CN101729910B - Data transmission method and device based on gradable bit streams - Google Patents
Data transmission method and device based on gradable bit streams Download PDFInfo
- Publication number
- CN101729910B CN101729910B CN200810224472XA CN200810224472A CN101729910B CN 101729910 B CN101729910 B CN 101729910B CN 200810224472X A CN200810224472X A CN 200810224472XA CN 200810224472 A CN200810224472 A CN 200810224472A CN 101729910 B CN101729910 B CN 101729910B
- Authority
- CN
- China
- Prior art keywords
- data
- error correction
- importance
- enhancement data
- correction coding
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
Images
Landscapes
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
The invention discloses data transmission method and device based on gradable bit streams. The data transmission method comprises the following steps of: dividing reinforcement layer data into a plurality of grades according to the significance for reconstructing an image; adopting error correction modes with different redundancy rates for the data of different grades, wherein the higher the significance for reconstructing the image is, the higher the redundancy rate of the error correction mode adopted by a grade; and carrying out error correction coding for base layer data, wherein the redundancy rate of the error correction coding is at least equal to that of the error correction coding of a grade with the highest significance. Compared with the traditional transmission method adopting equal error protection or unequal data protection, the invention can enhance the channel disturbance resistant capacity under the precondition of indistinctly increasing the redundancy rate and achieve higher transmission reliability.
Description
Technical field
The present invention relates to the communications field, relate in particular to a kind of data transmission method and device based on gradable bit streams.
Background technology
Be accompanied by the fast development of radio network technique, increasing user wish can be on wireless network the real-time Transmission video flowing.In the video stream traffic of IP based network and client/server framework, client often has different terminal processes and display capabilities, and network bandwidth resources changes at any time.Scalable video not only is particularly suitable for Network Transmission, and can also satisfies the specific demand of the different terminal of client because have many-sided flexibilities such as time domain, spatial domain and quality.Along with the requirement of the gradable ability in market more and more widely and strong, ISO (International Organization for Standardization, International Standards Organization)/IEC (International Electro Technical Commission, International Electrotechnical Commission) and ITU-T (ITU Telecommunication Standardization Sector, International Telecommunications Union's telecommunication standards group) JVT (Joint Video Team, the video joint working group) code set of jointly setting up has begun the scalable video Study on Technology.H.264, the JVT code set is proposing a kind of new FGS (Fine granularity scalable in the working draft, fine graded) the realization technology, and SVC (Scalable video coding, gradable code stream) but be defined as the H.264 continuity of Scalable expansion.SVC supports the gradable of spatial resolution well, and gradable and qualitative CGS of time domain resolution (Coarse grain scalability, rough classification) and two kinds of FGS are gradable, and can support dissimilar gradable mixing-classifyings.
Owing to become, be subject to characteristics of interference during wireless channel, how under wireless environment, to guarantee quality, the raising code efficiency of SVC video flowing and strengthen fault freedom, be one of current research focus.Many documents have carried out this respect research.Aspect the SVC of link layer video data management, some document has proposed the needs according to different user, to the deletion of adaptive video data.Aspect SVC forward error correction protection; some document just adopts different protected modes according to the time priority level; but do not consider spatial scalability and SNR (Signal to Noise Ratio; signal to noise ratio) classification; also with good grounds signal-to-noise ratio enhancement layer priority difference, but do not consider the room and time classification.
Summary of the invention
The technical problem to be solved in the present invention provides a kind of data transmission method and device based on gradable bit streams; can be according to the difference of enhancement layer bit-stream when carrying out the error correction protection to reconstructed image importance; spatial scalability, temporal scalability and signal to noise ratio classification are taken all factors into consideration, improved the ability of anti-channel disturbance.
In order to address the above problem, the invention provides a kind of data transmission method based on gradable bit streams, comprising:
According to the importance of building the multigraph picture is divided into some grades with enhancement data to be sent;
The data of different brackets are carried out the error correction coding of different redundancys, the high more grade of importance is carried out the high more error correction coding of redundancy;
Base layer data is carried out error correction coding, and the redundancy of this error correction coding equals the redundancy of error correction coding that the highest grade of importance is carried out at least;
Send data through error correction coding;
Wherein, according to the importance to reconstructed image enhancement data to be sent being divided into some grades specifically is meant:
Described enhancement data is divided into Three Estate;
The grade that importance is the highest comprises the enhancement data of minimum time domain layer;
Importance time high grade comprises signal to noise ratio first enhancement data and time domain second enhancement data except that the highest time domain rank;
The grade that importance is minimum comprises the signal to noise ratio second and second above enhancement data and the highest time domain layer data;
When an enhancement data can be assigned in a plurality of ranks simultaneously, this enhancement data is assigned to the highest rank of importance in described a plurality of rank.
Further, according to the importance to reconstructed image enhancement data to be sent being divided into some grades specifically is meant:
Read second byte in the network abstraction layer unit, described enhancement data is divided into Three Estate according to time domain layer, other flag information of signal to noise ratio level TL, the QL that this byte provided;
The grade that importance is the highest comprises the enhancement data of TL=0;
Importance time high grade comprises the enhancement data of the QL=1 except that the highest time domain rank and the enhancement data of TL=1;
The grade that importance is minimum comprises the enhancement data of QL 〉=2 and the data of high time domain layer.
