CN103475934A - Video coding stream control method facing network live broadcast - Google Patents

Abstract

The invention discloses a video coding stream control method facing network live broadcast. The video coding stream control method includes the steps that parameters are set at a broadcast end; a time length for receiving video data of an initial buffering zone of the broadcast end is preset to be M, and the lowest receiving bandwidth of the broadcast end is preset to be W; initial coding is performed; after live broadcast coding is started, an analog buffering zone is set according to the preset M and the preset W; coding is performed, wherein in the coding process, the data size of a frame undergoing coding and the data size of a frame to be coded are periodically predicated; the state of a video sequence in the analog buffering zone is calculated, a quantization coefficient is adjusted so that the data volume of the video sequence in the analog buffering zone cannot be empty, and subsequent coding is continuously performed; coding data are distributed; when the coding step is executed, the coding data are directly or indirectly distributed to the broadcast end at the same time; video playing is performed; a decoder is arranged at the broadcast end, live broadcast begins to be received at any time point, when the volume of the received data reaches the time length M, a video starts to be played, and meanwhile a bandwidth receiving live broadcast streams and arranged at the broadcast end is no lower than W.

Description

The Video coding flow control method that a kind of network-oriented is live

Technical field

The invention relates to a kind of video flowing control techniques, particularly, is Video coding flow control method and the system live about a kind of network-oriented.

Background technology

As everyone knows, the form of Video service can be divided into download, program request, live etc. usually.Download, demand (telecommunication) service characteristics are that established video data is sent to and plays end playback, and live broadcast service is in the process of video data generation in real time, to playing end, send and playback.

Video service can be based on multiple transmission network, as traditional DVB-C, DMB-T etc., the Video service of Internet-based progressively increases in recent years, and the video player on the Internet is generally the embedded flash play control of PC browser, the control that ios operating system is built-in, open source software VLC etc.

Flow Control is the important step of Video coding, and its effect normally guarantees that video and transmission, the decoding model after coding is complementary, and the video quality had.Video Flow Control main models comprises HRD, VBV, ABR etc. at present.

HRD (Hypothetical Reference Decoder) model, i.e. hypothetical reference decoding buffer model comprises: encoded picture buffer (CPB), instantaneous decode procedure, decoded picture buffer (DPB) and an output.

The major parameter of HRD model is defined as follows table:

hrd_parameters(){	C	Descriptor
			cpb_cnt_minusl	0	ue(v)
bit_rate_scale	0	u(4)
			cpb_size_scale	0	u(4)
for(SchedSelIdx=0；SchedSelIdx＜=cpb_cnt_minus1；SchedSelIdx++){	?	?
			bit_rate_value_minus1[SchedSelIdx]	0	ue(v)
cpb_size_value_minus1[SchedSelIdx]	0	ue(v)
			cbr_flag[SchedSelIdx]	0	u(1)
}	?	?
			initial_cpb_removal_delay_length_minus1	0	u(5)
cpb_removal_delay_length_minusl	0	u(5)
			dpb_output_delay_length_minus1	0	u(5)
time_offset_length	0	u(5)
			}	?	?

Cpb_cnt_minus1+1 has specified the number of the CPB standard that can select, and codomain scope [0,31] is not specified cpb_cnt_minus1 in SPS, and giving tacit consent to this value is 0.Bit_rate_scale and bit_rate_value_minus1[SchedSeIdx] together with, the maximum input bit rate of having specified SchedSelIdx CPB.

Cpb_size_scale and cpb_size_value_minus1[SchedSelIdx] together with, specified the CPBSize of SchedSelIdx CPB.

HRD model definition AUn means n the addressed location (access unit, AU) in decoding order, from AU0.HRD requires according to the following rules code stream to be verified:

To the combination of each BitRate and CPBSize, all need to be verified.The parameter of HRD model, according to SchedSelIdx, initial_cpb_removal_delay[Sched SelIdx], BitRate[SchedSelIdx] and CpbSize[SchedSelIdx] decide.When the checking code stream is compatible, need meet the following conditions:

1. for AUn, n>0, definition Δ t _{g, 90}(n)=90000* (t _{r, n}(n)-t _af(n-1))

Initial_cpb_removal_delay[SchedSelIdx] need to meet:

1) if cbr_flag[SchedSelIdx] equal 0,

initial_cpb_removal_delay[SchedSelIdx]<=Ceil(Δt _g,90(n))。

2) otherwise (that is, cbr_flag[SchedSelIdx] equal at 1 o'clock), Floor (Δ t _{g, 90}(n))<=initial_cpb_removal_delay[SchedSelIdx]<=Ceil (Δ t _{g, 90}(n)).

Attention: when any one-frame video data removes from CPB, in CPB, remaining bit number depends on the buffering period SEI message that initialization HRD uses.Encoder must be considered this point, and guarantees no matter decoder selects any one buffering period SEI message to carry out initialization HRD, and the code stream of encoder output all can meet the constraints of all requirements.

2.CPB overflow means that the total bit number in CPB has surpassed CPBSize, CPB never should overflow.

3.CPB underflow means for some AUn, t _r,n(n)<t _af(n).When low_delay_hrd_flag equals 0, CPB never should underflow.

The subject matter of HRD model is:

(1) HRD need provide corresponding initial buffer requirement to each AU that encodes, and player need read this information and just start broadcasting after the actual data volume received with initial buffer requirement coupling from video flowing;

(2) decode time (remove time) that needs each frame/field of dynamic calculation when HRD encodes, decoder also needs to control by this value the decode time of each frame/field data simultaneously;

By above-mentioned 2 contents, can be found out, the HRD model has increased the complexity of decoding end, and at present the video players such as common flash, the ios in the Internet, VLC are not easy to realize the support to this encoding model.

