CN101065780A - Reference picture management in video coding - Google Patents
Reference picture management in video coding Download PDFInfo
- Publication number
- CN101065780A CN101065780A CN 200580040403 CN200580040403A CN101065780A CN 101065780 A CN101065780 A CN 101065780A CN 200580040403 CN200580040403 CN 200580040403 CN 200580040403 A CN200580040403 A CN 200580040403A CN 101065780 A CN101065780 A CN 101065780A
- Authority
- CN
- China
- Prior art keywords
- reference picture
- parameter
- image
- rpn
- value
- 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.)
- Pending
Links
Images
Landscapes
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
A method for encoding a sequence of pictures comprising using one or more pictures as reference pictures, labeling the reference pictures with a first parameter, signaling the first parameter to a decoder, and using a reference picture management, wherein all the reference pictures are identified by a second parameter which is derived on the basis of the first parameter.
Description
Technical field
The present invention relates to the reference picture management in the video Code And Decode.
Background technology
There are a plurality of video encoding standards that comprise ITU-TH.261, ISO/IEC MPEG-1 vision, ITU-TH.262 or ISO/IEC MPEG-2 vision, ITU-TH.263, ISO/IEC MPEG-4 vision and ITU-TH.264 or ISO/IEC MPEG-4 AVC.H.264/AVC be the work product of the joint video team (JVT) of ITU-T video coding expert group (VCEG) and ISO/IEC MPEG.
In addition, also there is the achievement of being devoted to the new video coding standard.One of them is the exploitation of the scalable video (SVC) among the MPEG.It will become MPEG-21 the 13rd part.Second achievement is the exploitation of the Chinese video encoding standard worked out by Chinese audio-visual coding standard working group (AVS).AVS has finished it in February, 2004 and has been directed to first video coding criterion AVS1.0 of SDTV and HDTV application.From then on focus is transferred to the mobile video service.
Many available video encoding standards are utilized motion compensation, i.e. predictive coding is so that remove time redundancy between the vision signal for high code efficiency.In motion compensation, one or more previous decoded image is used as the reference picture of the present image that is being encoded or is decoding.When encoding, making the differential signal between current block and the reference block need the bit of minimal amount to represent from reference picture searching for reference piece to the block of pixels (current block) of present image.It is also conceivable that in the searching for reference piece displacement between current block and the reference block is encoded.And, can also in the searching for reference piece, consider the distortion of reconstructed block.
In coded video bit stream, when other image was encoded, some images can be used as reference picture, and some may be used as reference picture never.The image that is not used as reference picture is called non-reference picture.Whether be the signal of reference picture to scrambler if should send image to demoder then, makes that demoder does not need to store this image for motion compensation reference.At first, each image should be stored in the frame buffer of back decoder buffer or decoding and be labeled as " be used for reference to ".But, when reference picture be not used further to reference to the time, it will be marked as " be not used in reference to ".In other images, reference picture marking was handled by reference picture management and finished for " being used for reference " or " being not used in reference ".
For coding or the selected reference picture of piece of decoding can be the recently image (being commonly referred to short-term reference picture) of decoding, the perhaps image that amphi position was decoded before the image of present encoding in decoding order (being commonly referred to the long term reference image).An example that comprises the image stream 100 of reference picture 101,103,105,106,108,110 and non-reference picture 102,104,107,109 has been described in Fig. 1.Suppose that reference picture 101 is short-term reference picture (when image 103 and 102 is encoded), and supposition reference picture 105 is long term reference image (when image 106 is encoded).Long term reference image 105 and use long term reference image are not shown in Figure 1 as the image between the image 106 of reference image.