Further, the error correction coding that the data of different brackets are carried out different redundancys specifically is meant:
It is 18% forward error correction coding that the enhancement data of base layer data, importance highest ranking is carried out redundancy;
It is 12% forward error correction coding that importance time high-grade enhancement data is carried out redundancy;
It is 5% RS chnnel coding that the enhancement data of importance the lowest class is carried out redundancy.
Further, described data transmission method also comprises:
Enhancement data and base layer data, the inferior high-grade enhancement data of importance, the data of importance the lowest class to the importance highest ranking interweave respectively;
The error correction coding of the data of different brackets being carried out different redundancys specifically is meant: after interweaving, it is 18% forward error correction coding that the enhancement data of importance highest ranking and base layer data are carried out redundancy, it is 12% forward error correction coding that importance time high-grade enhancement data is carried out redundancy, and it is 5% forward error correction coding that the enhancement data of importance the lowest class is carried out redundancy;
The data that send through error correction coding specifically are meant: the data flow after will encoding is encapsulated as the RTP bag and sends.
The present invention also provides a kind of dispensing device based on gradable bit streams, comprising:
N error correction protected location, the redundancy difference of the error correction coding that each error correction protected location is adopted;
Gradable code stream encoder is used for the sequence of input is encoded, output spatial domain, time domain and the gradable data of signal to noise ratio;
Stage unit is used for being divided into N grade according to the enhancement data that the importance of reconstructed image is exported gradable code stream encoder; The enhancement data of different brackets is sent to different error correction protected locations, during transmission, the data of the grade that importance is high more send to the high more error correction protected location of error correction coding redundancy, and base layer data is sent to the highest error correction protected location of error correction coding redundancy;
Each error correction protected location is used for the data of input are carried out sending after the error correction coding.
Further, described stage unit is divided into some grades according to the importance to reconstructed image with enhancement data to be sent and specifically is meant:
Stage unit is divided into Three Estate with described enhancement data, and the grade that importance is the highest comprises the enhancement data of minimum time domain layer; Importance time high grade comprises signal to noise ratio first enhancement data and time domain second enhancement data except that the highest time domain rank; The grade that importance is minimum comprises the signal to noise ratio second and second above enhancement data and the highest time domain layer data;
When an enhancement data can be assigned in a plurality of ranks simultaneously, this enhancement data is assigned to the highest rank of importance in described a plurality of rank.
Further, described stage unit is divided into some grades according to the importance to reconstructed image with enhancement data to be sent and specifically is meant:
Stage unit reads second byte in the network abstraction layer unit, according to time domain layer, other flag information of signal to noise ratio level TL, the QL that this byte provided described enhancement data is divided into Three Estate; The grade that importance is the highest comprises the enhancement data of TL=0; Importance time high grade comprises the enhancement data of the QL=1 except that the highest time domain rank and the enhancement data of TL=1; The grade that importance is minimum comprises the enhancement data of QL 〉=2 and the data of high time domain layer.
Further, to adopt redundancy respectively be 18%, 12% and 5% forward error correction coding to the error correction coding module in three error correction protected locations;
Described stage unit sends to different error correction protected locations with the enhancement data of different brackets and specifically is meant: it is 18% error correction protected location that stage unit sends to the error correction coding redundancy with the enhancement data of base layer data, importance highest ranking; It is 12% error correction protected location that importance time high-grade enhancement data is sent to the error correction coding redundancy; It is 5% error correction protected location that the enhancement data of importance the lowest class is sent to the error correction coding redundancy.
Further, each error correction protected location specifically comprises: interleaving block is used for the data of input are interweaved;
The error correction coding module is used for the bit stream data after interweaving is carried out error correction coding;
The RTP packetization module is used for the data encapsulation after the error correction coding is sent for the RTP bag.
Technical scheme of the present invention is on the basis of the importance of taking all factors into consideration spatial scalability, temporal scalability and signal to noise ratio classification, adopt the error correcting system of different redundancys by importance degree, can improve the ability of anti-channel disturbance under the prerequisite of not obvious increase redundancy, transmission reliability is higher.
Description of drawings
Fig. 1 is based on the concrete implementing procedure figure of the anti-error code transmission method of classification bit stream;
Fig. 2 is the data structure schematic diagram of gradable code stream network abstraction layer unit;
Fig. 3 (a) and (b) are that different enhancement data are lost the schematic diagram of back to the influence of reconstructed image;
Fig. 4 is based on the concrete enforcement schematic block diagram of the dispensing device of gradable bit streams;
Fig. 5 (a) is among the embodiment to (f), the result that anti-error code transmission method of the present invention, EEP scheme and the emulation of traditional UEP scheme are compared.
Embodiment
Prior art is analyzed discovery, and these SVC unequal loss protections all are to adopt single index, such as according to time domain priority or SNR priority, do not take all factors into consideration three's relation.Given this, the invention provides a kind of anti-error code transmission plan based on SVC, at first analysis-by-synthesis spatial enhancement layer data, time enhancement data, signal-to-noise ratio enhancement layer data three are with respect to the importance of reconstructed image.According to the difference of importance, the data of enhancement layer are classified as some grades---such as being classified as 3 classes.For inhomogeneous enhancement data, adopt the forward error correction protected mode of different redundancys.