For above-mentioned (2) point, also increased the complexity of coding side, simultaneously due in the HRD model to the adjustment of remove time numerical value, also can make the video playback fluency be affected, the larger program of motion amplitude particularly, as physical culture, amusement etc.

In practical business, VBV (Video Buffer Verifier, video buffer verifier) model is more adopted, its basic characteristics are that the initial buffer of player is fixing data volume (vbv buffer size): when VBV model regulation player starts, since the first frame data, by speed R (n)=Rmax, to VBV buffering padding data, filling process continues vbv_delay second, and vbv_delay defines in picture header, and now decode procedure starts.

But the main flow video player (flash player, VLC) of the Internet generally is based on the time of buffered data at present, it is the time difference of buffering area the first frame data and last frame data, rather than data volume (being how many bit), the initial buffer district is set, and the time tolerance of the video data of buffering area starts to play after reaching set point again.Adopt the user of VBV model when live broadcast service, if calculate the buffer time of buffering area by vbv buffer size/vbv bitrate, and be set to the initial transmission duration, also can't guarantee smoothness, because in practical business, the bandwidth of code stream is non-constant, so player cushions by the vbv buffer size/vbv bitrate time video sequence obtained, its data volume can't guarantee to equal vbv buffersize.If too strict restriction vbv buffer size, also can significantly reduce the possibility that card pauses, but also can significantly reduce coding quality.

HRD, the VBV model is mostly is towards traditional broadcasting and TV business, and in practical application, normal what adopt is the scheme such as asic chip to its decoding end.

ABR (the Average Bit Rate in addition of more use in Video coding, mean bit rate) Flow Control Model, the ABR model does not have strict definition, concrete methods of realizing is also different, normally strictly do not limit the fluctuation range of code stream, but the Flow Control target is to make the average bit rate of video draw close to given target bit rate, the ABR model is usually used in the non-program volume/transcoding towards transmission.

Summary of the invention

The invention provides the live Video coding flow control method of a kind of network-oriented, with this, guarantee the fluency that internet video is live.

To achieve these goals, the embodiment of the present invention provides a kind of network-oriented live Video coding flow control method, and this Video coding flow control method comprises:

Play end parameter setting steps: the duration that will play end initial buffer district receiving video data is preset as M, will play the minimum reception bandwidth of end and be preset as W;

Initialization codes step: after live coding starts, according to default M, W, set up simulated cushioned district.

Coding step: in cataloged procedure, periodically prediction is just in the size of data of coded frame and the size of data of frame to be encoded; Calculate the state of video sequence in described simulated cushioned district, and adjust the data volume non-NULL that quantization parameter makes video sequence in described simulated cushioned district, then proceed next code;

The coded data distributing step: when carrying out described coding step, direct or indirect distributes coded data to described broadcasting end;

Video playback step: at described broadcasting end structure decoder, can put at any time and start to receive live, after receiving data and reach the video data volume of duration M, start displaying video, make the bandwidth of the reception live TV stream of described broadcasting end be not less than W simultaneously.

In one embodiment, described method also comprises: in the initialization codes step, after the coding duration reaches described M, the video data of coding is put into to described simulated cushioned district, the data volume that does not enter the video sequence in simulated cushioned district is defined as to SN, now, SN=0.

In one embodiment, described method also comprises: in the initialization codes step, the average bandwidth that video is set is A, and take A as the target bandwidth, by the ABR model, is encoded.

In one embodiment, described method also comprises: in coding step, take A as the target bandwidth, by the ABR model, calculate quantization parameter, and then adjust quantization parameter according to simulated cushioned district numerical value.

In one embodiment, described method also comprises: in coding step, after each frame Video coding completes, prediction is just in the size of data of coded frame and the size of data of frame to be encoded; Calculate the state of video sequence in described simulated cushioned district, and adjust the data volume non-NULL that quantization parameter makes video sequence in described simulated cushioned district, then proceed next code;

Further, described coding step also comprises:

Step 1: after i frame coding, by live limited data transmission model, calculate the described simulated cushioned zone state under live condition; The limited data transmission model: video sequence is when transmission, process of transmitting should meet the transmission data volume and be no more than limit on transmission bandwidth, and current transmission video time stamp is poor with the initial video time stamp of transmission, deducts process of transmitting after the time, should not surpass a predetermined constant.

Described limited data transmission model is:

SN _i=SN _i-1+F _i，

If SN _i>D _i* W, SN _i-=D _i* W,

Otherwise SN _i=0;

Wherein, SN _ifor the data volume of the i frame video sequence that do not enter described simulated cushioned district, F _iit is the size of data of i frame of video; I the duration that frame of video is corresponding:

unit is second; The DTS timestamp of i frame of video is DTS _i.

Further, described prediction is just in the size of data of coded frame and the size of data of frame to be encoded; Calculate the state of video sequence in described simulated cushioned district, and adjust the data volume non-NULL that quantization parameter makes video sequence in described simulated cushioned district, comprising:

Step 2: set frame of video F to be encoded _i+a+1the value of intermediate parameters q of quantization parameter:

$q_{i+a+1}=q_{i+a}+q_{i+a}\&times;({\mathrm{\&sigma;}}_i-A*\underset{j=0}{\overset{i}{\mathrm{\&Sigma;}}}{D}_j)/(A*\max (1,\sqrt{\underset{j=0}{\overset{i}{\mathrm{\&Sigma;}}}{D}_j}))$

Calculate quantization parameter according to q: $Q{P}_{i+a+1}=12+6*\mathrm{lo}{g}_2\left(\frac{q_{i+a+1}}{0.85}\right)$

Wherein, the quantity that a is current encoder system parallel encoding frame, span is between 4-8 usually; B is the constant between 10～30, and the sequence of frames of video length of presentation code forecast analysis, be made as 20 in the present embodiment; The cumulative data size that has the i frame from the 1st frame to the i frame altogether is σ _i=σ _i-1+ F _i.

Step 3: based on q _i+1to q _i+a, the prediction of video residual error and Video coding historical information is just in the size of coded frame: F _i+1, F _i+2..., F _i+a;

Step 4: δ is set _i+a+1=0;

Wherein, δ means the degree that q adjusts.

Step 5: with q _i+a+1+ δ _i+a+1* Δ, for prediction q value, is predicted frame to be encoded: F _i+a+1, F _i+a+2..., F _i+a+b;

Step 6: according to limited data transmission model SN ' _i+a+b=SN _i+ F _i+1+ F _i+2+ ... + F _i+a+bif, ${{\mathrm{SN}}^{\&prime;}}_{i+a+b}>\frac{({\mathrm{DTS}}_{i+a+b}-{\mathrm{DTS}}_i)}{{10}^7}*W,$ ? ${{\mathrm{SN}}^{\&prime;}}_{i+a+b}-=\frac{({\mathrm{DTS}}_{i+a+b}-{\mathrm{DTS}}_i)}{{10}^7}*W;$ Otherwise SN ' _i+a+b=0; DB ' _i+a+b=M-SN ' _i+a+b/ W; As DB ' _i+a+b, recalculate quantization parameter at>0 o'clock

then carry out the coding of subsequent frame.

Wherein, i the frame of video duration of video sequence in rear simulation buffering area of having encoded is DB _i, the encoded duration of rear video sequence of i frame of video of prediction is DB ' _i.

In one embodiment, in described i frame coding step, as described DB ' _i+a+b<=0, δ _i+a+1+=1; If δ _i+a+1<Max Δ/Δ, carry out above-mentioned steps 5 and step 6 again.

Wherein, the Max Δ is pre-set constant, means each maximum q adjusting range, can be made as 20, and Δ is constant, represents the minimum value that q changes, and can be made as 0.05.

Further, if δ _i+a+1>=Max Δ/Δ, provide the underflow prompting of simulated cushioned district, simultaneously assignment δ _i+a+1=Max Δ/Δ, then recalculate quantization parameter

then carry out the coding of subsequent frame.

In one embodiment, in described coding step, as described DB ' _i+a+b<=0, define δ _i+a+1be mapped to DB ' _i+a+bfunction be: y=f (x); The y implication is δ _i+a+1during=x, the DB ' calculated by preceding method _i+a+bvalue, x is integer;

Definition P-1 rank multinomial: y=g (x)=t ₀+ t ₁* x+t ₂* x ²+ ... + t _p-1* x ^p-1, with g (x), carry out approach f (x); P, for the target order of prediction fitting function, is constant, 5<P<Max Δ/Δ.

Y=g (x) is calculated to sample point, and carry out matching: make x=0,1,2 ..., P-1, calculate y=y ₀, y ₁, y ₂..., y _p-1, at y=y ₀, y ₁, y ₂..., y _p-1computational process in, if y occurs for the first time _k>0, make δ _i+a+1=k, recalculate quantization parameter

then carry out the coding of subsequent frame;

Otherwise, if all y _k<=0, by x=0,1,2 ..., P-1, and corresponding y=y ₀, y ₁, y ₂..., y _p-1, as the reference sample point, carry out least square fitting, draw every coefficient of polynomial function: t ₀, t ₁, t ₂..., t _p-1;

Ask minimum integer x,

make inequality g (x)>0 set up;

By newton's down-hill method to equation g (x)=t ₀+ t ₁* x+t ₂* x ²+ ... + t _p-1* x ^p-1=0 carries out iterative, wherein take x=P as iterative initial value, and 1 is iteration precision, and concentrates at solution of equation, chooses the minimum value x that meets following 2 conditions _root:

1) in interval in;

2) to the first derivative g ' of g (x) (x)=t ₁+ t ₂* 2x+ ... + t _p-1* (p-1) * x ^p-2, and second dervative g " (x)=t of g (x) ₂* 2+t ₃* 3x+ ... + t _p-1* (p-1) * (p-2) * x ^p-3, meet: g ' (x _root) ≠ 0 or g " (x _root) 0;

If there is not the x satisfied condition _root, provide the buffer underflow prompting, simultaneously assignment δ _i+a+1=Max Δ/Δ, then recalculate quantization parameter ${\mathrm{QP}}_{i+a+1}=12+6*{\log }_2\left(\frac{q_{i+a+1}+{\mathrm{\&delta;}}_{i+a+1}*\mathrm{\&Delta;}}{0.85}\right),$ Then carry out the coding of subsequent frame.

Further, if there is x _root:

If g ' is (x _root) 0, make δ _i+a+1=INT (x _root)+1; If g ' is (x _root)<0, make δ _i+a+1=INT (x _root); If g ' is (x _root)=0, g " (x _root) 0, make δ _i+a+1=INT (x _root);

Recalculate quantization parameter ${\mathrm{QP}}_{i+a+1}=12+6*{\log }_2\left(\frac{q_{i+a+1}+{\mathrm{\&delta;}}_{i+a+1}*\mathrm{\&Delta;}}{0.85}\right),$ Then carry out the coding of subsequent frame;

Wherein, INT (x) means to be not more than the maximum integer of x.