In the standard that allows to be used for short-term and long term reference image, for example H.263 and H.264/AVC, reference picture management is handled and is separated between short-term reference picture and long term reference image.In addition, handle designated being used for short-term reference picture is labeled as the long term reference image for one.In H.264/AVC, short-term reference picture is identified by variable PicNum, and the long term reference image is identified by variable L ongTermPicNum.In the specific item 8.2.4.1 of standard H.264/AVC, PicNum and LongTermPicNum have been stipulated.Therefore, make up every other the separating at short-term reference picture and long term reference image of (in the specific item 8.2.4 of standard H.264/AVC, stipulating) and reference picture marking (in the specific item 8.2.5 of standard H.264/AVC, stipulating) with reference to bookkeeping such as reference picture list.
In the H.263 Annex of standard N (reference picture preference pattern), the 10 bit time reference index TRI or the RTR of express time reference are used to identify reference picture.A shortcoming of this solution is to be restricted to less than 1024 unit in the time interval between reference picture and present image.Limit this unit according to the live image clock frequency.In other words, do not enable so-called long term reference image.
In the H.263 Annex of standard U (the reference picture preference pattern of enhancing), be used to identify short-term reference picture with 10 bit image that increase progressively number (PN) at each reference picture (hereinafter being called " image of storage ").The Variable Length Code LPIN that represents long-term image index is used to identify the long term reference image.
Standard H.264/AVC in, use PicNum and LongTermPicNum to identify short-term reference picture and long term reference image respectively.PicNum and LongTermPicNum are similar to H.263 PN and the LPIN among the Annex U of standard respectively, but PicNum and LongTermPicNum both expansions are used for progressively-encode and interleaved code.PicNum and PN also have another difference, and the value of PicNum can be a negative and along with differing from and successively decrease between the decoding order of the decoding order of present image and reference picture.For example, the PN of a row reference picture can be 1022,1023,0,1,2, and the PicNum of the reference picture of same row can be-2 ,-1,0,1,2.
For example, patented claim US-09/892977, WO 01/86960 and GB 2382403, and standard H.263 H.264/AVC Annex U and standard disclose some prior art solutions to reference picture management in the video coding.
The separate management of short-term reference picture and long term reference image has caused complicated reference picture management operation, has therefore improved the enforcement complicacy of hardware enforcement and software implementation.
Summary of the invention
The invention provides the reference picture management solution of the enforcement of a kind of for example video encoder that is used for whether supporting using the long term reference image method and/or demoder no matter.
According to an exemplary embodiment of the present invention, how far the present image no matter reference picture is being encoded or decoded at the decoding order middle distance has, and all manages reference picture in the same manner.Therefore reference picture does not need to be divided into short-term image or long-term image.Reference picture is identified by variable, runs through the video sequence of coding, and the value of variable is unique for reference picture.This variable can also be used for the management processing of all reference pictures except the sign reference picture.
In the present invention, disclose unified reference picture management and handled, it makes Video Decoder and/or the scrambler that can enable simplification when supporting that the long term reference image is implemented implement.
Standard H.264/AVC in, exist to be used for the syntax table that reference picture reorders.Eight syntactic elements (that is encoded point) are arranged in syntax table.When use is of the present invention, do not need two in these syntactic elements.Standard H.264/AVC in, also be useful on the syntax table of reference picture heavy label.Eight syntactic elements are arranged in this syntax table, do not need when implementing when of the present invention wherein four.
The present invention can be implemented as software to a great extent, and wherein this software can be simplified to a certain extent.
The reference picture that is proposed reorders and the mark processing can realize effectively the needed information of reference picture management process being sent as signal.
Description of drawings
Hereinafter will present invention is described in more detail with reference to appended accompanying drawing, wherein
Fig. 1 shows the example of the image stream that comprises reference picture and non-reference picture;
Fig. 2 shows the example of the image stream that comprises frame number;
Fig. 3 shows the example according to signal of the present invention;
Fig. 4 shows the example of the method according to this invention with process flow diagram;
Fig. 5 has drawn the embodiment according to an advantage of system of the present invention;
Fig. 6 has drawn the embodiment according to an advantage of scrambler of the present invention;
Fig. 7 has drawn the embodiment according to an advantage of demoder of the present invention.
Embodiment
Only to describe following enforcement of the present invention aspect at the mode of progressively-encode, one of them image is equivalent to a frame.But clearly they can be expanded to being used for progressively-encode and interleaved code in the mode that is similar to according to standard prior art H.264/AVC, and wherein image can be zone or frame.And following aspect of the present invention only is described at forward prediction.With their the expansion be used for as standard H.264/AVC defined bi-directional predicted also be clearly.
To come hereinafter that present invention is described in more detail with reference to the demoder 2 of the scrambler 1 of the system of figure 5, Fig. 6 and Fig. 7.Image to be encoded for example can be the image from the video flowing of the video source 3 of for example camera, video recorder etc.The image of video flowing (frame) can be divided into a plurality of less part such as sheet.Sheet can also further be divided into piece.In scrambler 1, video flowing encoded treat via transmission channel 4 transmission or the information that are passed to the storage medium (not shown) to be passed with minimizing.The image of video flowing is input to scrambler 1.Scrambler has the coded buffer 1.1 (Fig. 6) that is used for temporarily storing some images to be encoded.Scrambler 1 also comprises storer 1.3 and processor 1.2, wherein can implement according to coding task of the present invention.Transmission equipment 6 can shared storage 1.3 and processor 1.2 or transmission equipment 6 can have another processor and/or storer (not shown) to be used for other functions of transmission equipment 6.Scrambler 1 execution estimation and/or some other tasks are with compressing video frequency flow.In case as using with reference to image, then reference picture must be stored in (for example in the frame buffer 5.2 of decoding) in the impact damper.Scrambler 1 can also be inserted into the information about the DISPLAY ORDER of image in the transport stream.
According to cataloged procedure, if necessary, image encoded can move to the image impact damper 5.3 that interweaves.And the reference picture of coding is decoded and be inserted in the frame buffer 5.2 of decoding of scrambler.Image encoded is transferred to receiving equipment 8 from scrambler 1 via transmission channel 4 by transmitter 7.In receiving equipment 8, receiver 9 receive institute's information transmitted and carry out operations necessary with will become by the conversion of signals that transmitter 7 is transmitted the form of known suitable demoder 2.In demoder 2, image encoded is decoded to form as far as possible the unpressed image corresponding to image encoded.
Demoder 1 also comprises storer 2.3 and processor 2.2, wherein can implement decoding task.Receiving equipment 8 can shared storage 2.3 and processor 2.2 or receiving equipment 8 can have another processor and/or storer (not shown) to be used for other functions of receiving equipment 8.
Coding
Let us is considered coding-decoding processing in more detail now.When needs, be transfused to into scrambler 1 and be stored in the coded buffer 1.1 from the image of video source 3.Encoding process need not just get started after first image enters scrambler, but after the image of certain amount is available in coded buffer 1.1.Scrambler 1 is attempted the suitable candidate of search from the image of the reference frame that is used as estimation then.Scrambler 1 is carried out coding to form image encoded then.Image encoded for example can be predicted picture (P), bidirectional predictive picture (B) and/or intra-coded picture (I).Can not use any other image that intra-coded picture is decoded, but the image of other type needed a reference picture at least before they can be decoded.The image of any kind referred to above can be used as reference picture.
Scrambler 1 additional instance is given image as two timestamps: decoded time stamp (DTS) and output time stab (OTS).Demoder can stab to determine the correct decode time and the time of output (demonstration) image service time.But, these timestamps need not be transferred to demoder or demoder and not stab service time.
Next present buffer model.Precoding impact damper 1.0, decoded picture buffer 5.2 and the impact damper 5.3 that interweaves are empty at first.Unpressed image is inserted into the precoding impact damper to catch order.When using any time domain scalability schemes, before coding, be buffered in the precoding impact damper more than one not compressed image.After initial precoding buffering, encoding process begins.Scrambler 5 is carried out encoding process.As the result of encoding process, the reference picture of scrambler generation decoding and image encoded and removal are from precoding impact damper image encoded.The reference picture of decoding is inserted in the frame buffer 5.2 of decoding and and be inserted in the impact damper 5.3 that interweaves image encoded.Transmitting apparatus is selected the data cell of transmission from the image encoded of the impact damper that interweaves.Impact damper is removed the data cell of the transmission of image encoded from interweaving.
Transmission
The transmission of image encoded or storage (and optionally virtual decoded) can begin after first image encoded is ready immediately.This image need not be first in the demoder output order, because decoding order can be different with the output order.
When to first picture coding of video flowing, transmission can begin.Alternatively image encoded is stored in the impact damper 5.3 that interweaves.Transmission can also start from the stage after a while, for example behind certain the part coding to video flowing.
Decoding
Receiver 8 is collected all and is belonged to all data cells of (a plurality of) signal that is received of image, makes them have rational order.The severity of order depends on employed configuration.The data cell that receives is to receive sequential storage (pre-decoder buffer, deinterleaving impact damper) in reception buffer 9.1.Receiver 8 is abandoned any out of use, and remainder is passed to demoder 2.
Decode and it is stored in the decoded picture buffer 2.1 by 2.2 pairs of image encoded of processor.Decoded picture buffer 2.1 comprises the memory location that is used to store a plurality of images.Those positions can also be called the frame warehouse.Demoder 2 is decoded to the image that is received from the removed order of deinterleaving impact damper (that is decoding order) with them.As long as image needs then will be stored in the decoded picture buffer 2.1 as the image of reference picture as with reference to image.When reference picture is marked as " being not used in reference " (perhaps replacedly, mark " is used for reference " and is removed) time, if its output or demonstration time expire and/or the image of new decoding is stored on this reference picture, then this reference picture can be removed from decoded picture buffer 2.1.
The sign of reference picture
In the present invention, a variable that has unique value for all reference pictures in the encoded video sequence is used to identify reference picture, and no matter how far reference picture has apart from present image on time sequencing, decoding order or any other order in same encoded video sequence.This variable is called reference picture number and is abbreviated as RPN at this.
Encoded video sequence basically with standard H.264/AVC in defined term identical.The definition of encoded video sequence is: on decoding order by zero or comprise up to all follow-up images but do not comprise that instantaneous decoding that a plurality of non-IDR image of any follow-up IDR image is followed upgrades the sequence of the coded image of (IDR) image construction.The IDR image is an intra-coded picture, after it is decoded, can decode under not from the situation prior to the reference of any image of IDR picture decoding by all subsequent encoding images of decoding order.First image of each encoded video sequence is the IDR image.
Reference picture number (RPN) goes out from the information inference that sends at each image.For example, reference picture number can be derived from time reference (for example H.263 the TR the picture headers) or frame number (FN), this time reference or frame number at each reference picture in the modulo n arithmetic 1 to increase progressively (for example H.264/AVC frame number (frame_num) in the sheet head and the H.263Annex PN of defined among the U).
When reference picture RPN has some advantages when frame number FN derives.At first, frame number FN is only to the reference picture counting, and secondly non-reference picture is not stored in the demoder frame buffer of back and is used for reference.Clearly, similarly derivation method can be used for from the reference picture RPN that derives such as the out of Memory of time reference.
The frame number value of IDR image can be set to 0 and maximum frame number value MaxFN between any integer value, although it is set to 0 usually.Maximum frame number value MaxFN and 1 and be expressed as MaxFNplus1.Can indicate this MaxFNplus1 according to the information and/or the codec standard that send.The IDR image is reference picture naturally.For in the same encoded video sequence in the decoding order after image (no matter it is a reference picture also is non-reference picture), the FN value that FN value in the image equals the last reference picture in the decoding order adds 1 delivery MaxFNplus1, shown in the example of Fig. 2, wherein the image shown in all is that reference picture and MaxFNplus1 are 256.
