WO2012041055A1

WO2012041055A1 - Method and system for adjusting quantization parameters of frame level and macroblock level

Info

Publication number: WO2012041055A1
Application number: PCT/CN2011/072822
Authority: WO
Inventors: 舒倩
Original assignee: 深圳市融创天下科技股份有限公司
Priority date: 2010-09-27
Filing date: 2011-04-14
Publication date: 2012-04-05
Also published as: CN102420980B; CN102420980A

Abstract

A method and system for adjusting frame level and macroblock level quantization parameters are provided by the present invention. The method comprises the following steps: determining an adjustment range for a frame level quantization parameter on the basis of dynamic and static contrast information of an image; making further adjustments on the basis of a current bit or a QP output from an RC. The method enables effective control in moving a real bit rate closer to a target bit rate. Furthermore, a new method for adjusting the macroblock level quantization parameter is also provided by the present invention. The method enables self-adaptive adjustment of a current macroblock quantization parameter by using static and dynamic contrast information of the current macroblock and that of the current frame and by using the frame level quantization parameter, thereby improving the rate distortion performance of an encoder while maintaining the stability of the code rate volatility.

Description

Frame layer and macroblock layer quantization parameter adjustment method and system

Technical field

The present invention relates to the field of video coding, and in particular, to a frame layer and macroblock layer quantization parameter adjustment method and system.

Background technique

One of the core problems in video coding is RC (rate control), which is used to determine various coding modes, such as macroblock division, motion vector, quantization parameters, and so on. Among them, the quantization causes distortion, which makes the selection of quantization and the selection of quantization parameters especially important. At the same time, in video coding, the first coded frame is reconstructed and used as the reference frame of the subsequent coded frame, so that the quality of the subsequent coded frame is closely related to the quality of the reference frame.

The prior art generally adopts a fixed-period frame layer quantization parameter adjustment mode. Since it does not consider the characteristics of the image, it cannot be adaptively adjusted online, thereby restricting its performance.

Summary of the invention

The object of the embodiments of the present invention is to provide a frame layer quantization parameter adjustment method, which aims to solve the problem that the frame layer quantization parameter adjustment mode in the prior art cannot be adaptively adjusted online, the rate distortion performance of the encoder is not high, and the code rate fluctuation performance is unstable. The problem.

The frame layer quantization parameter adjustment method includes the following steps: extracting static contrast information and dynamic contrast information of a current frame; determining a frame according to the static contrast information of the current frame, the dynamic contrast information, and a quantization parameter of a code rate control output. a final adjustment function of the layer quantization parameter; controlling the output quantization parameter and the final adjustment function according to the code rate of the current frame to obtain a new quantization parameter of the frame layer. In the frame layer quantization parameter adjustment method of the present invention, the static contrast information of the current frame is:

Contrast static _t - meaniContrast static _nt \ Mb _nt € region)

Where t is the time serial number of the image in the video sequence, n is the label of the macroblock in the image, Contrast-static is the static contrast information of the _t- frame image; ^ContmSt - blood ^tic is the static contrast of the n-th macroblock of the t-th frame Information, ^M is the nth macroblock of the tth frame, ^g''. "For the image area for statistics, mean is the mean function.

In the frame layer quantization parameter adjustment method according to the present invention,

The CwtraW —static _nt = sum (δ (Contrast _ijt ) \ p _ijt e M _nt ) , f(x)>Thres _f(x)

) is the corresponding decision threshold _t

^Cont and t - ^'s' ^ "is a t-th frame the n-th macroblock static contrast information, C a _s t _Lj, the contrast of the pixel ^% t frame image i-th row j-th column, ^A is _Βί in. The pixel of the i-th row and the j-th column of the t-th frame image;

^ Contrast _i j _τ - max( fe(5 * Contrast — 3 * Contrast 5 , ))/15

Contrast, . is the contrast of the pixel ^Ρί small ' in the i-th row and the j-th column of the t-th frame image, Contrast-3; the small _tk is the three-contiguous k- _th mode of the pixel, and Q^r W_5 _i is the pixel ^Α .'· ''The five adjacency kth mode.