Experiment shows; the present invention has obtained reasonable effect to the unequal loss protection scheme of video data H.264SVC; with EEP (Equal error protection; Deng error protection) and common UEP (Unequal error protection; unequal error protection) protection scheme is compared, and the solution of the present invention has the gain of 2dB and 0.5dB respectively.
Data transmission method based on gradable bit streams provided by the invention comprises:
According to the importance to reconstructed image enhancement data to be sent is divided into some grades, described enhancement data can comprise spatial enhancement layer data, time enhancement data, signal-to-noise ratio enhancement layer data;
The data of different brackets are carried out the error correction coding of different redundancys, the high more grade of importance is carried out the high more error correction coding of redundancy;
Base layer data is carried out error correction coding, and the redundancy of this error correction coding equals the redundancy of error correction coding that the highest grade of importance is carried out at least;
Send data through error correction coding.
Optionally, can but be not limited to the data of base layer data, grade that importance is the highest are carried out the error correction coding of same redundant degree.
1,2 epimere words can be understood like this: suppose that enhancement data is divided into N grade is: ... N-1, N, wherein 1 grade importance is minimum, and the importance of N level is the highest; The redundancy R of the error correction coding of 1 grade The data then
1Be minimum in this N grade, the redundancy R of the error correction coding of 2 grades The data
2Be higher than 1 grade of error correction coding that data adopted, by that analogy, the redundancy R of the error correction coding of the The data of N level
NBe the highest in this N grade; And the redundancy of the error correction coding that base layer data adopts is equal to or greater than R
N
Optionally, described error correction coding can but be not limited to forward error correction coding; During practical application, do not get rid of yet and use other error correction coding.
Optionally, described error correction coding can but be not limited to adopt RS (Reed-Solomen) sign indicating number.
Optionally, can but be not limited to described enhancement data is divided into Three Estate, this Three Estate is arranged from high to low according to importance and is called category-A, category-B, C class (A, B, C only for the difference different brackets, do not play the qualification effect) successively.Certainly also can be divided into two grades, four grades etc. during practical application.
When described enhancement data being divided into category-A, category-B, C class Three Estate, described transmission method idiographic flow can be as shown in Figure 1:
At first, according to difference, will be divided into category-A through the enhancement layer bitstream (also can be described as enhancement layer bit-stream) of SVC coding output to reconstructed image importance, category-B, C class Three Estate, wherein category-A is the grade the highest to the importance of reconstructed image, the C class is to the minimum grade of the importance of reconstructed image;
Then respectively category-A and base layer data, category-B data, C class data are interweaved;
Bit stream data after three classes are interweaved adopts different forward error correction protected modes respectively, category-A data and base layer data adopt the protection scheme of high redundancy degree, the protection scheme of the medium redundancy of category-B The data, the protection scheme of the low redundancy of C class The data;
Send after will protecting the back data flow through RTP (Real-time Transport Protocol, RTP) packing at last.
Experimental result shows, compares with traditional UEP scheme with EEP, and the solution of the present invention has the ability of stronger anti-channel disturbance.
Optionally, the error correction coding that the data of different brackets are carried out different redundancys specifically can be meant:
It is 18% forward error correction coding that category-A data and base layer data are carried out redundancy, and it is 12% forward error correction coding that the category-B data are carried out redundancy, and it is 5% forward error correction coding that C class data are carried out redundancy.
To how describe the solution of the present invention below according to the grade of the importance of reconstructed image being divided enhancement data.
H.264SVC mix the coded system that has adopted hierarchical B-frame and strengthened gradually.For this reason, the data of each enhancement layer of SVC have different importance to image reconstruction, its importance has following relation: for the time enhancement data, the enhancement data of low time domain layer is than the importance height of high time domain layer enhancement data, because low time domain layer data will be as high time domain layer reference; For spatial enhancement layer data, the enhancement data of low-resolution layer (promptly low spatial domain layer) is than the importance height of high-resolution enhancement data, and is same because the low resolution layer data will be as high-resolution reference; Simultaneously, for the SNR enhancement data, the data of low SNR enhancement layer are than the importance height of the data of high SNR enhancement layer.
Judgement time, space, SNR enhancement layer in the present invention, also need this three combined and consider importance for reconstructed image for the importance of reconstructed image separately according to above-mentioned relation.
According to SNR, spatial domain and other relation of time domain layer level, introduce a parameter p
rCome each enhancement layer of index, p
rDefinition is shown in (1) formula:
Bi is an i bit of head the 2nd byte H.264SVC.
Fig. 3 has provided under direct channel, tests different enhancement data and loses the influence of back to reconstructed image.In test, the spatial domain is set is classified as 2, SNR is classified as 3, and GOP (Group of Pictures, image sets) length is respectively 32,16 and 8 (unit is a frame number), (s, l t) represent the rank of signal to noise ratio classification, spatial domain classification and time domain classification respectively with coordinate, s then, l and t scope are shown in (2) formula:
{0≤s≤2，0≤l≤1，0≤t≤5|GOP?Length＝32}
{0≤s≤2，0≤l≤1，0≤t≤4|GOP?Length＝16} (2)
{0≤s≤2，0≤l≤1，0≤t≤3|GOP?Length＝8}
Wherein, s, l and t are integer.
In this test, do not consider p
r＜64, because p
r＜64 basic layers of representative (s=0), its data have the highest priority.
In the description of this paper, the enhancement data of minimum time domain layer (being t=0) is equivalent to the minimum enhancement data of time domain rank, and other is analogized; The data of time domain enhancement layer 2 (being t=1) are that the time domain rank is 1 enhancement data, also are equivalent to time domain second enhancement layer (or second time domain layer enhancement layer) data, and other is analogized.And the enhancement data of s=0 is a base layer data, and the data of SNR enhancement layer 1 (s=1) are signal to noise ratio first enhancement layer (or first signal-to-noise ratio enhancement layer) data.
Can see at first that from Fig. 3 each enhancement data of minimum time domain layer (t=0) has the greatest impact to the decoded picture rebuild; As these enhancement layers (p
r=64,72,128,136) after data are lost, cause the decoded picture signal to noise ratio to descend significantly, because it can have influence on follow-up all reconstructions with reference to its image.The influence of time domain rank height is bigger than spatial domain and other influence of signal to noise ratio level.For example lose enhancement layer (s=1, l=0, data t=4) influence ratio lose enhancement layer (data t=1) are little for s=1, l=1, even the former is the spatial domain lowermost layer, but it does not do the reference frame of any other frame, thus the influence smaller; (data t=1) are lost influences ratio and loses enhancement layer (data t=5) are big for s=1, l=1, even the former is a SNR enhancement layer 2, but the frame number of its influence when rebuilding is much larger than the latter enhancement layer for s=2, l=0; It can also be seen that from Fig. 3 identical t under the l, loses SNR enhancement layer 1 (s=1, i.e. 64≤p
rThe ratio that influences of data≤127) is lost SNR enhancement layer 2 (s=2, i.e. 128≤p
rThe influence of data≤141) is big; Identical s is under the t, when losing spatial domain layer l=0 (64≤p among Fig. 3
rThe ratio that influences of data message≤68) is lost spatial domain layer l=1 (72≤p among Fig. 3
r≤ 77) data are big.
From above-mentioned analysis as can be known, different enhancement data have different importance to decoded picture.Optionally, described basis to the importance of reconstructed image with enhancement data to be sent be divided into some grades can but be not limited to be meant by following mode divided rank: enhancement data is divided into category-A, category-B, C class Three Estate, category-A has limit priority, comprise the minimum enhancement data of time domain rank (such as t=0, p
r=64,72,128,136); Category-B takes second place, comprise the SNR enhancement layer 1 except that the highest time domain rank data (such as s=1,65≤p
r≤ 67,73≤p
r≤ 76) and the time domain rank be 1 enhancement data (such as t=1, p
r=129,137); The C class has lowest priority, and the enhancement data that comprises time domain rank the highest (in this test, highest level is 3, is 4 during Gop Length=16, is 5 during Gop Length=32) when Gop Length=8 is (such as p
r=68,77, t=4), the data of SNR enhancement layer 2 and above (s 〉=2) (such as s=2,130≤p
r≤ 132 and 138≤p
r≤ 141).Certainly also be not limited to adopt other dividing mode in the practical application.
It is pointed out that if a certain enhancement data simultaneously can assign in a plurality of ranks the time, is assigned to the highest rank of importance in described a plurality of rank with this enhancement data; (s=2, l=1 t=1), can be divided into category-B according to t=1, and can be divided into the C class according to s=2, then are divided in the category-B such as enhancement layer.Also can be understood as, when dividing, begin to divide from the highest grade of described importance earlier, such as earlier with t=0 all be divided into category-A after, begin the category-B that is divided into again with s=1 and t=1, the rest may be inferred.
As seen, it is standard that one of breakthrough of the present invention is exactly with the importance to reconstructed image, to data from the spatial domain, time domain and SNR three aspects take all factors into consideration and sort out, rather than use single classification.
Need adopt the protected mode of the forward error correction FEC of different redundancys for described three class data; category-A bitstream data and base layer data adopt the forward error correction of highly redundant degree; the category-B bitstream data adopts the forward error correction of medium redundancy, and C class bitstream data adopts the forward error correction of low redundancy.
Optionally, can but be not limited to utilize spatial domain layer, time domain layer, other flag information of SNR level among the NALU (network abstraction layer unit), to the enhancement data of video flowing by importance divided rank to reconstructed image; The NALU cellular construction of each coding provides SNR, spatial domain and time domain layer class emblem information by second byte of its head as shown in Figure 2.When sorting out, mean: the enhancement data of minimum time domain layer is divided into category-A (TL=0) according to mode mentioned above; Signal to noise ratio first enhancement data (QL=1, signal to noise ratio first enhancement data except that the highest time domain rank) and the second time domain layer enhancement data (TL=1) are divided into category-B; The signal to noise ratio second and second above enhancement data (QL 〉=2) and the highest time domain layer data are divided into the C class.
Certainly in the practical application,, then also can utilize these information that enhancement data is sorted out if other unit or sign also can indicate spatial domain layer, time domain layer, SNR level other.
The also corresponding dispensing device that provides based on gradable bit streams of the present invention comprises:
N error correction protected location, the redundancy difference of the error correction coding that each error correction protected location is adopted;
The SVC encoder is used for the sequence of input is encoded, output spatial domain, time domain and the gradable data of signal to noise ratio;
Stage unit is used for being divided into N grade according to the enhancement data that the importance of reconstructed image is exported the SVC encoder, and described enhancement data can comprise spatial enhancement layer data, time enhancement data, signal-to-noise ratio enhancement layer data; The enhancement data of different brackets is sent to different error correction protected locations, during transmission, the data of the grade that importance is high more send to the high more error correction protected location of error correction coding redundancy, and base layer data is sent to the highest error correction protected location of error correction coding redundancy;
Each error correction protected location is used for the data of input are carried out sending after the error correction coding.
Optionally, can but the data and the base layer data that are not limited to the grade that importance is the highest sends to same error correction coding unit.
Optionally, each error correction protected location can but be not limited to adopt forward error correction coding; During practical application, do not get rid of yet and adopt other error correction coding.
Optionally, each error correction protected location can but be not limited to adopt RS (Reed-Solomen) sign indicating number.
Optionally, N can but be not limited to 3, described stage unit is divided into Three Estate with described enhancement data, this Three Estate is arranged from high to low according to importance and is called category-A, category-B, C class (A, B, C only for the difference different brackets, do not play the qualification effect) successively.Certainly N also can be worth for other during practical application.
Optionally, each error correction protected location can specifically comprise: interleaving block, error correction coding module, RTP packetization module;
Interleaving block is used for the data of input are interweaved;
The error correction coding module is used for the bit stream data after interweaving is carried out error correction coding;
It is that RTP wraps transmission that the RTP packetization module is used for the data after the error correction coding are beaten.
Optionally, when N=3, the error correction coding module in each error correction protected location can but to be not limited to adopt respectively redundancy be 18%, 12% and 5% forward error correction coding.
Described stage unit sends to different error correction protected locations with the enhancement data of different brackets and specifically is meant:
It is 18% error correction protected location that stage unit sends to the error correction coding redundancy with the enhancement data of base layer data, importance highest ranking; It is 12% error correction protected location that importance time high-grade enhancement data is sent to the error correction coding redundancy; It is 5% error correction protected location that the enhancement data of importance the lowest class is sent to the error correction coding redundancy.
Optionally, described stage unit can but be not limited to enhancement data at different levels are divided into Three Estate by following scheme:
Category-A comprises the enhancement data of minimum time domain layer (t=0); Category-B comprises signal to noise ratio first enhancement layer (s=1, signal to noise ratio first enhancement data except that the highest time domain rank) data and second time domain layer enhancement layer (t=1) data; The C class comprises signal to noise ratio second and above enhancement layer (s 〉=2) data and the highest time domain layer data; When an enhancement data can be assigned in a plurality of ranks simultaneously, this enhancement data is assigned to the highest rank of importance in described a plurality of rank.
Optionally, described stage unit can but be not limited to utilize spatial domain layer in the NALU unit, time domain layer, other flag information of SNR level, the enhancement data of video flowing by the importance divided rank to reconstructed image, be that is to say: the enhancement data of minimum time domain layer is divided into category-A (TL=0); Signal to noise ratio first enhancement data (QL=1) except that the highest time domain rank and the second time domain layer enhancement data (TL=1) are divided into category-B; The signal to noise ratio second and second above enhancement data (QL 〉=2) and the highest time domain layer data are divided into the C class.
Certainly in the practical application, if other unit or sign also can indicate spatial domain layer, time domain layer, SNR level other, then described stage unit also can utilize these information that enhancement data is sorted out.
A kind of concrete enforcement schematic block diagram of the dispensing device based on gradable bit streams as shown in Figure 4.
A specific embodiment of the present invention is as described below:
Referring to Fig. 1, be the transmission method that the present invention is based on the SVC data.At first the sequence with different resolution is input to the SVC encoder, coding output spatial domain, time domain and the gradable bit stream of signal to noise ratio.Based on the difference of each enhancement layer bit-stream to reconstructed image importance, enhancement data is divided into category-A, category-B, C class Three Estate, wherein category-A is the grade the highest to the importance of reconstructed image, the C class is to the minimum grade of the importance of reconstructed image.Then respectively category-A and base layer data, category-B data, C class data are interweaved respectively; the different forward error correction protected mode of The data after three classes are interweaved is carried out error correction coding; category-A and base layer data adopt the protection scheme of high redundancy degree; the protection scheme of the medium redundancy of category-B The data, the protection scheme of the low redundancy of C class The data.To protect the back data flow through RTP packing back output at last.
The step of this embodiment is:
(1) SVC coding: the sequence of high-resolution and low-resolution is input to the SVC encoder, coding output spatial domain, time domain and the gradable bit stream of signal to noise ratio;
(2) to SVC enhancement layer bit-stream divided rank: the SNR, spatial domain and the time domain layer class emblem information that provide according to second byte of NALU head, reference is the mode of described divided rank above, difference by to reconstructed image importance is divided into the SVC enhancement layer bit-stream Three Estates such as category-A, category-B and C class;
(3) unequal loss protection of SVC bit stream: category-A data and base layer data, category-B data and C class data are interweaved respectively; interweaving method is as described below, k the bag continuously that align in advance, and each bag reads 1 element then; vertically interweave and read k element formation RS (n, k) coding.