The reference picture of reference picture is number following to be derived based on frame number FN.Equal FN and be stored in the reference picture that is used for reference in the decoder buffer 5.2,2.1 of back for having frame number, make parameter p revFN equal the frame number of the last reference picture in the decoding order, and make parameter p revRPN equal the reference picture number of last parametric image.The reference picture of following then calculating parameter image number:
if(prevFN<=FN)
then?RPN=prevRPN+FN-prevFN
else
RPN=prevRPN+FN-prevFN+MaxFNplus1
The reference picture list initialization
Initial reference picture list index stores is used for the reference picture of reference in the decoder buffer of back, make that reference picture starts from having the reference picture of the highest RPN value and continuing to proceed to the reference picture with minimum RPN value in order.For example, be used for reference if stored four images, and their RPN value is 255,502,1027 and 1029, initial tab sequential is 1029,1027,502,255.Utilize the tab sequential of this acquiescence, Variable Length Code (VLC) code 0 can be used to indicate that to have the RPN value be 1029 reference picture, and code 1 can be used to indicate that to have the RPN value be 1027 reference picture, or the like.
Reference picture list reorders
Each predicted picture can have a plurality of reference pictures.These reference pictures sort in being called two reference picture lists of RefPicList0 and RefPicList1.Each reference picture list has an initial order, and this order can reorder to handle and changes by reference picture list.For example, the initial order of supposing RefPicList0 is r0, r1, r2...rm, and it uses the variable length code coding.Code 0 expression r0, code 1 expression r1 or the like.If scrambler know r1 than r0 use more frequent, then demoder can be by exchange r0 and r1 this tabulation of resequencing, and makes code 1 expression r0, code 0 expression r1.Because code 0 is shorter than code 1 on code length, so realized improved code efficiency.Reference picture reorders to handle and must be transmitted in the bit stream, makes demoder can be each reference picture list correct reference picture of deriving in proper order.
Being used for a kind of method that reference picture list reorders is that the RPN value is sent with indication reference picture to be reordered as signal.For example, be 255,1027,1029,502 if tab sequential 1029,1027,502,255 is rearranged preface, the tabulation information of reordering that then will be sent as signal is (by their appearance ordering):
VLC code at 255
VLC code at 1027
After handling second code, to have the RPN value and be 1027 reference picture and put into the second place of this order, and will put into this order according to initial order except the order of other reference picture the above-mentioned reference picture of having handled behind second reference picture.This tab sequential becomes 255,1027,1029,502 then.
The problem of said method is that the bit number that original RPN value is sent as signal may be very big, because higher value has long code length usually in the VLC coding.
In order to save the reorder bit of information of expression tabulation, can use the predictive coding of RPN value.A kind of possible method is similar to and is used for the method that reorders in standard short-term reference picture tabulation H.264/AVC.Replacement directly sends the RPN value at reference picture to be reordered as signal, absolute difference between prediction and the RPN value is subtracted 1 to be sent as signal, be expressed as AbsDIFFminus1, the indication of absolute value that adds or deduct this difference together with predicted value is expressed as ASidc with derivation RPN value.For first reference picture to be reordered, predicted value equals RPNcurr, is expressed as predRPN.After information was reordered in the tabulation of handling each reference picture to be reordered, predRPN was set to the RPN value of the reference picture that equals just to have reordered.
The following RPN value of deriving reference picture to be reordered:
if(ASidc==0)
RPN=predRPN-(AbsDIFFminus1+1)
else?if(ASidc==1)
RPN=predRPN+(AbsDIFFminus1+1)
For above example, suppose that RPNcurr equals 1030, the tabulation information of reordering of signal to be sent becomes:
AbsDIFFminus1=774,ASidc=0
AbsDIFFminus1=771,ASidc=1
Can derive first reference picture to be reordered and have the RPN value and equal (1030-(774+1)=255), and second reference picture to be reordered has the RPN value and equals (255+ (771+1)=1027).
But being seen as us, said method is not very effective, because may be still very big as the value that signal sent.
The invention provides the reorder coding of information of a kind of effective reference picture list.The prediction of the RPN value of the reference picture that use is to be reordered.In order to indicate the RPN value that three information are sent as signal:
1) absolute difference between prediction and the RPN value subtracts 1, is expressed as AbsDIFFminus1,
2) indication that adds or deduct, be used to derive predicted value and RPN value are expressed as ASidc, and
3) numerical range of predicted value is expressed as PS.Should select to make the only PS value in the scope between 0 to MaxFNplius1 of AbsDIFFminus1.
For first reference picture to be reordered, the following calculating of predicted value predRPN:
predRPN=RPNcurr-PS*MaxFNplus1
After information is reordered in the tabulation of handling each reference picture to be reordered, predicted value predRPN at first is set to the RPN value of the reference picture that equals just to have reordered.The following renewal of predRPN then:
if(ASidc==0)
predRPN=predRPN-PS*MaxFNplus1
else?if(PNidc==1)
predRPN=predRPN+PS*MaxFNplus1
The RPN value of reference picture to be reordered is derived as follows:
if(ASidc==0)
RPN=predRPN-(AbsDIFFminus1+1)
else?if(ASidc==1)
RPN=predRPN+ (AbsDIFFminus1+1) then
For above-mentioned example, suppose RPNcurr equal 1030 and MaxFNplus1 equal 256, the tabulation information of reordering that sends as signal in signal 300 becomes as follows:
AbsDIFFminus1=6, ASidc=0, PS=3 (it illustrates with the label among Fig. 3 301)
AbsDIFFminus1=3, ASidc=1, PS=3 (it illustrates with the label among Fig. 3 302)
Can derive first reference picture to be reordered and have the RPN value and equal 1030-3*256-(6+1)=255, and second reference picture to be reordered has the RPN value and equals 255+3*256+ (3+1)=1027.
The value that sends as signal is less as can be seen, therefore can save the bit that the expression reference picture list reorders and handles.
The simple change that should state said method is fine usually.For example, three information can be included in two syntactic elements (by combination ASidc and PS in a syntactic element) and three syntactic elements.Estimation range PS can be based on the value except MaxFNplius1, as long as can indicate this value according to codec standard and/or the relevant information that sends.
Reference picture marking
Reference picture marking is handled to be mainly used in some reference picture markings is " be not used in reference to ", if make they output or show that time is up the phase, then they can be removed from the decoder buffer 2.1,5.2 of back.There are two kinds with reference to image tagged mechanism, first in first out moving window method and special-purpose self-adaptation labeling method.
Be similar to those methods that in H.264/AVC, are used for moving window marking operation and self-adaptation marking operation and can be applied to it is used to identify reference picture at RPN situation.
For the moving window marking operation, the sum that no matter when is used for the image of reference after being stored in the decoder buffer equals maximal value and new reference picture will be stored, and an image that then has minimum RPN value is marked as " being not used in reference ".
For the self-adaptation marking operation, the required information of RPN of the reference picture that derivation is to be marked sends as signal.The information that sends as signal is that the RPNcurr of reference picture to be marked and the difference between the RPN value subtract 1, is expressed as diffRPNminus1.
The RPN value of reference picture to be marked is derived as follows:
RPN=RPNcurr-(diffRPNminus1+1)
For same example before, equal 255 reference picture and be marked as " be not used in reference to " if having RPN, then the signal that sends as signal is diffRPNminus1=774.
Can derive reference picture to be marked has the RPN value and equals (1030-(774+1)=255).
The problem of prior art moving window marking operation mentioned above illustrates by following example.Suppose that RPNcurr equals 200, three and has the RPN value and equal 60,198 and 199 image and be stored in the decoder buffer of back and be used for reference, the maximum number that is used for the image of being stored of reference is 3.For next one image to be encoded, scrambler 1 will have RPN and equal 60 reference pictures store being used for the use of back, is " be not used in reference to " and RPN equaled 199 reference picture marking.In this case, it will be effective using the moving window marking operation.But the moving window marking operation of prior art equals RPN to 60 reference picture marking and is " be not used in reference to ".
The invention provides a kind of solution at the problems referred to above.For the moving window marking operation, another the size of the information indication moving window that additionally sends as signal, be expressed as SSW.Only operate SSW reference picture with maximum RPN value according to first-in first-out rule.Do not comprise reference picture with smaller value.
For example, poor between the maximum number of the information that additionally the sends image that is used for reference that equals to be stored and the SSW as signal.In above-mentioned example, then the information that additionally sends as signal is exactly the code of an expression 1 (equaling 3-2).
It can also be seen that the self-adaptation marking operation of prior art is not very effective, because may be very big as the value of signal transmission.Unfortunately, the signal that directly sends the RPN value of reference picture to be marked also is invalid.
The present invention also provides a kind of method that is used for effective transmission signal of adaptive marking operation.The signal that sends two information is used for a reference picture marking is " being not used in reference ":
1) the prediction RPN of reference picture to be marked and the difference between the RPN value subtract 1, are expressed as diffRPNminus1, and
2) indicating this prediction is the estimation range of how to derive, and is expressed as PS.
Should select to make the only PS value in the scope between 0 to MaxFNplus1 of AbsDIFFminus1.
Prediction is derived as follows, is expressed as predRPN:
predRPN=RPNcurr-PS*MaxFNplus1
The RPN value of reference picture to be marked is derived as follows:
RPN=predRPN-(diffRPNminus1+1)
=RPNcurr-PS*MaxFNplus1-(diffRPNminus1+1)
For same example before, equal 255 reference picture marking and be " be not used in reference to " if will have RPN, then the information that sends as signal is diffRPNminus1=6, PS=3 (it illustrates with the label among Fig. 3 303).
Can derive reference picture to be marked has the RPN value and equals (1030-3*256-(6+1)=255).
Again, should state that the simple change of said method is always passable.For example, estimation range PS can be based on the value except MaxFNplus1, as long as can indicate this value according to codec standard and/or the relevant information that sends.
In the example system of Fig. 5, the value that the coding of scrambler 1 carries out image stream and calculating are used for parameter.If scrambler 1 is gone back the initializing signal transmission so that its demonstration or output time expire, then notify demoder 2 reference pictures of receiving equipment 8 from the back decoder buffer 2.1 of demoder, to remove.This signal comprises the reorder parameter of information and/or reference picture marking information of indication reference picture number, reference picture list.This signal is by transmitter 7 transmission of transmission equipment 6.
The present invention can be applied in the system and equipment of many types.Transmission equipment 6 for example can be the computing equipment such as server apparatus, video transmitter, Wireless Telecom Equipment etc.Receiving equipment 8 can be the computing equipment such as workstation, Wireless Telecom Equipment, video receiver etc.The transmission equipment 6 that comprises scrambler 1 also advantageously comprise transmitter 7 with the image of transfer encoding to transmission channel 4.Receiving equipment 8 comprises that receiver 9 is with image, the demoder 2 of received code and the display 10 that can show the image of decoding alternatively thereon.Transmission channel for example can be ground link communication channel and/or radio communication channel.Transmitting apparatus and receiving equipment also comprise one or more processor 1.2,2.2, and it can carry out the steps necessary that is used for the coding/decoding process of control of video stream according to the present invention.