In one embodiment, described method also comprises, in the video playback step, to the flash player, by the member property NetConnection::bufferTime that the NetConnection class is set, specify the flash play control before starting broadcasting, need the video data of buffer memory M duration, specifically comprise:

varnc：NetConnection=new?NetConnection；

var?flvbuftime：Number=stage.loaderInfo.parameters[″buftime″]；

Nc.bufferTime=flvbuftime, flvbuftime is M.

The beneficial effect of the embodiment of the present invention is:

The Flow Control target is combined with average bandwidth, guarantee to play the level and smooth of end, also guaranteed that average discharge is no more than expection, can control the flow cost of distribution link.

Because the present invention has defined the data volume SN of the video sequence that does not enter simulated cushioned district, need not record the specifying information of every frame video, this specifying information is including but not limited to the timestamp of every frame video, byte number, frame type etc., only need the not byte number in the simulated cushioned district of transmission of record, can calculate the duration of video in simulated cushioned district.

By Video coding flow control method of the present invention, coding side, without adopting HRD and VBV model, not changing the decode time of playing end, can realize playing the level and smooth broadcasting of end.

Adopt the limited data transmission model with adaptive surface to video real-time live broadcast business, as adopt traditional mode, in the live process of reality, because the coding side data are to produce in real time, therefore can cause sending the data volume deficiency, and then produce deviation with simulation process, and the risk that causes actual appearance card to pause.

Video coding flow control method of the present invention has been simplified flow control algorithm, particularly, without carrying out calculating again of decode time, is convenient to the realization of coding side, plays the video of end simultaneously and also can play by video time stamp, has guaranteed result of broadcast.

Video coding flow control method of the present invention adopt the time span of buffers video data but not data volume as the initial buffer foundation, be easy to the realization of the Internet main flow player, and can when guaranteeing fluency, need not adopt too strict Flow Control restriction, and then guaranteed video quality.

The present invention can guarantee to play the random time point of end after live coding starts and start receiver, video stream, all can guarantee smooth playing.

Flow Control Model of the present invention is not considered the overflow restriction that traditional Flow Control model is considered, has simplified flow control procedure.

The accompanying drawing explanation

In order to be illustrated more clearly in the embodiment of the present invention or technical scheme of the prior art, in below describing embodiment, the accompanying drawing of required use is briefly described, apparently, accompanying drawing in the following describes is only some embodiments of the present invention, for those of ordinary skills, under the prerequisite of not paying creative work, can also obtain according to these accompanying drawings other accompanying drawing.

Fig. 1 is the live Video coding flow control method flow chart of embodiment of the present invention network-oriented;

Fig. 2 is the method flow diagram after the every frame Video coding of the embodiment of the present invention.

Embodiment

Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is clearly and completely described, obviously, described embodiment is only the present invention's part embodiment, rather than whole embodiment.Embodiment based in the present invention, those of ordinary skills, not making under the creative work prerequisite the every other embodiment obtained, belong to the scope of protection of the invention.

As shown in Figure 1, the embodiment of the present invention provides a kind of network-oriented live Video coding flow control method, and this Video coding flow control method comprises:

S101: play end parameter setting steps: the duration that will play end initial buffer district receiving video data is preset as M, will play the minimum reception bandwidth of end and be preset as W;

S102: initialization codes step: after live coding starts, according to default M, W, set up simulated cushioned district;

S103: coding step: in cataloged procedure, periodically prediction is just in the size of data of coded frame and the size of data of frame to be encoded; Calculate the state of video sequence in described simulated cushioned district, and adjust the data volume non-NULL that quantization parameter makes video sequence in described simulated cushioned district, then proceed next code;

S104: coded data distributing step: when carrying out coding step, direct or indirect distributes coded data to described broadcasting end;

S105: video playback step: at described broadcasting end structure decoder, after receiving data and reach the video data volume of duration M, start displaying video, make the bandwidth of the reception live TV stream of described broadcasting end be not less than W simultaneously.

Flow process by Fig. 1 is known, in the embodiment of the present invention, duration and the broadcasting of playing end initial buffer district receiving video data is held to minimum reception bandwidth, video frame rate is preset as respectively to M and W; Set up simulated cushioned district according to default M, W; Periodically the prediction state of video sequence in the size of data of the size of data of coded frame and frame to be encoded and in calculating simulated cushioned district just, adjust the data volume non-NULL that quantization parameter makes video sequence in described simulated cushioned district; To described broadcasting end distribution coded data, and playing end structure decoder when carrying out described coding step, after the video data volume that receives data and reach duration M, starting displaying video, and the bandwidth that makes to play the reception live TV stream of end is not less than W.Play the end simple structure, need not rely on HRD, VBV model, need not change the decode time of playing end, Flow Control Model need not be considered the overflow restriction, also need not record the specifying information of every frame video while setting up simulated cushioned district, the byte number in the simulated cushioned district that only needs record not send, can realize the live smoothness of internet video.

When S101 specifically implements, before live beginning, following two parameters need to be set: play the duration M of end initial buffer district receiving video data and play the minimum reception bandwidth W of end.Preferably, coding average bandwidth A can also be set.

When S102 specifically implements, after live coding starts, set up simulated cushioned district according to default M, W, and after the coding duration reaches described M, the video data of coding is put into to described simulated cushioned district B, and now in B, video sequence duration DB is M, and the data volume that does not enter the video sequence in simulated cushioned district is defined as to SN, now, SN=0.Because this step has defined the data volume SN of the video sequence that does not enter simulated cushioned district, need not record the specifying information of every frame video, this specifying information is including but not limited to the timestamp of every frame video, byte number, frame type etc., only need the not byte number in the simulated cushioned district of transmission of record, can calculate the duration of video in simulated cushioned district.

In this step, initial simulated cushioned district is set to have the video sequence of M duration, but not a fixing the video data volume.In this step, the average bandwidth that video can also be set is A, and take A as the target bandwidth, by the ABR model, is encoded.

After S102 completes, can enter tri-steps of S103, S104 and S105, can find out from the flow chart of Fig. 1, tri-steps of S103, S104 and S105 do not have sequencing, and this three steps below are described respectively:

In S103, after each frame Video coding completes, need prediction just in the size of data of coded frame and the size of data of frame to be encoded, calculate the state of video sequence in simulated cushioned district, and adjust the data volume non-NULL that quantization parameter makes video sequence in described simulated cushioned district, then just carry out next code.

Concrete, while adjusting quantization parameter, take A as the target bandwidth, first calculate quantization parameter by the ABR model, and then adjust quantization parameter according to simulated cushioned district numerical value.The Flow Control target is combined with average bandwidth, guarantee smoothly also to have guaranteed that average discharge is no more than expection, can control the flow cost of distribution link.

As shown in Figure 2, when S103 specifically implements, coding step can also comprise:

S201: after i frame coding, calculate the simulated cushioned zone state under live condition by live limited data transmission model, the limited data transmission model: video sequence is when transmission, process of transmitting should meet the transmission data volume and be no more than limit on transmission bandwidth, and current transmission video time stamp is poor with the initial video time stamp of transmission, deduct process of transmitting after the time, should not surpass a predetermined constant.This limited data transmission model is:

SN _i=SN _i-1+F _i，

If SN _i>D _i* W, SN _i-=D _i* W,

Otherwise SN _i=0;

Wherein, SN _ifor the data volume of the i frame video sequence that do not enter described simulated cushioned district, F _iit is the size of data of i frame of video; I the duration that frame of video is corresponding:

unit is second; The DTS timestamp of i frame of video is DTS _i.

The present invention is owing to adopting the limited data transmission model, can adaptive surface to video real-time live broadcast business.As adopt traditional mode,, in the live process of reality, because the coding side data are real-time generations, therefore can produce the data volume deficiency, and then produce deviation with simulation process, and the risk that causes actual appearance card to pause.Therefore, the introducing of limited data transmission model has solved the above-mentioned problems in the prior art.

After S201, prediction is just in the size of data of coded frame and the size of data of frame to be encoded, as shown in Figure 2, prediction is just in the size of data of coded frame and the size of data of frame to be encoded, calculate the state of video sequence in described simulated cushioned district, and adjust the data volume non-NULL that quantization parameter makes video sequence in described simulated cushioned district, specifically comprise S202, S203, S204 and S205:

S202: need to press abr mode, set frame of video F to be encoded _i+a+1the q value:

$q_{i+a+1}=q_{i+a}+q_{i+a}\&times;({\mathrm{\&sigma;}}_i-A*{\mathrm{\&Sigma;}}_{j=0}^i{D}_j)/(A*\max (1,\sqrt{{\mathrm{\&Sigma;}}_{j=0}^i{D}_j}));$

Calculate quantization parameter according to q: ${\mathrm{QP}}_{i+a+1}=12+6*\mathrm{lo}{g}_2\left(\frac{q_{i+a+1}}{0.85}\right).$

Wherein, the quantity that a is current encoder system parallel encoding frame, span is between 4-8 usually; B is the constant between 10～30, and the sequence of frames of video length of presentation code forecast analysis, be made as 20 in the present embodiment; The cumulative data size that has the i frame from the 1st frame to the i frame altogether is σ _i=σ _i-1+ F _i.

S203: based on q _i+1to q _i+a, the prediction of video residual error and Video coding historical information is just in the size of coded frame: F _i+1, F _i+2..., F _i+a;

S204: δ is set _i+a+1=0.

Wherein, δ means the degree that q adjusts.

S205: with q _i+a+1+ δ _i+a+1* Δ, for prediction q value, is predicted frame to be encoded: F _i+a+1, F _i+a+2..., F _i+a+b.

As shown in Figure 2, in S103, calculate the duration of video sequence in simulated cushioned district, specifically comprise:

S206: show that according to S203 and S205 i+1 is after the prediction size of these frames of i+a+b, according to limited data transmission model SN' _i+a+b=SN _i+ F _i+1+ F _i+2+ ... + F _i+a+bif,

?

otherwise SN' _i+a+b=0; DB ' _i+a+b=M-SN' _i+a+b/ W.

As shown in Figure 2, in S103, after calculating the duration of video sequence in simulated cushioned district, also comprise:

S207: judgement DB ' _i+a+bwhether>0 set up, and if so, enters S208; If DB ' _i+a+b<=0,, δ _i+a+1+=1; Judgement δ _i+a+1with the size (S210) of Max Δ/Δ, if δ _i+a+1<Max Δ/Δ, with q _i+a+1+ δ _i+a+1* Δ, for prediction q value, is predicted frame to be encoded: F _i+a+1, F _i+a+2..., F _i+a+b; According to limited data transmission model SN ' _i+a+b=SN _i+ F _i+1+ F _i+2+ ... + F _i+a+bif, ${{\mathrm{SN}}^{\&prime;}}_{i+a+b}>\frac{({\mathrm{DTS}}_{i+a+b}-{\mathrm{DTS}}_i)}{{10}^7}*W,$ ? ${{\mathrm{SN}}^{\&prime;}}_{i+a+b}-=\frac{({\mathrm{DTS}}_{i+a+b}-{\mathrm{DTS}}_i)}{{10}^7}*W;$ Otherwise SN ' _i+a+b=0; DB ' _i+a+b=M-SN ' _i+a+b/ W; As DB ' _i+a+b, recalculate quantization parameter at>0 o'clock

then carry out the coding of subsequent frame.