Bit stream data after three classes are interweaved adopts different forward error correction protected modes; Category-A data and base layer data adopt the forward error correction protected mode (the employing redundancy is 18% RS chnnel coding) of high redundancy degree; to the forward error correction protected mode (the employing redundancy is 12% RS chnnel coding) of the medium redundancy of category-B The data, to the forward error correction mode (the employing redundancy is 5% RS chnnel coding) of the low redundancy of C class The data.
(4) video stream data output:
Being provided with of the wireless channel parameter of emulation experiment is as shown in table 1.With JSVM 6.0 encoded videos, list entries is " Foreman ", " Hall ".The GOP size is respectively 32,16,8, and the space layer number of plies (being the spatial domain classification) is 2, and resolution is respectively QCIF (176 * 144), CIF (352 * 288), and the signal-to-noise ratio enhancement layer number of plies (being the SNR classification) is 2.
Table 1, wireless simulated channel parameter
In order to verify SVC UEP of the present invention guard method, EEP method and traditional UEP method and its contrast in emulation experiment, have been adopted.Each layer bit stream protected mode of three kinds of methods is as shown in table 2.Can observe from experiment, the pass of three kinds of whole redundancys of method is R
EEP＞R
Proposed≈ R
UEP(R represents redundancy).Forward error correction coding adopts RS (Reed-Solomen) sign indicating number, the error correction of the unexpected error that the RS sign indicating number is suitable in the wireless channel being produced.
Table 2, the used forward error correction protection scheme of emulation experiment
To " Forman " and " Hall " video sequence coding of input once after, then coded bit stream is carried out forward error correction coding, coding afterwards passes through wireless simulator 10 times.The original position difference of each bit-error code pattern, table 3 show 10 different original positions of each " bit-error code pattern ", and wherein sequence number 1 and 2 is represented the emulation at " Forman " video sequence, " Hall " video sequence respectively.The position that error code takes place subsequently is at random.Read last of file when the bit error code pattern, roll back to file header again and begin to read.
The error code original position of table 3, corresponding bit-error code pattern
Fig. 5 is relatively the solution of the present invention, EEP transmission method and traditional UEP method of example with video sequence " Forman " and " Hall ".The simulation result of Fig. 5 shows, channel packet loss l＜5% o'clock, all do not have loss of data when adopting three kinds of schemes basically.When the channel packet loss is between 5%＜l＜10%, there is partial data to lose when adopting the EEP method, video quality has obvious decline; Begin that loss of data is arranged when adopting traditional UEP method, but the data of losing are little to the video quality influence; And do not have loss of data when adopting the solution of the present invention basically.When channel packet loss l＞10%, there is lot of data to lose when adopting the EEP method, video quality declines to a great extent.Loss of data is also all arranged when adopting traditional UEP and the present invention program, and the data of losing increase along with the increase of packet loss, and the data of losing when still adopting the present invention program are less relatively, compare with traditional UEP, and the gain of 0.5～2dB is arranged.From " Forman " of different GOP length, the experiment comparative result of " Hall " sequence equates that at packet loss GOP is big more under the situation as can be seen, and corresponding video quality is high more.At the l=12% place, GOP=32,16,8 respectively than original video quality decline 1.9db, 2.05db, and 2.3db, so GOP is big more, corresponding anti-channel disturbance ability is strong more.
The curve of three kinds of methods can draw in the comparison diagram 5, and the UEP scheme is than EEP scheme performance good (as for " Foreman " sequence, when l=10%, the PSNR of UEP on average wants high 0.5db, and when l=15%, the PSNR of UEP on average wants high 3db) usually; In three kinds of methods, the solution of the present invention obtains best effect.Compare with UEP with EEP and to compare, the solution of the present invention on average has 2dB and 0.5dB gain respectively.
Certainly; the present invention also can have other various embodiments; under the situation that does not deviate from spirit of the present invention and essence thereof; those of ordinary skill in the art work as can make various corresponding changes and distortion according to the present invention, but these corresponding changes and distortion all should belong to the protection range of claim of the present invention.
Claims (9)
1. data transmission method based on gradable bit streams comprises:
According to the importance of building the multigraph picture is divided into some grades with enhancement data to be sent;
The data of different brackets are carried out the error correction coding of different redundancys, the high more grade of importance is carried out the high more error correction coding of redundancy;
Base layer data is carried out error correction coding, and the redundancy of this error correction coding equals the redundancy of error correction coding that the highest grade of importance is carried out at least;
Send data through error correction coding;
Wherein, according to the importance to reconstructed image enhancement data to be sent being divided into some grades specifically is meant:
Described enhancement data is divided into Three Estate;
The grade that importance is the highest comprises the enhancement data of minimum time domain layer;
Importance time high grade comprises signal to noise ratio first enhancement data and time domain second enhancement data except that the highest time domain layer;
The grade that importance is minimum comprises the signal to noise ratio second and second above enhancement data and the highest time domain layer data;
When an enhancement data can be assigned in a plurality of ranks simultaneously, this enhancement data is assigned to the highest rank of importance in described a plurality of rank.
2. data transmission method as claimed in claim 1 is characterized in that, according to the importance to reconstructed image enhancement data to be sent is divided into some grades and specifically is meant:
Read second byte in the network abstraction layer unit, described enhancement data is divided into Three Estate according to time domain layer, other flag information of signal to noise ratio level TL, the QL that this byte provided;
The grade that importance is the highest comprises the enhancement data of TL=0; Importance time high grade comprises the enhancement data of the QL=1 except that the highest time domain rank and the enhancement data of TL=1; The grade that importance is minimum comprises the enhancement data of QL 〉=2 and the data of high time domain layer.
3. data transmission method as claimed in claim 1 or 2 is characterized in that, the error correction coding of the data of different brackets being carried out different redundancys specifically is meant:
It is 18% forward error correction coding that the enhancement data of base layer data, importance highest ranking is carried out redundancy;
It is 12% forward error correction coding that importance time high-grade enhancement data is carried out redundancy;
It is 5% RS chnnel coding that the enhancement data of importance the lowest class is carried out redundancy.
4. data transmission method as claimed in claim 1 is characterized in that, also comprises:
Enhancement data and base layer data, the inferior high-grade enhancement data of importance, the data of importance the lowest class to the importance highest ranking interweave respectively;
The error correction coding of the data of different brackets being carried out different redundancys specifically is meant: after interweaving, it is 18% forward error correction coding that the enhancement data of importance highest ranking and base layer data are carried out redundancy, it is 12% forward error correction coding that importance time high-grade enhancement data is carried out redundancy, and it is 5% forward error correction coding that the enhancement data of importance the lowest class is carried out redundancy;
The data that send through error correction coding specifically are meant: the data flow after will encoding is encapsulated as the RTP bag and sends.
5. the dispensing device based on gradable bit streams is characterized in that, comprising:
N error correction protected location, the redundancy difference of the error correction coding that each error correction protected location is adopted;
Gradable code stream encoder is used for the sequence of input is encoded, output spatial domain, time domain and the gradable data of signal to noise ratio;
Stage unit is used for being divided into N grade according to the enhancement data that the importance of reconstructed image is exported gradable code stream encoder; The enhancement data of different brackets is sent to different error correction protected locations, during transmission, the data of the grade that importance is high more send to the high more error correction protected location of error correction coding redundancy, and base layer data is sent to the highest error correction protected location of error correction coding redundancy;
Each error correction protected location is used for the data of input are carried out sending after the error correction coding.
6. dispensing device as claimed in claim 5 is characterized in that, described stage unit is divided into some grades according to the importance to reconstructed image with enhancement data to be sent and specifically is meant:
Stage unit is divided into Three Estate with described enhancement data, and the grade that importance is the highest comprises the enhancement data of minimum time domain layer; Importance time high grade comprises signal to noise ratio first enhancement data and time domain second enhancement data except that the highest time domain rank; The grade that importance is minimum comprises the signal to noise ratio second and second above enhancement data and the highest time domain layer data;
When an enhancement data can be assigned in a plurality of ranks simultaneously, this enhancement data is assigned to the highest rank of importance in described a plurality of rank.
7. dispensing device as claimed in claim 6 is characterized in that, described stage unit is divided into some grades according to the importance to reconstructed image with enhancement data to be sent and specifically is meant:
Stage unit reads second byte in the network abstraction layer unit, according to time domain layer, other flag information of signal to noise ratio level TL, the QL that this byte provided described enhancement data is divided into Three Estate; The grade that importance is the highest comprises the enhancement data of TL=0; Importance time high grade comprises the enhancement data of the QL=1 except that the highest time domain rank and the enhancement data of TL=1; The grade that importance is minimum comprises the enhancement data of QL 〉=2 and the data of high time domain layer.
8. as claim 6 or 7 described dispensing devices, it is characterized in that:
It is 18%, 12% and 5% forward error correction coding that error correction coding module in three error correction protected locations adopts redundancy respectively;
Described stage unit sends to different error correction protected locations with the enhancement data of different brackets and specifically is meant: it is 18% error correction protected location that stage unit sends to the error correction coding redundancy with the enhancement data of base layer data, importance highest ranking; It is 12% error correction protected location that importance time high-grade enhancement data is sent to the error correction coding redundancy; It is 5% error correction protected location that the enhancement data of importance the lowest class is sent to the error correction coding redundancy.
9. dispensing device as claimed in claim 5 is characterized in that, each error correction protected location specifically comprises: interleaving block is used for the data of input are interweaved;
The error correction coding module is used for the bit stream data after interweaving is carried out error correction coding;
The RTP packetization module is used for the data encapsulation after the error correction coding is sent for the RTP bag.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN200810224472XA CN101729910B (en) | 2008-10-15 | 2008-10-15 | Data transmission method and device based on gradable bit streams |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN200810224472XA CN101729910B (en) | 2008-10-15 | 2008-10-15 | Data transmission method and device based on gradable bit streams |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101729910A CN101729910A (en) | 2010-06-09 |
| CN101729910B true CN101729910B (en) | 2011-11-23 |
Family
ID=42449974
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN200810224472XA Expired - Fee Related CN101729910B (en) | 2008-10-15 | 2008-10-15 | Data transmission method and device based on gradable bit streams |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101729910B (en) |
Families Citing this family (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP5768332B2 (en) * | 2010-06-24 | 2015-08-26 | ソニー株式会社 | Transmitter, receiver and communication system |
| KR20140092935A (en) * | 2011-08-18 | 2014-07-24 | 브이아이디 스케일, 인크. | Methods and systems for packet differentiation |
| CN104620525B (en) * | 2012-07-09 | 2018-07-27 | 瑞典爱立信有限公司 | Method and apparatus for distributing information during broadcast delivery |
| CN109089122B (en) * | 2013-01-18 | 2022-08-05 | 弗劳恩霍夫应用研究促进协会 | Forward error correction data generator and forward error correction decoder |
| CN103107863B (en) * | 2013-01-22 | 2016-01-20 | 深圳广晟信源技术有限公司 | Digital audio source coding method and device with segmented average code rate |
| CN103067719A (en) * | 2013-01-31 | 2013-04-24 | 南京邮电大学 | Real-time video communication method based on unequal error protection |
| CN104036826B (en) * | 2014-06-12 | 2018-08-28 | 上海新储集成电路有限公司 | The selecting method of error correction circuit in memory |
| CN111541514B (en) * | 2015-06-29 | 2021-10-26 | 华为技术有限公司 | Message transmission method and device |
| CN107592540B (en) * | 2016-07-07 | 2020-02-11 | 腾讯科技（深圳）有限公司 | Video data processing method and device |
| CN109669917B (en) * | 2018-12-24 | 2020-07-10 | 北京华大九天软件有限公司 | Waveform storage method based on priority grouping |
| CN113395521A (en) * | 2020-03-13 | 2021-09-14 | 华为技术有限公司 | Image encoding and decoding method and device |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1729696A (en) * | 2002-12-19 | 2006-02-01 | 皇家飞利浦电子股份有限公司 | System and method for drift-free fractional multiple description channel coding of video using forward error correction codes |
-
2008
- 2008-10-15 CN CN200810224472XA patent/CN101729910B/en not_active Expired - Fee Related
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1729696A (en) * | 2002-12-19 | 2006-02-01 | 皇家飞利浦电子股份有限公司 | System and method for drift-free fractional multiple description channel coding of video using forward error correction codes |
Non-Patent Citations (1)
| Title |
|---|
| 孙刚，等.带宽约束下FGS视频传输非均等错误保护.《浙江大学学报（工学版）》.2006,第40卷(第12期),见摘要、2059页－2062页第3部分. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101729910A (en) | 2010-06-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101729910B (en) | Data transmission method and device based on gradable bit streams | |
| Apostolopoulos et al. | Modeling path diversity for multiple description video communication | |
| CN102131083B (en) | Method and device for transmitting scalable videos based on priorities | |
| Lee et al. | Layered coded vs. multiple description coded video over error-prone networks | |
| CN101707701B (en) | Multi-stream information transmission system and transmission method thereof | |
| DE60223147T2 (en) | METHOD OF MEDIA STREAMING WITH BITSTROME MULTIPLE DESCRIPTIONS | |
| Zhang et al. | Unequal packet loss protection for layered video transmission | |
| CN1910932A (en) | Method of spatial and SNR fine granular scalable video encoding and transmission | |
| US8300705B2 (en) | Method for generating and processing hierarchical PES packet for digital satellite broadcasting based on SVC video | |
| CN101777960A (en) | Audio encoding method, audio decoding method, related device and communication system | |
| US10469202B2 (en) | Fec mechanism based on media content | |
| CN101917625A (en) | Joint source-network coding-based gradable video stream transmission method | |
| Monteiro et al. | Evaluation of the H. 264 scalable video coding in error prone IP networks | |
| Lee et al. | Performance comparisons of layered and multiple description coded video streaming over error-prone networks | |
| CN104137455B (en) | The method and apparatus that streaming service is provided | |
| CN101595733A (en) | Be used for the transmission of hierarchical modulation of scalable video bitstream and the apparatus and method of reception | |
| Hellge et al. | Multidimensional layered forward error correction using rateless codes | |
| Nejati et al. | Distortion optimal transmission of multi-layered FGS video over wireless channels | |
| Sardari et al. | Multilevel diversity coding via rateless codes for reliable and scalable video multicasting | |
| Pliakas et al. | Scalable Video Streaming Traffic Delivery in IP/UMTS Networking Environments. | |
| Korhonen et al. | Bit-error resilient packetization for streaming H. 264/AVC video | |
| CN100391264C (en) | Multi-descriptive coding-decoding method based on AVS video stardard | |
| CN101325717A (en) | H.264/SVC code flow regrouping method for OFDM protection | |
| Wang et al. | Channel-adaptive error protection for scalable video over channels with bit errors and packet erasures | |
| Baqai et al. | Error resilience of EZW coder for image transmission in lossy networks |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20111123 Termination date: 20211015 |