In the frame layer quantization parameter adjustment method according to the present invention, the Q^raw-3, specifically: the three adjacency first mode: ^Co coffee ^st - ³ , ', i = + Pi-u, _t + Pi-

Pi, "The three adjacencies of the second mode: ^CQntmst - ³ "'" = -ij, + Pi- _W + / i, j + Ut Pi, "three adjacencies of the third mode: ^Contrast - ³ ", ', 3 = Pi-w + + A i ₊ +l,j+l,t

Pu, _t three adjacent fourth pattern: ^Co coffee ^st - ³ ", ', 4 = Pw + A ₊₁ , / ₊ i, t + P _{i + l} j, _t Pi, "The third mode of the third adjacency: ^Contrast - ³ , ', 5 = Pi ₊ w + + PM - t Pi," the third adjacency sixth mode: ^{Co st} - ³ , ', 6 = Pi ₊ y , t + _+U -i,t + P l,t Pi," the three adjacent modes of the seventh mode: ^Contrast - = P^w + Pi -ut + Pi-Pi," the three adjacency eighth mode: ^Contrast - ³ ,',8 = P' -ut + Pi-uj- t +

The Contrast _5 _{ij tk is} specifically:

Pij, the five adjacency modes of _t : Contrast _5 _iJ = sum _iJt - Contrast _ > _iJ where _匪 ij, _t = Pi-u- t + Pi-i ,, + Pi- _lJ+ t + Pij-ut + Pi _J+ ut + Pi _+lJ -ut + PMJ, + P _i+W , _t represents the time serial number of the image in the video sequence, ij represents the row and column label in the current pixel image, and the pixel is the current macroblock. The pixel of the i-th row and the j-th column of the t-frame image.

In the frame layer quantization parameter adjustment method of the present invention, the current frame dynamic contrast information is:

Contrast moving _τ - mean Contrast _movingj _{η τ} ) | Mb _nt € region,] - \or2)

Contrast _m _OV ing _t is the dynamic contrast information of the _t-th frame image; ^M is the n-th macroblock of the _t-th frame, for the statistical image area, mean is the mean function; Contrast _ movingj _n , _t means different The dynamic contrast information of the nth macroblock of the tth frame obtained by the method, j=l, indicating the dynamic contrast of the nth macroblock M of the tth frame when the motion vector mv information is not used, and j=2, indicating use The dynamic contrast of the nth macroblock „ of the tth frame in the motion vector mv information;

Contrast _ moving\ _nt = sum(S(Contrast _i j _t ) \ p _{t Jt} ^ Mb _nt && p _{t } t} = p _{; t} - p _; , Contrast, , _t = max(abs(5 * Contrast 3 , -3 * Contrast 5 , , , ))/15

^J ' l ≤ k ≤ & ~ ^J " ~ ' ⁷ ''

Contrast; "The contrast of the pixel ^Α ·, for the i-th row and the j-th column of the t-frame image, Contrast is the three-contiguous k- _th mode of the pixel A, Q^ _ra w_5 _i , which is the five-contiguous k- _th pattern of the pixel';

Contrast moving! _nt - sum(S(mvx _nt + mvy _nt )) | mv _nt € Mb _nt ) Wherein the foot represents the horizontal component of the motion vector of the nth macroblock of the tth frame, and represents the vertical component of the motion vector of the nth macroblock of the tth frame.

In the frame layer quantization parameter adjustment method of the present invention, the determining the final adjustment function of the frame layer quantization parameter specifically includes the following steps: determining a picture group fluctuation period according to a frame rate and an image sequence number of the image;

Determining an initial adjustment function of the frame layer quantization parameter according to the image group fluctuation period;

Obtaining a frame layer quantization parameter correction function according to the static contrast information of the current frame, the dynamic contrast information, and the initial adjustment function;

And adjusting the frame layer quantization parameter correction function according to a current bit or code rate control output quantization parameter to obtain a final adjustment function of the frame layer quantization parameter.

In the frame layer quantization parameter adjustment method of the present invention, the image group fluctuation period is:

Period = f (id _t , fps) where per iod is the image group fluctuation period, which is the coding sequence number of the t-th frame, fp _S , and the image group fluctuation period function determined by the frame rate plus the coding order ^id of the image;

Λ ijd _t , fps) = k _t , Thres^ < fps < Thres _i & 8dd _t e Ω? Where is a statistical constant determined according to different frame rates and a set of coding sequence numbers, 73⁄4r% - ^ΤΤ»^. is the corresponding threshold value, Ω, which is a category set of coding sequence numbers;

The initial adjustment function of the frame layer quantization parameter is:

dQP _xt = f _x (id period )

^(^y) _; o

Where / ! y) satisfies, d _x , A (x + ky, y) = Ά y) , ^d Q ^p is the initial adjustment function of the quantization parameter of the t-th frame, k is the period constant, 3 is the partial derivative; The layer quantization parameter correction function is:

dQP _lt = f ₂ (Contrast _ static _t , Contrast _ moving _t , dQP _lt ) Represents the t-frame quantization parameter correction function; / ₂₍ , _z) satisfies ^a/ 2 ^(x , ^y , ^z) <o ,

Where dx is the partial derivative,

a ^m j is a division set of different fluctuation periods and coding sequence numbers, and is a statistical adjustment constant determined according to the above division set; the final adjustment function of the frame layer quantization parameter is:

dQPr = MdQP _2l , QP _t , bit _t ) / χ, satisfy ^ ^ > 0 ^ ^ > 0, _Λ ^, _ζ) = _{Λ (} χ, υ ^ _Ω function, its ox ΟΖ

Medium W is the final adjustment function of the t-frame quantization parameter, and Bitt is the current bit value.

Another object of the embodiments of the present invention is to provide a frame layer quantization parameter adjustment system, where the system includes: a current frame static and dynamic contrast information, and is used to extract static contrast information and dynamic contrast information of a current frame. ;

a final adjustment function obtaining module of the frame layer quantization parameter, configured to determine a final adjustment function of the frame layer quantization parameter according to the static contrast information of the current frame, the dynamic contrast information, and the quantization parameter of the code rate control output;

The frame layer new quantization parameter obtaining module is configured to obtain a new quantization parameter of the frame layer according to the sum of the quantization parameter outputted by the code rate control of the current frame and the final adjustment function.

In the frame layer quantization parameter adjustment system of the present invention, the final adjustment function acquisition module of the frame layer quantization parameter includes: an image group fluctuation period determining module, configured to determine image group fluctuation according to a frame rate and an image sequence number of the image. Cycle

a frame layer quantization parameter initial adjustment function determining module, configured to determine an initial adjustment function of the frame layer quantization parameter according to the image group fluctuation period; a frame layer quantization parameter correction function determining module, configured to acquire a frame layer quantization parameter correction function according to the static contrast information of the current frame, the dynamic contrast information, and the initial adjustment function; a frame layer quantization parameter final adjustment function determining module, And configured to adjust the frame layer quantization parameter correction function according to a current bit or a code rate control output quantization parameter to obtain a final adjustment function of the frame layer quantization parameter.

Another object of the embodiments of the present invention is to provide a macroblock layer quantization parameter adjustment method, the method comprising: extracting dynamic and static contrast information of a current macroblock and a current frame; and determining, by a frame layer, a current macroblock and a current macroblock, The dynamic and static contrast information of the frame determines the quantization parameters of the current macroblock. In the macroblock layer quantization parameter adjustment method of the present invention, the quantization parameter of the current macroblock is:

^^S^ Weight _mb where is the quantization parameter of the current macroblock, ^ is the quantization parameter of the original t-frame, ^eight _{mb movmg} , which is the weight of the static contrast information and the dynamic contrast information respectively; f _mb static (Contrast - static _Nt , Contrast _ static _t ) ^ is a quantization based on static contrast information d Λχ, Ύ _{; 0} df _mb x, y)

Parameter adjustment function, , ^ ^ ^ ;

Fm _b ― mo _Ving (Contrast _ movingj^ Contrast _ moving _t ) _? is quantization based on dynamic contrast information

3f _mb — _mg ^x , y < ₀ df _mb (x, y)

■0

The parameter adjustment function, , ^ .

Another object of the embodiments of the present invention is to provide a macroblock layer quantization parameter adjustment system, where the system includes: a dynamic block and a dynamic contrast information of a current macroblock and a frame, which are used to extract the current macroblock and the current frame. Static contrast information and dynamic contrast information; The quantization parameter obtaining module of the current macroblock is configured to determine, by the frame layer quantization parameter, the current macroblock, the static contrast information of the current frame, and the dynamic contrast information, the quantization parameter of the current macroblock.

The beneficial effects of the invention:

The embodiment of the present invention determines the adjustment range of the frame layer quantization parameter by using the dynamic and static contrast information of the image, and further adjusts the QP condition of the current bit or RC output, which can effectively control the actual bit rate direction. The target bit rate is close. By using the frame layer quantization parameter adjustment strategy of the present invention, not only the rate distortion performance but also the code rate fluctuation of the encoder output code stream can be made more gradual, thereby improving the network transmission performance, and also enabling the encoder to actually output the bit rate. Closer to the encoding target bit rate.

In addition, the embodiment of the present invention adaptively adjusts the quantization parameter of the current macroblock by using the current macroblock, the static and dynamic contrast information of the current frame, and the frame layer quantization parameter, thereby maintaining the code rate while improving the rate distortion performance of the encoder. Stability of wave performance.

DRAWINGS

1 is a flowchart of a method for adjusting a frame layer quantization parameter according to a preferred embodiment of the present invention; FIG. 2 is a flowchart of a method for acquiring a final adjustment function of a frame layer quantization parameter in the method of FIG. 1. FIG. 3 is a preferred embodiment of the present invention. FIG. 4 is a detailed structural diagram of a final adjustment function of a frame layer quantization parameter in the system of FIG. 3; FIG. 5 is a flow chart of a macroblock layer quantization parameter adjustment method according to a preferred embodiment of the present invention; Figure 6 is a block diagram showing the structure of a macroblock layer quantization parameter adjustment system in accordance with a preferred embodiment of the present invention.

Detailed ways

The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. For the convenience of description, only the parts related to the embodiments of the present invention are shown. It is understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. The embodiment of the invention adopts a new frame layer quantization parameter adjustment method. On one hand, the dynamic amplitude and static contrast information of the image are used to determine the adjustment range of the frame layer quantization parameter, and on the other hand, the current bit or the QP of the RC output is used. Further, this can effectively control the actual bit rate to be close to the target bit rate; in addition, the embodiment of the present invention also proposes a new macroblock layer self-quantization parameter adjustment method, which utilizes the current macro block, the static and dynamic of the current frame. The contrast information and the frame layer quantization parameter adaptively adjust the quantization parameter of the current macroblock, thereby improving the rate distortion performance of the encoder while maintaining the stability of the rate fluctuation performance.

FIG. 1 is a flowchart of a method for adjusting a frame layer quantization parameter according to an embodiment of the present invention, where the method includes:

S101: extract static contrast information and dynamic contrast information of the current frame;

Contrast static _t - mean{Contrast static _nt | Mb _nt € region)

Contrast moving _t = meaniContrast moving] _nt ) | Mb _nt € region, ] - \or2) where t is the time sequence number of the image in the video sequence, n is the label of the macroblock in the image, and the static contrast information of the Contrast- static frame image ; ^C - ^St - blood ^tic is the static contrast information of the nth macroblock of the tth frame; ^M is the nth macroblock of the tth frame; _{C_t_m} _OV i _ngl is the dynamic contrast information of the image of the tth frame; Contrast _ moving] _nt indicates the dynamic contrast information of the nth macroblock of the tth frame obtained by different methods, and j=l indicates the dynamic contrast of the nth macroblock „ of the tth frame when the motion vector mv information is not used. When j=2, it indicates the dynamic contrast of the nth macroblock „ of the tth frame when the motion vector mv information is used; for the image area to be statistically, mean is the mean function.

(1) The method for extracting the static contrast information of the current frame is as follows:

Let the current macroblock be Mb _n , _t , be the pixel in Mb _n , _t , t denote the time serial number of the image in the video sequence, n is the label of the macroblock in the image, and i and j respectively represent the current pixel image The row and column labels, M is the nth macroblock of the tth frame, that is, the current macroblock, and is the pixel of the i th row and the jth column of the tth frame image in M, that is, the current pixel. The ^Ρ '' of the 8 pixels adjacent to the following list:

Sumsum;, = Pi- _W + Pi-u, _t + Pi-i ₊ i, t + Pi -i, t + P _{i +} i, t + P _i+W + P _i+ i , _t + P _{i+ W}

The three adjacencies are:

^Pi '" three adjacency first mode: Contrast j,t, ―

+ Pi-\,j,t + Pi-\,j+\,t

Α·, _Λί 's three adjacencies second mode: Contrast ~ Pi + Pi + Pi

Α·, _Λί 's three adjacencies, the third mode: Contrast /V ~ Pi + Pi + Pi

^Pi '" three adjacencies fourth mode: Contrast /V ~

^Ρ ''Three adjacencies fifth mode: Contrast j,t ~

^Pi '" three adjacency sixth mode: Contrast j,t ~

+ Pi + Pi,jl,t

^Ρ ''Three adjacency seventh mode: Contrast /V ~ Pi + Pi,jl,t + Pi-\,j-\,t

^Pi '" three adjacency eighth mode: Contrast ~ Pi, jl, t + Pi +

Five adjacencies are:

The fifth neighboring first mode: Contrast _ .5 i , t, k ' = sum _ijt - Contrast — 3 _ijtk The contrast of the current pixel ^A is:

Contrast i'j = max(abs(5 * Contrast _ 3ι _ί]ίΙ( -3 * Contrast _ 5 _{i J tk} )) / 15 C _on tm _S t _{i jt} ^ _{tPixel of} the i-th row and j-th column of the frame image That is, the contrast of the current pixel ', Contrast _ 3 _{ijt lc} is the three-contiguous k- ^th mode of the current pixel ^{Ρ ί} '", and Contrast _5 _{ijt lc} is the five-contiguous k- ^th mode of the current pixel. The static contrast of the current macroblock M (the nth macroblock of the tth frame) is:

Contrast static _nt - sum (δ (Contrast _i j _t ) p _{i } t} e M _nt )

1 , f(x) > Thres _f{x)

S(f(x)) =

Where ⁰ , ^dse , and ⁷⁷ "^/w is the corresponding decision threshold

Static contrast information t n-th frame macroblocks ^Contrast, C a _sti, j, _{t t} frame image i-th row pixel column j i.e. the current pixel ^contrast, ρ ", 'as Mb _n, _t frame t The pixel of the i-th row and the j-th column of the image is also the current pixel.

Then the current frame static contrast is:

Contrast _ static _t = mean(Contrast _ static _nt | Mb _nt e region)

(2) The current frame dynamic contrast information extraction method is as follows:

Embodiments of the present invention provide two methods for acquiring the contrast of dynamic information of a video image macroblock.

Method 1: Without the motion vector mv information, the dynamic contrast of the current macroblock Mb _nt is:

Contrast _ moving\ _nt = sum{d {Contrast _tjt ) | p _ijt e Mb _nt && p _ijt = p _ijt — p This method does not require motion search and is computationally intensive.

Method 2: Using the motion vector mv information, the dynamic contrast of the current macroblock Mb _nt is:

Contrast _ moving! _nt = sum(S(mvx _nt + mvy _nt )) | mv _nt e Mb _nt ) The foot represents the horizontal component of the motion vector of the nth macroblock of the tth frame, representing the nth macroblock of the tth frame The vertical component of the motion vector.

The method is based on motion vectors obtained by motion search with high precision.

Based on the above two methods for obtaining the dynamic information contrast of the video image macroblock, the current frame dynamic contrast is obtained as follows:

Contrast _ moving! = mean(Contrast _ moving) _nt ) | Mb _nt e region, j = lor 2) Contrast _ movingj _nt represents the dynamic contrast information of the nth macroblock of the tth frame obtained by different methods; j = l, when the motion vector mv information is not used, the dynamics of the nth macroblock „ of the tth frame Contrast, when j=2, indicates the dynamic contrast of the nth macroblock „ of the tth frame when the motion vector mv information is used.

S102: Determine a final adjustment function ^rf W of the frame layer quantization parameter QP from the dynamic static contrast information of the current frame image and the RC output. Where RC is the rate control, QP is the quantization parameter, 'is the quantization parameter of the original t-frame, and is the adjustment function of the quantization parameter of the t-th frame. 2 is a flowchart of a method for acquiring a final adjustment function of a frame layer quantization parameter, where the method for obtaining a final adjustment function of a frame layer quantization parameter includes the following steps:

S1021: First, the image group fluctuation period is determined by the frame rate plus the code number of the image.

Period = f ₀ (id _t , fps) where period is the image group fluctuation period, which is the coding sequence number of the t-th frame, : birthday, fp _S , frame rate plus the image coding order ^id 'determined image group fluctuation period function . f ₀ (id _t , fps) can be selected as follows:

Λ « , fP ^s ) = , Thres^ < fps < Thres _i 8c 8cid _t e Ω? Wherein is a statistical constant determined according to different frame rates and a set of coding sequence numbers, ΤΤ^, ΤΤ»^. is a corresponding threshold value, Ω, which is a category set of coding sequence numbers;

S 1022: determining an initial adjustment function of the frame layer quantization parameter by the image group fluctuation period

dQP _xt = f _x (id period ) (x,y) . ₀

Where /xj) satisfies, dx , f (x + k , y) = f (x, y) _? d ^QP , ' is the initial adjustment function for the quantization ^mask of % t frame, k is the period constant, and 3 is the partial derivative. S1023: Acquire a frame layer quantization parameter correction function by combining dynamic and static contrast information of the current frame image and the initial adjustment function.

dQP _{2 t} = f ₂ (Contrast _ static _t , Contrast _ moving _t , dQP _lt ) represents the t-frame quantization parameter correction function; / ₂ (x, _y , _z ) optionally satisfies ^a/ 2 ( ^X , ^y , ^Z )<0 , = ^, _Ζε Ω, for example, satisfying the above conditional cylinder

The linear function of a single / ₂ ^, _z ) = -x _{+ y +} where 3 is the partial derivative, il ^Tid j is the set of partitions of different fluctuation periods and coding numbers, and kkj is the statistical adjustment constant determined according to the above division set;

S1024: Adjust the frame layer quantization parameter correction function ^dQP " according to the current bit or the QP condition of the RC output, and obtain the frame layer QP final adjustment function ^^.

dQPr = MdQP _2l , QP _t , bit _t )

/ ₃ ^, ₂₎ Optional satisfies ^ ^〉0^ ^〉0, M _X , y, _z) = _Mx , kkk _m , _z) , ye^ _m , for example

OX oz satisfies the above condition single-segment function / ₃ ( , y, z) = x + kkk _m + (z > Thres _buffer Ία'.-b) , where is the final adjustment function of the _t- frame quantization parameter, Bitt is the current The bit value, Ω is a set of partitions of different quantization parameters β, „ is a statistical adjustment constant determined according to the quantization parameter set, Th _reSbuffer is a buffer _overflow limit, ^^ is a constant, _2> 73⁄4 means that if z is greater than the adjustment threshold 77 ^ , 贝'] ζ = α, No Bay' jz = _6.

S103: Obtain a new quantization parameter of the frame layer ^

QP _t ^∞w = QP _t + dQP; ^av FIG. 3 is a schematic structural diagram of a frame layer quantization parameter adjustment system according to an embodiment of the present invention, where the system includes: a current frame static and dynamic contrast information. Used to extract static contrast information and dynamic contrast information of the current frame;

The frame layer new quantization parameter obtaining module is configured to obtain a new quantization parameter of the frame layer according to the sum of the quantization parameter outputted by the code rate control of the current frame and the final adjustment function. FIG. 4 is a schematic structural diagram of a frame layer quantization parameter final adjustment function acquiring module in the system of FIG. 3, where the frame layer quantization parameter final adjustment function obtaining module includes: an image group fluctuation period determining module, configured to use a frame rate according to a frame rate And a coding sequence number of the image, determining a picture group fluctuation period; a frame layer quantization parameter initial adjustment function determining module, configured to determine a frame layer quantization parameter initial adjustment function according to the image group fluctuation period; a frame layer quantization parameter correction function determining module, configured to The static contrast information of the current frame, the dynamic contrast information, and the initial adjustment function acquire a frame layer quantization parameter correction function; a frame layer quantization parameter final adjustment function determining module for controlling quantization of the output according to a current bit or code rate In the case of the parameter, the frame layer quantization parameter correction function is adjusted to obtain a final adjustment function of the frame layer quantization parameter. The frame layer quantization parameter adjustment method of the embodiment of the invention can not only improve the rate distortion performance of the encoder, but also can make the code rate fluctuation of the encoder more gradual, improve the network transmission performance of the code stream, and can also make The actual output bit rate of the encoder is closer to the encoding target bit rate.

FIG. 5 is a flowchart of a macroblock layer quantization parameter adjustment method according to an embodiment of the present invention.

The current macroblock layer quantization parameters are determined by various modes such as energy determination method based on AC coefficient, macroblock layer rate control determination method, and visual characteristic adjustment method. The method based on the AC coefficient energy calculates the cylinder list and has good preservation characteristics for high frequency detail information. In the embodiment of the present invention, the static and dynamic contrast information of the macroblock, the static and dynamic contrast information of the frame, and the frame layer quantization parameter are combined to determine the quantization parameter of the current macroblock. Among them, the static and dynamic contrast information are used to protect the high frequency information of the motion macroblock and the static macroblock, respectively.

The macroblock layer quantization parameter adjustment method includes the following steps:

S501: extracting static contrast information and dynamic contrast information of the current macroblock and the current frame; static contrast of the current macroblock ^Co- ^st , blood, dynamic contrast of the current macroblock Contrast— moving! _nt ( _{j = 1 or 2} ) and the static contrast of the current frame and the dynamic contrast of the current frame. The method of acquiring Contrast moving _t has been described in detail above and will not be described here.

S502: Determine a quantization parameter ^{QP of the} current macroblock by the frame layer quantization parameter and the current macroblock, the static contrast information of the current frame, and the dynamic contrast information.

Weight _mb

Where ^ is the quantization parameter of the original t-frame, which is the weight of the static contrast and dynamic contrast information, respectively;

(Contrast — static _nt , Contrast —static _t , is the amount based on static contrast information ^匕

^L (x,y) _{; 0} df _{mb static} (x, y)

Parameter adjustment function, , ^ ^ ^ ;

(Contrast _ movingj^ Contrast _ moving _t ) _? is quantization based on dynamic contrast information

3f _mb — _mg ^x , y < ₀ df _mb (x, y)

■0

Parameter adjustment function, , ^

FIG. 6 is a schematic structural diagram of a macroblock layer quantization parameter adjustment system according to an embodiment of the present invention, where the system includes: a dynamic and static contrast information extraction module of a current macroblock and a frame, configured to extract a current macroblock and a current frame. Static contrast information and dynamic contrast information; a quantization parameter acquisition module of the current macroblock, configured to determine a quantization parameter of the current macroblock from the frame layer quantization parameter and the current macroblock, the static contrast information of the current frame, and the dynamic contrast information.

The macroblock layer quantization parameter adjustment method of the embodiment of the present invention can maintain the stability of the rate fluctuation performance while improving the rate distortion performance of the encoder. The above is only the preferred embodiment of the present invention, and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection of the present invention. Within the scope.

Claims

Rights request

A method for adjusting a frame layer quantization parameter, the method comprising the steps of: extracting static contrast information and dynamic contrast information of a current frame;

Determining a final adjustment function of the frame layer quantization parameter according to the static contrast information of the current frame, the dynamic contrast information, and the quantization parameter of the code rate control output;

The new quantization parameter of the frame layer is obtained by controlling the sum of the output quantization parameter and the final adjustment function according to the code rate of the current frame.

2. The frame layer quantization parameter adjustment method according to claim 1, wherein:

The static contrast information of the current frame is:

Contrast _ static _t - niecm{Contrast _ static _nt | Mb, _lt G region)

Where t is the time sequence number of the image in the video sequence, n is the label of the macroblock in the image, Contrast-static! is the static contrast information of the _t- frame image; ^{C t} and t - ^static «, the n-th macroblock of the % t frame Static contrast information, ^M is the nth macroblock of the tth frame, '. "For the image area for statistics, mean is the mean function.

3. The frame layer quantization parameter adjustment method according to claim 2, wherein:

The Contrast _ static _nt = sum {δ {Contrast _ijt ) \ p _ijt e M _nt )

1 , f(x) > Thres _f{x) else and -r T^hre

Where ^s w is the corresponding decision threshold;

^ContmSt t frame n-th static contrast information of the macro block, Q ^ ^ the contrast of pixels in the t th frame image of the i-th row j-th column, is _{Β ί} pixels in the t th frame image of the i-th row j-th column.;

Contrast _it - max( fe(5 * Contrast _k — 3 * Contrast 5 _i ^^ _k )) / 15 where Contrast _{i jt} ^% t frame image i-th row j-th column pixel ^Α ·.'·. 'Contrast, Contrast _ , _jtk is the three adjacent k- _{th pattern} of pixel A, Q^ _ra w_5^ is the pixel of five Adjacent to the kth mode.

The frame layer quantization parameter adjustment method according to claim 3, wherein the Contrast _3ι , .

Pi , _t Contrast

The first mode of the three adjacencies: - ^i,],t,\ = Pi-l,j,t + Pi-l ₊ l,t

Pi ,, Cont

The third adjacency second mode: rast -

The third mode of P's three adjacencies: Contrast

Pi, _t three adjacencies fourth mode: Contrast ―

'J,'A _ Pi +ΪΛ + + Pi ₊ l ,,

Pi , _t Contrast

The third mode of the third adjacency: ~ Pi + Pi+l,j,t + P _i+W

Pi , _t Contrast

The third mode of the third adjacency: _ Pi+l,j,l + P -l,t + Pi -l,t

Pi ,, Contrast

The seventh mode of the three adjacencies: ― Pi+l,j-l,t + Pi -l,t + Pi

Pi , _t Contrast

The third mode of the third adjacency: = Pi'j - l't + Pi-l, jl, t + Pi-u, _{t The} Contrast _5 _{ij tk is} specifically:

The fifth mode of u, the adjacent mode: Contrast > _ijtk = sum _{ijt -} Contrast _ _Μ where L = Pi-w + Pi-u, t + A - P -i, t + P _{iJ +} i, t + PM - + P _i+ u, _t + ΡΜ, Λ , _T denotes the time serial number of the image in the video sequence, i and j respectively represent the row and column labels in the current pixel image, and the pixel is the i-th row of the t-th frame image in the current macroblock. The pixel of the jth column.

5. The frame layer quantization parameter adjustment method according to claim 1, wherein:

The current frame dynamic contrast information is:

Contrast moving _t = meaniContrast moving _nt ) | Mb _nt e region, j = \or2)

Where Contrast—m _OV ing _t is the dynamic contrast information of the _t-th frame image; ^M is the n-th macroblock of the _t- frame, and for the statistical image region, mean is the mean value function; Contrast _ movingj _nt represents the dynamic contrast information of the nth macroblock of the tth frame obtained by different methods, and j=l indicates the dynamic contrast of the nth macroblock M of the tth frame when the motion vector mv information is not used, When j=2, it indicates the dynamic contrast of the nth macroblock „ of the tth frame when the motion vector mv information is used;

Contrast _ moving\ _nt = sum{d{Contrast _i j _t ) | p _{i ] t} e Mb _nt && p _{i ] t} = p _{i ] t} — p ', —] where Contrast i _{} t} = max(abs( 5 * Contrast _ 3 _; . _{; tk} -3 * Contrast _5 _{t jtk} ))l\5

Contrast _ijt is the contrast of the pixel P", ' of the i-th row and the j-th column of the t-frame image, Contrast > _u , _t , _k is the three adjacent k- _th pattern of the pixel P"'', and C ₀ "m^_5 is the pixel 'five adjacency kth mode;

Contrast moving! _nt - sum(d(mvx _nt + mvy _nt )) | mv _nt e Mb _nt )

Where mvx _Bi represents the horizontal component of the motion vector of the nth macroblock of the tth, and m represents the vertical component of the motion vector of the nth macroblock of the tth frame.

The frame layer quantization parameter adjustment method according to claim 1, wherein the determining the final adjustment function of the frame layer quantization parameter comprises the following steps:

Determining the image group fluctuation period according to the frame rate and the code serial number of the image;

The frame layer quantization parameter adjustment method according to claim 6, wherein the image group fluctuation period is:

Period = f. (id _t , fps)

Where period is the image group fluctuation period, which is the coding sequence number of the t-th frame, plus) Coding sequence plus the frame rate and image ^id 'image group determined as a function fluctuation _{_{cycle; f 0 (id t, fps}} ) = k i, Thres <fps <Thres t & 8dd t e Ω? Wherein is a statistical constant determined according to different frame rates and a set of coding sequence numbers, 73⁄4r% - ^77»^. is a corresponding threshold value, Ω, which is a category set of coding sequence numbers;

The initial adjustment function of the frame layer quantization parameter is:

dQP _xt = f _x (id period ) (x,y) _; o

Where / ! y) satisfies, d _x , A(x + ky, y) = Ά y) , ^d Q ^p is the initial adjustment function of the quantization parameter of the t-th frame, k is a period constant, and 3 is a partial derivative; The layer quantization parameter correction function is:

dQP _{2 t} = f ₂ (Contrast _ static _t , Contrast _ moving _t , dQP _lt ) represents the t-frame quantization parameter correction function; / _{2 (} ^ ₂₎ satisfies ^3/ ^, ^) < 0 , dx where 3 is Partial guidance,

dQPr = MdQP _2l , QP _t , bit _t ) / ₃ ^, _{2) a function} satisfying ^ ^ > ₀ ^ ^ > ₀ , / ₃₍ w, _{z) =} / _{3 z)} j _e Q ,

OX OZ

8. A frame layer quantization parameter adjustment system, wherein the system comprises: a current frame static and dynamic contrast information, and is used for extracting static contrast information and dynamic contrast information of a current frame; a final adjustment function obtaining module of the parameter, configured to determine a frame layer quantization parameter according to the static contrast information of the current frame, the dynamic contrast information, and a quantization parameter of a code rate control output Final adjustment function;

The frame layer quantization parameter adjustment system according to claim 8, wherein the final adjustment function acquisition module of the frame layer quantization parameter comprises: an image group fluctuation period determining module, configured to encode according to a frame rate and an image. Sequence number, determining the fluctuation period of the image group;

a frame layer quantization parameter initial adjustment function determining module, configured to determine an initial adjustment function of the frame layer quantization parameter according to the image group fluctuation period;

a frame layer quantization parameter correction function determining module, configured to acquire a frame layer quantization parameter correction function according to the static contrast information of the current frame, the dynamic contrast information, and the initial adjustment function; a frame layer quantization parameter final adjustment function determining module, And configured to adjust the frame layer quantization parameter correction function according to a current bit or a code rate control output quantization parameter to obtain a final adjustment function of the frame layer quantization parameter.

A method for adjusting a macroblock layer quantization parameter, the method comprising: extracting static contrast information and dynamic contrast information of a current macroblock and a current frame;

The quantization parameter of the current macroblock is determined by the frame layer quantization parameter and the current macroblock, the static contrast information of the current frame, and the dynamic contrast information.

The macroblock layer quantization parameter adjustment method according to claim 10, wherein the quantization parameter of the current macroblock is:

+ f ^^S^^ + f ^^S^"^

Where is the quantization parameter of the current macroblock, ^ is the quantization parameter of the original t-th frame, and ^ight _{mb movms is the weight} of the static contrast information and the dynamic contrast information, respectively; (Contrast — static _nt , Contrast —static _t ,, is the amount based on static contrast information ^匕

^L (x,y) _;0 df _{mb static} (x,y)

Parameter adjustment function, , ^ ^ ^ ;

Swelling (Contrast _ moving / ^ Contrast _ moving t)? Dynamic contrast information based on a quantization

( , y) _ΐ0 df _mb x,y)

Parameter adjustment function, ^ , ^ .

12. A macroblock layer quantization parameter adjustment system, wherein the system comprises: a dynamic and static contrast information extraction module of a current macroblock and a frame, configured to extract static contrast information and dynamics of a current macroblock and a current frame. Contrast information;

The quantization parameter obtaining module of the current macroblock is configured to determine a quantization parameter of the current macroblock by the frame layer quantization parameter and the current macroblock, the static contrast information of the current frame, and the dynamic contrast information.