If δ _i+a+1>=Max Δ/Δ, enter S209, then jumps to S208.Wherein, the Max Δ is pre-set constant, means each maximum q adjusting range, and Δ is constant, represents the minimum value that q changes.

S208: the quantization parameter that recalculates frame of video ${\mathrm{QP}}_{i+a+1}=12+6*{\log }_2\left(\frac{q_{i+a+1}+{\mathrm{\&delta;}}_{i+a+1}*\mathrm{\&Delta;}}{0.85}\right),$ Then start the actual coding of a new frame, repeat afterwards S103.

S209: provide the buffer underflow prompting, mean that there is the possibility of pausing in this code stream, simultaneously assignment δ _i+a+1=Max Δ/Δ.

S210: judgement δ _i+a+1size with Max Δ/Δ.

In a preferred embodiment, in order to reduce amount of calculation, S207, S209 can be amended as follows:

As DB ' _i+a+b>0, enter S208;

As described DB ' _i+a+b<=0, define δ _i+a+1be mapped to DB ' _i+a+bfunction be: y=f (x); The y implication is δ _i+a+1during=x, the DB ' that process S205 and S206 calculate _i+a+bvalue, wherein, the x implication is δ _i+a+1, x is integer.

From S205 and S206, when x is larger, prediction q value is larger, from coding criterion H.264, and the frame F to be encoded of prediction _i+a+1, F _i+a+2..., F _i+a+bjust less, SN' _i+a+balso just less, so DB ' _i+a+blarger, y is larger.That is to say that function y=f (x) is monotonically increasing function.

For reducing amount of calculation, rapid solving meets DB ' _i+a+b>0 minimum δ _i+a+1, definition P-1 rank multinomial: y=g (x)=t ₀+ t ₁* x+t ₂* x ²+ ... + t _p-1* x ^p-1, with g (x), carry out approach f (x);

To y=g (x), calculate sample point, and carry out matching: make x=0,1,2 ..., P-1, calculate y=y ₀, y ₁, y ₂..., y _p-1, at y=y ₀, y ₁, y ₂..., y _p-1computational process in, if y occurs for the first time _k>0, make δ _i+a+1=k, enter S208, recalculates quantization parameter

then carry out i+1 frame coding, wherein, the x implication is δ _i+a+1, x is integer, P, for the target order of prediction fitting function, is constant, 5<P<Max Δ/Δ.

Otherwise, if all y _k<=0, by x=0,1,2 ..., P-1, and corresponding y=y ₀, y ₁, y ₂..., y _p-1, as the reference sample point, carry out least square fitting, draw every coefficient of polynomial function: t ₀, t ₁, t ₂..., t _p-1;

For calculating target, find minimum δ _i+a+1, make DB ' _i+a+b>0, and meet

can mathematical description be:

Ask minimum integer x,

make inequality g (x)>0 set up;

After obtaining the approximating function g (x) of f (x), calculate target and further be reduced to: ask minimum integer x,

make inequality g (x)>0 set up.

The method solved according to the monobasic inequality of higher degree, the disaggregation space of inequality g (x)>0 is determined by disaggregation, sign of inequality direction and g (x) first derivative at the equattion root place of equation g (x)=0, therefore at first to g (x)=0 equation solution.

Then, by newton's down-hill method to equation g (x)=t ₀+ t ₁* x+t ₂* x ²+ ... + t _p-1* x ^p-1=0 carries out iterative (wherein take x=P as iterative initial value, 1 is iteration precision), and concentrates at solution of equation, chooses the minimum value x that meets following 2 conditions _root:

1) in interval

in;

2) to the first derivative g ' of g (x) (x)=t ₁+ t ₂* 2x+ ... + t _p-1* (p-1) * x ^p-2, and second dervative g " (x)=t of g (x) ₂* 2+t ₃* 3x+ ... + t _p-1* (p-1) * (p-2) * x ^p-3, meet: g ' (x _root) ≠ 0 or g " (x _root) 0;

If there is not the x satisfied condition _root, provide the buffer underflow prompting, simultaneously assignment δ _i+a+1=Max Δ/Δ, then enter S208 and recalculate quantization parameter

then carry out i+1 frame coding.

If there is the x satisfied condition _root:

If g ' is (x _root) 0, make δ _i+a+1=INT (x _root)+1; If g ' is (x _root)<0, make δ _i+a+1=INT (x _root); If g ' is (x _root)=0, g " (x _root) 0, make δ _i+a+1=INT (x _root); Then enter S208 and recalculate quantization parameter

then carry out i+1 frame coding; Wherein, INT (x) means to be not more than the maximum integer of x.

S104 carries out when S103 carries out, and the actual video data that coding is produced, to playing the end distribution, so that play the end displaying video.

The broadcasting end can receive live video stream to carry out when carrying out S103, S104, when S105 specifically implements, play end structure decoder, it is put at any time receives when live, should after reaching the video data volume of duration M, the reception data just start displaying video, player removes video data decoding by the DTS timestamp of video from buffering area, also should guarantee that the bandwidth of playing the reception live TV stream of holding is not less than fixed value W simultaneously.

S105, in the video playback step, can adopt the flash play control, to the flash player, with q _i+a+1+ δ _i+a+1* △, for prediction q value, predicts frame to be encoded: F _i+a+1, F _i+a+2..., F _i+a+bstep in, by the member property NetConnection::bufferTime that the NetConnection class is set, specify the flash play control before starting to play, need the video data of buffer memory M duration, specifically comprise:

By the M assignment in bufferTime:

varnc：NetConnection=new?NetConnection；

var?flvbuftime：Number=stage.loaderInfo.parameters[″buftime″]；

Nc.bufferTime=flvbuftime, flvbuftime is M.

By after as above step, can guarantee to play end at any time during the reception live video, all can guarantee the buffering area non-NULL, get final product smooth playing.

In addition, the present invention also possesses following technique effect:

By Video coding flow control method of the present invention, coding side, without adopting the HRD model, not changing the decode time of playing end, can realize playing the level and smooth broadcasting of end.

Video coding flow control method of the present invention has been simplified flow control algorithm, particularly, without carrying out calculating again of decode time, is convenient to the realization of coding side, plays the video of end simultaneously and also can play by video time stamp, has guaranteed result of broadcast.

Video coding flow control method of the present invention adopt the time span of buffers video data but not data volume as the initial buffer foundation, be easy to the realization of the Internet main flow player.

Flow Control Model of the present invention is not considered the overflow restriction, has simplified flow control procedure.

One of ordinary skill in the art will appreciate that all or part of step realized in above-described embodiment method can come the hardware that instruction is relevant to complete by program, this program can be stored in a computer read/write memory medium, such as ROM/RAM, magnetic disc, CD etc.

Above-described specific embodiment; purpose of the present invention, technical scheme and beneficial effect are further described; institute is understood that; the foregoing is only specific embodiments of the invention; the protection range be not intended to limit the present invention; within the spirit and principles in the present invention all, any modification of making, be equal to replacement, improvement etc., within all should being included in protection scope of the present invention.

Claims (14)

1. the Video coding flow control method that network-oriented is live, is characterized in that, described Video coding flow control method comprises:

Play end parameter setting steps: the duration that will play end initial buffer district receiving video data is preset as M, will play the minimum reception bandwidth of end and be preset as W;

Initialization codes step: after live coding starts, according to default M, W, set up simulated cushioned district;

Coding step: in cataloged procedure, periodically prediction is just in the size of data of coded frame and the size of data of frame to be encoded; Calculate the state of video sequence in described simulated cushioned district, and adjust the data volume non-NULL that quantization parameter makes video sequence in described simulated cushioned district, then proceed next code;

The coded data distributing step: when carrying out described coding step, direct or indirect distributes coded data to described broadcasting end;

Video playback step: at described broadcasting end structure decoder, start at any time to receive live, start displaying video reach the video data volume of duration M in the reception data after, make the bandwidth of the reception live TV stream of described broadcasting end be not less than W simultaneously.

2. method according to claim 1, it is characterized in that, described method also comprises: in the initialization codes step, after the coding duration reaches described M, the video data of coding is put into to described simulated cushioned district, the data volume that does not enter the video sequence in simulated cushioned district is defined as to SN, now, SN=0.

3. method according to claim 1, is characterized in that, described method also comprises: in the initialization codes step, the average bandwidth that video is set is A, and take A as the target bandwidth, by the ABR model, is encoded.

4. method according to claim 1, is characterized in that, described method also comprises: in coding step, take A as the target bandwidth, by the ABR model, first calculate quantization parameter, and then adjust quantization parameter according to simulated cushioned district numerical value.

5. method according to claim 1, is characterized in that, described method also comprises: in coding step, after each frame Video coding completes, prediction is just in the size of data of coded frame and the size of data of frame to be encoded; Calculate the state of video sequence in described simulated cushioned district, and adjust the data volume non-NULL that quantization parameter makes video sequence in described simulated cushioned district, then proceed next code.

6. method according to claim 1, is characterized in that, described coding step also comprises:

Step 1: after i frame coding, by live limited data transmission model, calculate the described simulated cushioned zone state under live condition; Described limited data transmission model is:

SN _i＝SN _i-1+F _i，

If SN _i>D _i* W, SN _i-=D _i* W,

Otherwise SN _i=0; Wherein, F _ibe the size of data of i frame of video, i the duration that frame of video is corresponding: the DTS timestamp of i frame of video is DTS _i, SN _iit is the data volume that does not enter the video sequence in described simulated cushioned district after i frame coding.

7. method according to claim 6, it is characterized in that, described prediction is just in the size of data of coded frame and the size of data of frame to be encoded, calculate the state of video sequence in described simulated cushioned district, and adjust the data volume non-NULL that quantization parameter makes video sequence in described simulated cushioned district, comprising:

Step 2: set frame of video F to be encoded _i+a+1the value of intermediate parameters q of quantization parameter:

$q_{i+a+1}=q_{i+a}+q_{i+a}\&times;({\mathrm{\&sigma;}}_i-A*{\mathrm{\&Sigma;}}_{j=0}^i{D}_j)/(A*\max (1,\sqrt{{\mathrm{\&Sigma;}}_{j=0}^i{D}_j}));$

Calculate quantization parameter according to q: $Q{P}_{i+a+1}=12+6*{\log }_2\left(\frac{q_{i+a+1}}{0.85}\right);$

Step 3: based on q _i+1to q _i+a, the prediction of video residual error and Video coding historical information is just in the size of coded frame: F _i+1, F _i+2..., F _i+a;

Step 4: δ is set _i+a+1=0;

Step 5: with q _i+a+1+ δ _i+a+1* Δ, for prediction q value, is predicted frame to be encoded: F _i+a+1, F _i+a+2..., F _i+a+b;

Wherein, Δ is constant, represents the minimum value that q changes, F _i+a+bbe the size of data of i+a+b frame of video, the cumulative data size that has the i frame from the 1st frame to the i frame altogether is σ _i=σ _i-1+ F _i, the quantity that a is current encoder system parallel encoding frame, the sequence of frames of video length of b presentation code forecast analysis, δ means the degree that q adjusts.

8. method according to claim 7, it is characterized in that, the state of video sequence in the described simulated cushioned district of described calculating, and adjust the data volume non-NULL that quantization parameter makes video sequence in described simulated cushioned district, then proceed next code, also comprise:

Step 6: according to limited data transmission model: SN ' _i+a+b=SN _i+ F _i+1+ F _i+2+ ... + F _i+a+bif, ${{\mathrm{SN}}^{\&prime;}}_{i+a+b}>\frac{({\mathrm{DTS}}_{i+a+b}-{\mathrm{DTS}}_i)}{{10}^7}*W,$ ? ${{\mathrm{SN}}^{\&prime;}}_{i+a+b}-=\frac{({\mathrm{DTS}}_{i+a+b}-{\mathrm{DTS}}_i)}{{10}^7}*W;$ Otherwise SN ' _i+a+b=0; DB ' _i+a+b=M-SN ' _i+a+b/ W; As DB ' _i+a+b, recalculate quantization parameter at>0 o'clock

then carry out the coding of subsequent frame;

Wherein, i the frame of video duration of video sequence in rear simulation buffering area of having encoded is DB _i, the encoded duration of rear video sequence of i frame of video of prediction is DB ' _i.

9. method according to claim 8, is characterized in that, in described coding step, as described DB ' _i+a+b<=0, δ _i+a+1+=1;

If δ _i+a+1<Max Δ/Δ, perform step 5 and step 6;

Wherein, the Max Δ is pre-set constant, means each maximum q adjusting range.

10. method according to claim 9, is characterized in that, if δ _i+a+1>=Max Δ/Δ, provide the underflow prompting of simulated cushioned district, simultaneously assignment δ _i+a+1=Max Δ/Δ, then recalculate quantization parameter

then carry out the coding of subsequent frame.

11. method according to claim 8, is characterized in that, in described coding step, as described DB ' _i+a+b<=0, define δ _i+a+1be mapped to DB ' _i+a+bfunction be: y=f (x); The y implication is δ _i+a+1during=x, the DB ' calculated _i+a+bvalue;

Definition P-1 rank multinomial: y=g (x)=t ₀+ t ₁* x+t ₂* x ²+ ... + t _p-1* x ^p-1, with g (x), carry out approach f (x);

Y=g (x) is calculated to sample point, and carry out matching: make x=0,1,2 ..., P-1, calculate y=y ₀, y ₁, y ₂..., y _p-1, at y=y ₀, y ₁, y ₂..., y _p-1computational process in, if y occurs for the first time _k>0, make δ _i+a+1=k, recalculate quantization parameter then carry out i+1 frame coding;

Wherein, the x implication is δ _i+a+1, x is integer, P, for the target order of prediction fitting function, is constant, 5<P<Max Δ/Δ.

12. method according to claim 11, is characterized in that, if all y _k<=0, by x=0,1,2 ..., P-1, and corresponding y=y ₀, y ₁, y ₂..., y _p-1, as the reference sample point, carry out least square fitting, draw every coefficient of polynomial function: t ₀, t ₁, t ₂..., t _p-1;

Ask minimum integer

make inequality g (x)>0 set up;

By newton's down-hill method to equation g (x)=t ₀+ t ₁* x+t ₂* x ²+ ... + t _p-1* x ^p-1=0 carries out iterative, wherein take x=P as iterative initial value, and 1 is iteration precision, and concentrates at solution of equation, chooses the minimum value x that meets following 2 conditions _root:

A) in interval

in;

B) to the first derivative g ' of g (x) (x)=t ₁+ t ₂* 2x+ ... + t _p-1* (p-1) * x ^p-2, and second dervative g " (x)=t of g (x) ₂* 2+t ₃* 3x+ ... + t _p-1* (p-1) * (p-2) * x ^p-3, meet: g ' (x _root) ≠ 0 or g " (x _root)>0;

If there is not the x satisfied condition _root, provide the buffer underflow prompting, simultaneously assignment δ _i+a+1=Max Δ/Δ, then recalculate quantization parameter

then carry out i+1 frame coding.

13. method according to claim 12, is characterized in that, if there is x _root:

If g ' is (x _root)>0, make δ _i+a+1=INT (x _root)+1; If g ' is (x _root)<0, make δ _i+a+1=INT (x _root); If g ' is (x _root)=0, g " (x _root)>0, make δ _i+a+1=INT (x _root);

Recalculate quantization parameter ${\mathrm{QP}}_{i+a+1}=12+6*{\log }_2\left(\frac{q_{i+a+1}+{\mathrm{\&delta;}}_{i+a+1}*\mathrm{\&Delta;}}{0.85}\right),$ Then carry out i+1 frame coding;

Wherein, INT (x) means to be not more than the maximum integer of x.

14. method according to claim 1, it is characterized in that, described method also comprises, in the video playback step, to the flash player, by the member property NetConnection::bufferTime that the NetConnection class is set, specify the flash play control before starting broadcasting, need the video data of buffer memory M duration, specifically comprise:

By the M assignment in bufferTime:

varnc：NetConnection=new?NetConnection；

var?flvbuftime：Number=stage.loaderInfo.parameters["buftime″]；

Nc.bufferTime=flvbuftime, flvbuftime is M.

Images