010519720.4A 2020-03-25 2020-06-09 Quantization and entropy coding method and apparatus Active CN113453002B (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN2020102202342 2020-03-25
CN202010220234 2020-03-25

Publications (2)

Publication Number Publication Date
CN113453002A true CN113453002A (en) 2021-09-28
CN113453002B CN113453002B (en) 2022-08-09

Family

ID=77808392

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010519720.4A Active CN113453002B (en) 2020-03-25 2020-06-09 Quantization and entropy coding method and apparatus

Country Status (1)

Country Link
CN (1) CN113453002B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2022174477A1 (en) * 2021-02-22 2022-08-25 浙江大学 Encoding method, decoding method, encoder, decoder, and storage medium

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5283646A (en) * 1992-04-09 1994-02-01 Picturetel Corporation Quantizer control method and apparatus
CN1608284A (en) * 2001-12-25 2005-04-20 株式会社Ntt都科摩 Signal encoding apparatus, signal encoding method, and program
CN105979263A (en) * 2015-07-24 2016-09-28 渤海大学 Novel quantification control system coding method
US20170310999A1 (en) * 2016-04-25 2017-10-26 Magnum Semiconductor, Inc. Method and apparatus for rate-distortion optimized coefficient quantization including sign data hiding
WO2019185769A1 (en) * 2018-03-29 2019-10-03 Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. Dependent quantization

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5283646A (en) * 1992-04-09 1994-02-01 Picturetel Corporation Quantizer control method and apparatus
CN1608284A (en) * 2001-12-25 2005-04-20 株式会社Ntt都科摩 Signal encoding apparatus, signal encoding method, and program
CN105979263A (en) * 2015-07-24 2016-09-28 渤海大学 Novel quantification control system coding method
US20170310999A1 (en) * 2016-04-25 2017-10-26 Magnum Semiconductor, Inc. Method and apparatus for rate-distortion optimized coefficient quantization including sign data hiding
WO2019185769A1 (en) * 2018-03-29 2019-10-03 Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. Dependent quantization

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
郑莉华: "H.264/AVC视频编码的码率控制及并行处理研究", 《中国博士学位论文全文数据库（电子期刊）》 *

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2022174477A1 (en) * 2021-02-22 2022-08-25 浙江大学 Encoding method, decoding method, encoder, decoder, and storage medium

Also Published As

Publication number Publication date
CN113453002B (en) 2022-08-09

Similar Documents

Publication Publication Date Title
US11750841B2 (en) Methods and apparatuses for coding transform blocks
RU2673392C1 (en) Method and device for entropy video encoding and entropy video decoding method and device
RU2589382C2 (en) Method of updating parameters for entropy encoding and decoding of transformation coefficient level, as well as device for entropy encoding and device for entropy decoding of transformation coefficient level using same
US7454071B2 (en) System and method for using pattern vectors for video and image coding and decoding
US7336713B2 (en) Method and apparatus for encoding and decoding data
KR100736104B1 (en) Video coding method and apparatus supporting independent parsing
CN100488254C (en) Entropy coding method and decoding method based on text
CN107835421B (en) Method and apparatus for encoding video and method and apparatus for decoding video
CN101090495B (en) Method and system for processing video data
US8767823B2 (en) Method and apparatus for frame memory compression
RU2565877C2 (en) Method and apparatus for determining correlation between syntax element and codeword for variable length coding
CN102045560A (en) Video encoding and decoding method and video encoding and decoding equipment
CN102238387A (en) Video entropy code as well as entropy coding method, device and medium
KR20080008923A (en) Method and apparatus for entropy encoding/decoding
US20070133677A1 (en) Method and apparatus for encoding and decoding video signals on group basis
US20070071088A1 (en) Method and apparatus for entropy encoding and entropy decoding fine-granularity scalability layer video data
CN112399181B (en) Image coding and decoding method, device and storage medium
JP4163680B2 (en) Adaptive method and system for mapping parameter values to codeword indexes
CN113453002B (en) Quantization and entropy coding method and apparatus
CN102355578B (en) A kind of entropy decoding method, device
CN113365048A (en) Method, decoding end, encoding end and system for encoding and decoding
CN111988618B (en) Decoding and encoding method, decoding end and encoding end
KR20100089546A (en) Method and apparatus for successively encoding/decoding image
KR20090122390A (en) A method and a device for processing bit symbols generated by a data source, a computer readable medium, a computer program element
CN112055223A (en) Image coding and decoding method and coder-decoder

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant