CN103716623B - Video compression encoding-and-decoding method and encoder-decoder on the basis of weighting quantification - Google Patents

Abstract

The application discloses a video compression encoding-and-decoding method on the basis of weighting quantification. The encoding-and-decoding process comprises that an attribute component of an image to be encoded is divided into multiple attribute blocks; the attribute blocks are predicted so that residual blocks are obtained, and a transformation coefficient of each frequency point in the residual blocks is obtained after transformation; one default matrix is selected, and weighting calculation is performed on initial quantization step length of the frequency points so that a weighting quantification step length matrix is obtained; and the transformation coefficient is quantified by using the weighting quantification step length matrix so that quantification blocks are obtained, then the quantification blocks are written-in a code stream and other information written-in the code stream is confirmed according to values of the quantification blocks. The application also discloses a video compression encoder-decoder on the basis of weighting quantification. A preset weighting quantification matrix set comprises multiple default matrixes so that transformation coefficient matrixes under different situations can be effectively weighted and quantified, and thus code rate required by encoding can be effectively reduced without reducing subjective quality.

Description

Video compression coding-decoding method based on weight quantization and codec

Technical field

The application is related to video data process field, more particularly, to a kind of video compression coding-decoding side based on weight quantization Method and codec.

Background technology

In the video frequency compress coding standard of current main-stream, the residual block of prediction block and original block generation needs to be become Get transform block in return, transform block coefficient block of obtaining after quantifying is restored code stream.The process quantifying can cause the mistake of coefficient Very, lead to the decline of picture quality.But the subjective distortion that in transform block, the distortion of the conversion coefficient numerical value of diverse location causes is each Differ.Weight quantization technology is exactly to utilize this phenomenon, and the conversion coefficient of diverse location is carried out using different quantization steps The technology quantifying.This technology can reach in the case of not causing subjective quality to decline, and reduces encoder bit rate to greatest extent, but Using existing weight quantization coefficient matrix, subjective quality is not ideal enough, fails effectively to reduce the code check required for coding.

Content of the invention

The application technical problem to be solved is for the deficiencies in the prior art, provides a kind of video based on weight quantization Compression coding and decoding method and codec.

According to the application's in a first aspect, the application provides a kind of video compression coding-decoding method based on weight quantization, Including cataloged procedure, described cataloged procedure includes：

Extract the attribute component of image to be encoded, described attribute component is divided into some attribute blocks；

Described attribute block is predicted obtaining residual block, described residual block is entered with line translation and obtains in described residual block respectively The conversion coefficient of frequency；

Concentrate from default weighted quantization matrices and choose a default matrix, first to described frequency using described default matrix Beginning quantization step is weighted and obtains weight quantization Step matrix；

Using described weight quantization Step matrix, described conversion coefficient is quantified, obtain quantifying block, by described quantization The number information write code stream of block and described default matrix, and according to the described value quantifying block, confirm to write its of described code stream His information.

According to the second aspect of the application, the application provides a kind of video-compression codecs based on weight quantization, bag Include encoder, described encoder includes：

Division module, for extracting the attribute component of image to be encoded, described attribute component is divided into some attribute blocks；

Predictive transformation module, for being predicted obtaining residual block to described attribute block, enters line translation to described residual block Obtain the conversion coefficient of each frequency in described residual block；

Computing module, chooses a default matrix for concentrating from default weighted quantization matrices, using described default matrix The initial quantization step of described frequency is weighted and obtains weight quantization Step matrix；

Quantization modules, for being quantified to described conversion coefficient using described weight quantization Step matrix, are quantified Block, by described quantization block write code stream, and according to the described value quantifying block, confirms to write the other information of described code stream.

Due to employing above technical scheme, make what the application possessed to have the beneficial effects that：

In the specific embodiment of the application, concentrated due to default weighted quantization matrices and include multiple default matrixs, energy Effectively the transform coefficient matrix under different situations is weighted quantifying, in the case of not reducing subjective quality, effectively Reduce code check needed for coding；Also provide for encoder voluntarily to select using different default weighting amounts in image attributes block one-level simultaneously Change coefficient matrix, default weighting quantization matrix can more neatly be selected according to picture material, in the feelings not reducing subjective quality Under condition, reduce code check needed for coding further.

Brief description

Fig. 1 is the cataloged procedure applied for flow chart in one embodiment；

Fig. 2 is the decoding process flow chart in one embodiment of the application；

Fig. 3 is the application encoder structural representation in one embodiment；

Fig. 4 is the application decoder structural representation in one embodiment.

Specific embodiment

Combine accompanying drawing below by specific embodiment the application is described in further detail.

Embodiment one：

As shown in figure 1, the video compression coding-decoding method based on weight quantization of the application, a kind of its embodiment, bag Include cataloged procedure, described cataloged procedure includes：

Step 102：Extract the attribute component of image to be encoded, attribute component is divided into some attribute blocks.Attribute component Including brightness, colourity etc..

In one embodiment, attribute block can be divided in the following manner：

First all properties block is divided into 16 × 16 coding unit, a coding unit can be divided into 48 × 8, Or 44 × 16 or 4 16 × 4, or 1 16 × 16 predicting unit.

Step 104：Attribute block is predicted obtaining residual block, residual block is entered with line translation and obtains each frequency in residual block Conversion coefficient.

Step 106：Concentrate from default weighted quantization matrices and choose a default matrix, first to frequency using default matrix Beginning quantization step is weighted and obtains weight quantization Step matrix.

In one embodiment, default matrix can be selected by resolution ratio, if picture altitude is less than 480, select 4 × 4 default matrix 1,8 × 8 default matrix 1 and its 16 × 16,16 × 4,4 × 16 default matrixs derived；If picture altitude is little In 720, then select 4 × 4 default matrix 2,8 × 8 default matrix 2 and its 16 × 16,16 × 4,4 × 16 default matrixs derived； Other situations, then select 4 × 4 default matrix 3,8 × 8 default matrix 3 and its 16 × 16,16 × 4,4 × 16 acquiescence squares derived Battle array.Also default matrix can be selected by brightness or other factors,

Step 108：Using weight quantization Step matrix, conversion coefficient is quantified, obtain quantifying block, write quantifying block Enter code stream, and according to quantifying the value of block, confirm the other information of write code stream.

In one embodiment, if quantization block is full null matrix, the number information of default matrix is not write described Code stream, otherwise, the number information of default matrix is write code stream.

The video coding-decoding method based on weight quantization of the application, default weighted quantization matrices collection includes 4 × 4 acquiescence squares Battle array 1 and/or 4 × 4 default matrixs 2 and/or 4 × 4 default matrixs 3 and/or 8 × 8 default matrixs 1 and/or 8 × 8 default matrixs 2 And/or 8 × 8 default matrixs 3.

4 × 4 default matrixs 1:

4 × 4 default matrixs 2：

4 × 4 default matrixs 3：

8 × 8 default matrixs 1：

8 × 8 default matrixs 2：

8 × 8 default matrixs 3：

In one embodiment, the video coding-decoding method based on weight quantization of the application, default weight quantization square Battle array collection can also include 16 × 16 default matrixs and/or 32 × 32 default matrixs.

16 × 16 default matrixs and 32 × 32 default matrixs pass through formula respectively

Calculate and obtain；

Wherein WQ₈(x, y) is the value of (x, y) position in 8 × 8 default matrixs, and WQ (i, j) is the weighting matrix being derived by In (i, j) position value, k be objective matrix width divided by 8 multiple, a, b, c and d be constant.

In one embodiment, the video coding-decoding method based on weight quantization of the application, default weight quantization square Battle array collection can also include：2^e×2^e+2Default matrix and/or 2^e+2×2^eDefault matrix.

2^e×2^e+2Default matrix and 2^e+2×2^eDefault matrix passes through 2^e+2Default matrix is derived, wherein e=1, and 2,3；

Derivation formula is：WQ ' (i, j)=WQ (i*m, j*n)；

Wherein WQ (i*m, j*n) is the coefficient value of (i*m, j*n) position in default matrix, and WQ ' (i, j) is in default matrix The coefficient value of (i, j) position, m is the ratio of the width of the same objective matrix derived of width of the square matrices for deriving, n It is the ratio of the height of the same objective matrix derived of height of square matrices for deriving.

During as e=1, default matrix is 2 × 8 default matrixs and 8 × 2 default matrixs, now can use 8 × 8 default matrixs Derived；

During e=2, default matrix is 4 × 16 default matrixs and 16 × 4 default matrixs, now can be using 16 × 16 acquiescence squares Battle array is derived；

During e=3, default matrix is 8 × 32 default matrixs, and 32 × 8 default matrixs now can use 32 × 32 default matrixs Derived.

In one embodiment, default weighted quantization matrices collection can also include default weighted quantization matrices concentration arbitrarily One default matrix is multiplied by the new default matrix that A obtains, and wherein A is a constant.

As shown in Fig. 2 the video compression coding-decoding method based on weight quantization of the application, a kind of its embodiment, also Decoding process can be included, decoding process includes：

Step 202：It is predicted according to information of forecasting, obtain prediction block；

Step 204：From code stream, parsing obtains quantifying block；

Step 206：According to the value quantifying block, it is confirmed whether to parse the other information of write code stream, and rebuilds attribute block.

In one embodiment, the decoding process of the application can also include：

Step 208：If described quantization block is full null matrix, described prediction block is the described attribute block rebuild, and terminates solution Code process.Otherwise, 210 are gone to step；

Step 210：Default matrix is obtained according to the number information of default matrix, initial to each frequency using default matrix Inverse quantization step-length be weighted obtain weight inverse quantization Step matrix；

Step 208：Carry out inverse quantization using weighting inverse quantization Step matrix to quantifying block, obtain inverse transform block；

Step 210：Inverse transformation is carried out to inverse transform block and obtains residual block, residual block is added with prediction block and obtains reconstruction Attribute block.

The video compression coding-decoding method based on weight quantization of the application, in an encoding process, if current attribute block amount Coefficient matrix after change is full null matrix, then no longer weighted quantization matrices number information is stored in code stream.The application is in decoding During, if the coefficient matrix that current attribute block parses after the quantization obtaining is full null matrix, no longer parse weight quantization Matrix is numbered.

Embodiment two：

Embodiment two is a kind of concrete application example of the video compression coding-decoding method based on weight quantization of the application.This In embodiment, the default following matrix of weighted quantization matrices centralized procurement, numbering 0,1,2 respectively

8 × 8 default matrixs 1：

8 × 8 default matrixs 2：

8 × 8 default matrixs 3：

In the present embodiment, cataloged procedure obtains image to be encoded first, treats coded image and extracts attribute component, and will belong to Property component is divided into some 16 × 16 attribute block.

First by Predicting Technique, attribute block is predicted, such as infra-frame prediction, inter prediction etc., according to the life that predicts the outcome Become sub- residual block, sub- residual block is probably 8 × 8,4 × 16,16 × 4,16 × 16 block；

Sub- residual block obtains transform block to be quantified accordingly through conversion；

According to picture characteristics and transform block size from 8 × 8 default matrix, 1,8 × 8 default matrix 2,8 × 8 default matrix 3 Select suitable default matrix, and the default matrix using according to needed for selected default matrix and block size are derived：

If transform block is 8 × 8 pieces, directly using 8 × 8 default matrixs chosen；

If transform block is 16 × 16 pieces, derived according to below equation：

Here actual be exactly one common average, be equivalent to handleWithThe value of this four points is added together, then divided by 4.Also linear interpolation can be done, for formula

In a, tetra- parameters of b, c, d, if i, j is 2 multiple, then a be 1, b, c, d be 0；If j be 2 times Number, i is not 2 multiple, then a, b are 0.5, c, d is 0, if i is 2 multiple, j is not 2 multiple, then a, c are 0.5, b, d For 0；Otherwise it is simply that a, b, c, d are 0.25)

If transform block is 4 × 16, derived according to below equation：

WQ ' (i, j)=WQ (i*4, j)

If transform block is 16 × 4, derived according to below equation：

WQ ' (i, j)=WQ (i, j*4)

Calculate the weight quantization step-length of different frequent points according to the default matrix derived, obtain weight quantization Step matrix；

Using weight quantization Step matrix, transform block is quantified, obtain quantifying block；

If the quantization block non-full zero block obtaining, weighted quantization matrices are numbered write code stream.

In the present embodiment, decoding process is decoded to each 16 × 16 attribute block, and obtains final reconstruction image.

It is predicted according to information of forecasting first, obtain prediction block；

Parse and obtain to quantify block from code stream, quantify block and may include many height quantization blocks；

If son quantify block be stand growth model, prediction block directly as rebuild sub- attribute block, otherwise：

Parsing default matrix numbering further；

Obtain the default matrix of this attribute block selection according to default matrix numbering, and pushed away according to the size calculating of sub- quantization block Lead to obtain weighted quantization matrices：

It it is 8 × 8 pieces as fruit quantifies block, then directly using 8 × 8 default matrixs chosen；Be 16 as fruit quantifies block × 16 pieces, then derived according to below equation：

It is 4 × 16 as fruit quantifies block, then derived according to below equation：

WQ ' (i, j)=WQ (i*4, j)

It is 16 × 4 as fruit quantifies block, then derived according to below equation：

WQ ' (i, j)=WQ (i, j*4)

Weighting inverse quantization Step matrix is calculated according to the weight quantization block derived；

Carry out inverse quantization using weighting inverse quantization Step matrix to quantifying block, obtain inverse transform block；

Inverse transform block is obtained with inverse transformation and obtains residual block, residual block is added, with prediction block, the attribute block obtaining rebuilding.

Embodiment three：

As Fig. 3 such as shows, the video-compression codecs based on weight quantization of the application, a kind of its embodiment, including Encoder, encoder includes：

Division module, for extracting the attribute component of image to be encoded, attribute component is divided into some attribute blocks；

Predictive transformation module, for being predicted obtaining residual block to attribute block, enters line translation to residual block and obtains residual error The conversion coefficient of each frequency in block；

Computing module, chooses a default matrix for concentrating from default weighted quantization matrices, using default matrix to frequency The initial quantization step of point is weighted and obtains weight quantization Step matrix；

Quantization modules, for being quantified to conversion coefficient using weight quantization Step matrix, are obtained quantifying block, will quantify Block writes code stream, and according to the described value quantifying block, confirms to write the other information of described code stream.

In one embodiment, quantization modules are additionally operable to when quantifying block for full null matrix, then not by default matrix Number information writes code stream, otherwise, the number information of default matrix is write code stream.

The video-compression codecs based on weight quantization of the application, concentrate in default weighted quantization matrices and include 4 × 4 Default matrix 1 and/or 4 × 4 default matrixs 2 and/or 4 × 4 default matrixs 3 and/or 8 × 8 default matrix 1 and/or 8 × 8 acquiescence Matrix 2 and/or 8 × 8 default matrixs 3.

4 × 4 default matrixs 1:

4 × 4 default matrixs 2：

4 × 4 default matrixs 3：

8 × 8 default matrixs 1：

8 × 8 default matrixs 2：

8 × 8 default matrixs 3：

The video-compression codecs based on weight quantization of the application, default weighted quantization matrices collection can also include： 16 × 16 default matrixs and/or 32 × 32 default matrixs.

Computing module includes default matrix computing unit, and default matrix computing unit is used for by formula

Calculate acquisition 16 × 16 default matrix and 32 × 32 default matrixs.

Wherein WQ₈(x, y) is the value of (x, y) position in 8 × 8 default matrixs, and WQ (i, j) is the weighting matrix being derived by In (i, j) position value, k be objective matrix width divided by 8 multiple, a, b, c and d are constant, and b, c, d are not all 0.

In one embodiment, default weighted quantization matrices collection also includes：2^e×2^e+2Default matrix and/or 2^e+2×2^e Default matrix；

Default matrix computing unit, is additionally operable to by formula W Q ' (i, j)=WQ (i*m, j*n) derivation acquisition 2^e×2^e+2Silent Recognize matrix and 2^e+2×2^eDefault matrix, wherein e=1,2,3.Wherein WQ (i*m, j*n) is (i*m, j*n) position in default matrix Coefficient value, WQ ' (i, j) be default matrix in (i, j) position coefficient value, m be for derive square matrices width With the ratio of the width of objective matrix derived, n is the same objective matrix derived of height of the square matrices for deriving The ratio of height.

During as e=1, default matrix is 2 × 8 default matrixs and 8 × 2 default matrixs, now can use 8 × 8 default matrixs Derived；

During e=2, default matrix is 4 × 16 default matrixs and 16 × 4 default matrixs, now can be using 16 × 16 acquiescence squares Battle array is derived；

During e=3, default matrix is 8 × 32 default matrixs, and 32 × 8 default matrixs now can use 32 × 32 default matrixs Derived.

In one embodiment, default matrix computing unit can be additionally used in will default weighted quantization matrices concentrate any one Individual default matrix is multiplied by A, obtains a new default matrix, and this new default matrix is added default weighted quantization matrices collection In, wherein A is a constant.

As shown in figure 4, the video-compression codecs based on weight quantization of the application, a kind of its embodiment, also may be used To include decoder.Decoder includes prediction module, parsing module：

Prediction module, for being predicted according to information of forecasting, obtains prediction block, inverse quantization computing module, inverse quantization mould Block and reconstruction module.

Parsing module, obtains quantifying the number information of block and default matrix for parsing from code stream；

Rebuild module, for according to the value quantifying block, being confirmed whether to parse the other information of write code stream, and rebuilding attribute Block.

In one embodiment, decoder also includes inverse quantization computing module and inverse quantization module,

Rebuild module, be additionally operable to judge to quantify whether block is full null matrix, if quantization block is full null matrix, by prediction block As the attribute block rebuild；

Otherwise, inverse quantization computing module, obtains default matrix for the number information according to default matrix, using acquiescence square The battle array inverse quantization step-length initial to each frequency is weighted and obtains weighting inverse quantization Step matrix；

Inverse quantization module, for carrying out inverse quantization using weighting inverse quantization Step matrix to quantifying block, obtains inverse transform block；

Rebuild module, obtain residual block for inverse transformation is carried out to inverse transform block, residual block is added with prediction block and obtains The attribute block rebuild.

Above content is further description the application made with reference to specific embodiment it is impossible to assert this Shen Being embodied as please is confined to these explanations.For the application person of an ordinary skill in the technical field, do not taking off On the premise of the application design, some simple deduction or replace can also be made.

Claims (12)

1. a kind of video compression coding-decoding method based on weight quantization, including cataloged procedure it is characterised in that described encoded Journey includes：

Extract the attribute component of image to be encoded, described attribute component is divided into some attribute blocks；

Described attribute block is predicted obtaining residual block, described residual block is entered with line translation and obtains each frequency in described residual block Conversion coefficient；

Concentrate from default weighted quantization matrices and choose a default matrix, using the primary quantity to described frequency for the described default matrix Change step-length is weighted and obtains weight quantization Step matrix；

Using described weight quantization Step matrix, described conversion coefficient is quantified, obtain quantifying block, described quantization block is write Enter code stream, and according to the described value quantifying block, confirm to write the other information of described code stream；

If described quantization block is full null matrix, the number information of described default matrix is not write described code stream, otherwise, by institute The number information stating default matrix writes described code stream；

Described default weighted quantization matrices collection includes：

4 × 4 default matrixs 1:

$\begin{array}{cccc}16& 16& 16& 17\\ 16& 16& 17& 18\\ 16& 17& 19& 20\\ 18& 19& 21& 24\end{array}$

And/or 4 × 4 default matrixs 2：

$\begin{array}{cccc}16& 16& 16& 18\\ 16& 17& 19& 22\\ 17& 19& 23& 27\\ 21& 24& 29& 36\end{array}$

And/or 4 × 4 default matrixs 3：

$\begin{array}{cccc}16& 16& 17& 22\\ 17& 18& 24& 30\\ 19& 24& 32& 42\\ 28& 36& 48& 64\end{array}$

And/or 8 × 8 default matrixs 1：

$\begin{array}{cccccccc}16& 16& 16& 16& 17& 17& 18& 19\\ 16& 16& 16& 17& 18& 19& 21& 23\\ 16& 16& 17& 18& 19& 20& 22& 25\\ 16& 17& 18& 20& 21& 23& 25& 28\\ 17& 18& 20& 21& 23& 26& 28& 32\\ 19& 20& 21& 23& 26& 29& 33& 38\\ 24& 25& 26& 29& 31& 35& 41& 47\\ 26& 27& 29& 32& 38& 43& 48& 54\end{array}$

And/or 8 × 8 default matrixs 2：

$\begin{array}{cccccccc}16& 16& 16& 17& 18& 19& 22& 24\end{array}$

$\begin{array}{cccccccc}16& 16& 17& 18& 21& 24& 28& 35\\ 16& 17& 18& 22& 24& 27& 32& 39\\ 17& 19& 22& 26& 29& 33& 38& 46\\ 19& 22& 26& 29& 35& 41& 48& 57\\ 24& 27& 30& 35& 41& 50& 60& 72\\ 36& 39& 43& 49& 56& 66& 81& 96\\ 43& 45& 49& 57& 72& 86& 101& 115\end{array}$

And/or 8 × 8 default matrixs 3：

$\begin{array}{cccccccc}16& 16& 16& 18& 20& 24& 30& 36\\ 16& 17& 18& 20& 27& 35& 46& 62\\ 16& 18& 22& 31& 36& 43& 55& 72\\ 18& 24& 31& 41& 48& 57& 70& 89\\ 24& 31& 41& 48& 62& 76& 93& 116\\ 36& 42& 50& 62& 76& 97& 122& 152\\ 64& 72& 80& 96& 112& 136& 174& 210\\ 81& 85& 96& 116& 152& 186& 220& 255\end{array}.$

2. the video compression coding-decoding method based on weight quantization as claimed in claim 1 is it is characterised in that described presetting adds Power quantization matrix collection also includes：16 × 16 default matrixs and/or 32 × 32 default matrixs；

Described 16 × 16 default matrixs and described 32 × 32 default matrixs pass through formulaCalculating obtains ?；

Wherein WQ₈(x, y) is the value of (x, y) position in 8 × 8 default matrixs, WQ (i, j) be in the weighting matrix being derived by (i, J) value of position, k be objective matrix width divided by 8 multiple, a, b, c and d be constant.

3. the video compression coding-decoding method based on weight quantization as claimed in claim 2 is it is characterised in that described presetting adds Power quantization matrix collection also includes：2^e×2^e+2Default matrix and/or 2^e+2×2^eDefault matrix, described 2^e×2^e+2Default matrix and institute State 2^e+2×2^eDefault matrix uses 2^e+2Default matrix is derived, wherein e=1, and 2,3；

Derivation formula is：WQ ' (i, j)=WQ (i*m, j*n)；

Wherein WQ (i*m, j*n) is 2^e+2The coefficient value of (i*m, j*n) position in default matrix, WQ ' (i, j) is 2^e×2^e+2Acquiescence Matrix or 2^e+2×2^eThe coefficient value of (i, j) position in default matrix, m is the same derivation of width of the square matrices for deriving The width of objective matrix ratio, n be for derive square matrices height with derive objective matrix height Ratio.

4. the video compression coding-decoding method based on weight quantization as claimed any one in claims 1 to 3, its feature exists In described default weighted quantization matrices collection also includes concentrating any one default matrix to be multiplied by described default weighted quantization matrices The new default matrix that A obtains, wherein A is a constant.

5. the video compression coding-decoding method based on weight quantization as claimed any one in claims 1 to 3, its feature exists In also including decoding process, described decoding process includes：

It is predicted according to information of forecasting, obtain prediction block；

From described code stream, parsing obtains described quantization block；

According to the described value quantifying block, it is confirmed whether to parse the other information writing described code stream, and rebuilds attribute block.

6. the video compression coding-decoding method based on weight quantization as claimed in claim 5 was it is characterised in that described decoded Journey also includes：

If described quantization block is full null matrix, described prediction block is the described attribute block rebuild；

Otherwise, described default matrix is obtained according to the described number information of described default matrix, using described default matrix to every The initial inverse quantization step-length of individual frequency is weighted and obtains weighting inverse quantization Step matrix；

Using described weighting inverse quantization Step matrix, inverse quantization is carried out to described quantization block, obtain inverse transform block；

Inverse transformation is carried out to described inverse transform block and obtains described residual block, described residual block is added with described prediction block and obtains weight The described attribute block built.

7. a kind of video-compression codecs based on weight quantization are it is characterised in that include encoder；Described volume Code device includes：

Division module, for extracting the attribute component of image to be encoded, described attribute component is divided into some attribute blocks；

Predictive transformation module, for being predicted obtaining residual block to described attribute block, enters line translation to described residual block and obtains The conversion coefficient of each frequency in described residual block；

Computing module, chooses a default matrix for concentrating from default weighted quantization matrices, using described default matrix to institute The initial quantization step stating frequency is weighted and obtains weight quantization Step matrix；

Quantization modules, for quantifying to described conversion coefficient using described weight quantization Step matrix, obtain quantifying block, will Described quantization block write code stream, and according to the described value quantifying block, confirm to write the other information of described code stream；

Described quantization modules are additionally operable to when described quantization block is full null matrix, then not write the number information of described default matrix Enter described code stream, otherwise, the number information of described default matrix is write described code stream；

Described default weighted quantization matrices collection includes：

4 × 4 default matrixs 1:

$\begin{array}{cccc}16& 16& 16& 17\\ 16& 16& 17& 18\\ 16& 17& 19& 20\end{array}$

$\begin{array}{cccc}18& 19& 21& 24\end{array}$

And/or 4 × 4 default matrixs 2：

$\begin{array}{cccc}16& 16& 16& 18\\ 16& 17& 19& 22\\ 17& 19& 23& 27\\ 21& 24& 29& 36\end{array}$

And/or 4 × 4 default matrixs 3：

$\begin{array}{cccc}16& 16& 17& 22\\ 17& 18& 24& 30\\ 19& 24& 32& 42\\ 28& 36& 48& 64\end{array}$

And/or 8 × 8 default matrixs 1：

$\begin{array}{cccccccc}16& 16& 16& 16& 17& 17& 18& 19\\ 16& 16& 16& 17& 18& 19& 21& 23\\ 16& 16& 17& 18& 19& 20& 22& 25\\ 16& 17& 18& 20& 21& 23& 25& 28\\ 17& 18& 20& 21& 23& 26& 28& 32\\ 19& 20& 21& 23& 26& 29& 33& 38\\ 24& 25& 26& 29& 31& 35& 41& 47\\ 26& 27& 29& 32& 38& 43& 48& 54\end{array}$

And/or 8 × 8 default matrixs 2：

$\begin{array}{cccccccc}16& 16& 16& 17& 18& 19& 22& 24\\ 16& 16& 17& 18& 21& 24& 28& 35\\ 16& 17& 18& 22& 24& 27& 32& 39\\ 17& 19& 22& 26& 29& 33& 38& 46\\ 19& 22& 26& 29& 35& 41& 48& 57\\ 24& 27& 30& 35& 41& 50& 60& 72\\ 36& 39& 43& 49& 56& 66& 81& 96\\ 43& 45& 49& 57& 72& 86& 101& 115\end{array}$

And/or 8 × 8 default matrixs 3：

$\begin{array}{cccccccc}16& 16& 16& 18& 20& 24& 30& 36\\ 16& 17& 18& 20& 27& 35& 46& 62\\ 16& 18& 22& 31& 36& 43& 55& 72\\ 18& 24& 31& 41& 48& 57& 70& 89\\ 24& 31& 41& 48& 62& 76& 93& 116\\ 36& 42& 50& 62& 76& 97& 122& 152\\ 64& 72& 80& 96& 112& 136& 174& 210\\ 81& 85& 96& 116& 152& 186& 220& 255\end{array}.$

8. the video-compression codecs based on weight quantization as claimed in claim 7 are it is characterised in that described default weighting Quantization matrix collection also includes：16 × 16 default matrixs and/or 32 × 32 default matrixs；

Described computing module includes default matrix computing unit, and described default matrix computing unit is used for by formula

$WQ(i,j)=a*{\mathrm{WQ}}_8(\frac{i}{k},\frac{j}{k})+b*{\mathrm{WQ}}_8(\frac{i}{k}+1,\frac{j}{k})+c*{\mathrm{WQ}}_8(\frac{i}{k},\frac{j}{k}+1)+d*{\mathrm{WQ}}_8(\frac{i}{k}+1,\frac{j}{k}+1)$

Calculate and obtain described 16 × 16 default matrixs and described 32 × 32 default matrixs；

Wherein WQ₈(x, y) is the value of (x, y) position in 8 × 8 default matrixs, WQ (i, j) be in the weighting matrix being derived by (i, J) value of position, k be objective matrix width divided by 8 multiple, a, b, c and d be constant.

9. the video-compression codecs based on weight quantization as claimed in claim 8 are it is characterised in that described default weighting Quantization matrix collection also includes：2^e×2^e+2Default matrix and/or 2^e+2×2^eDefault matrix；

Described default matrix computing unit, is additionally operable to by formula W Q ' (i, j)=WQ (i*m, j*n) derivation acquisition described 2^e×2^{e +2}Default matrix and described 2^e+2×2^eDefault matrix, wherein e=1,2,3；Wherein WQ (i*m, j*n) is 2^e+2(i* in default matrix M, j*n) position coefficient value, WQ ' (i, j) be 2^e×2^e+2Default matrix or 2^e+2×2^eThe coefficient of (i, j) position in default matrix Value, m is the ratio of the width of the same objective matrix derived of width of the square matrices for deriving, and n is for just deriving The height of square matrix is with the ratio of the height of objective matrix derived.

10. the video-compression codecs based on weight quantization as any one of claim 7 to 9 it is characterised in that Described default matrix computing unit is additionally operable to concentrate any one default matrix to be multiplied by A described default weighted quantization matrices, obtains To a new default matrix, and described new default matrix is added described default weighted quantization matrices to concentrate, wherein A is one Individual constant.

11. video-compression codecs based on weight quantization as any one of claim 7 to 9 it is characterised in that Also include decoder, described decoder includes：

Prediction module, for being predicted according to information of forecasting, obtains prediction block；

Parsing module, obtains described quantization block for parsing from described code stream；

Rebuild module, for according to the described value quantifying block, being confirmed whether to parse the other information writing described code stream, and rebuilding Attribute block.

12. video-compression codecs based on weight quantization as claimed in claim 11 are it is characterised in that described decoder Also include inverse quantization computing module and inverse quantization module,

Described reconstruction module, is additionally operable to judge whether described quantization block is full null matrix, if described quantization block is full null matrix, Using described prediction block as the described attribute block rebuild；

Otherwise,

Described inverse quantization computing module, obtains described default matrix for the described number information according to described default matrix, makes It is weighted with the described default matrix inverse quantization step-length initial to each frequency and obtain weighting inverse quantization Step matrix；

Described inverse quantization module, for carrying out inverse quantization using described weighting inverse quantization Step matrix to described quantization block, obtains Inverse transform block；

Described reconstruction module, obtains described residual block for carrying out inverse transformation to described inverse transform block, by described residual block and institute State prediction block and be added the attribute block obtaining rebuilding.