Therefore, but the method according to this invention can mainly be embodied as the machine execution in step of processor.The buffering of image can be implemented in the storer 1.3,2.3 of equipment.The program code 1.4 of scrambler can be stored in the storer 1.3.Correspondingly, the program code 2.4 of demoder can be stored in the storer 2.3.
Claims (22)
1. the sequence to image is carried out Methods for Coding, and this method comprises:
Use one or more image as the reference image;
Utilize first parameter that described reference picture is carried out mark;
Described first parameter is sent to demoder as signal; And
Use reference picture management;
Wherein all described reference pictures are discerned by second parameter of deriving based on described first parameter.
2. method according to claim 1 comprises:
Use frame number FN as described first parameter, and
Use reference picture RPN as described second parameter.
3. method according to claim 2 comprises:
Define the decoding order of the image in the described image sequence;
Defined parameters prevFN equals the frame number of the last reference picture in the described decoding order;
Defined parameters prevRPN equals the reference picture number of last reference picture;
Define the maximal value of described frame number;
Defined parameters maxFNplus1 equals described maximal value+1 of described frame number; And
The described reference picture of the described reference picture of following calculating number:
if(prevFN<=FN)
RPN=prevRPN+FN-prevFN
else
RPN=prevRPN+FN-prevFN+MaxFNplus1
4. method according to claim 1, described reference picture management comprise that reference picture list initialization and reference picture list reorder.
5. method according to claim 4, described method comprise following parameter are sent as signal:
Parameter A bsDIFFminus1 represents the prediction of described RPN and the absolute difference between the described RPN value, and the prediction of wherein said RPN is the desired value of described RPN;
Parameter A Sidc represents that the described predicted value of described RPN adds or deduct described absolute difference with the described RPN value of deriving; And
Parameter PS represents the scope of the described predicted value of described RPN.
6. method according to claim 5 comprises:
Parameters R PNcurr is arranged to the value of the described RPN of first reference picture to be reordered;
The described predicted value predRPN of the reference picture that following calculating described first is to be reordered:
predRPN=RPNcurr-PS*MaxFNplus1
The RPN value that described predicted value predRPN at first equals the last reference picture that reorders is set; And
The described predRPN of following renewal:
if(ASidc==0)
predRPN=predRPN-PS*MaxFNplus1
else?if(PNidc==1)
predRPN=predRPN+PS*MaxFNplus1
7. method according to claim 1, described reference picture management comprises reference picture marking.
8. method according to claim 7, described method comprise following parameter are sent as signal:
Parameter d iffRPNminus1 represents that the described prediction of described RPN of described reference picture to be marked and the difference between the described RPN value subtract 1; And
Parameter PS represents the scope of described predicted value.
9. method according to claim 8 comprises:
Parameters R PNcurr is set to the described value of the described RPN of reference picture to be marked; And
The described reference picture of the described reference picture to be marked of following calculating number value RPN:
RPN=predRPN-(diffRPNminus1+1)
=RPNcurr-PS*MaxFNplus1-(diffRPNminus1+1)
10. one kind is used for method that the image encoded sequence is decoded, comprising:
Use one or more image as the reference image, described reference picture is carried out mark with first parameter;
Obtain described first parameter from described image encoded; And
The operation parameter image management;
Wherein all described reference pictures are discerned by second parameter of deriving based on described first parameter.
11. method according to claim 10, described reference picture management comprise that reference picture list initialization and reference picture list reorder.
12. method according to claim 10, described reference picture management comprises reference picture marking.
13. method according to claim 10, described reference picture management comprise that reference picture reorders and reference picture marking.
14. a signal that comprises the image encoded sequence, described sequence comprises one or more reference picture, with first parameter described reference picture is carried out mark; Use described signal according to claim 1.
15. hardware that is used to implement claim 1.
16. one kind is used for module that image sequence is encoded, comprises:
First module is used to select one or more image as reference picture;
Unit second is used to utilize first parameter that described reference picture is carried out mark;
Unit the 3rd is used for described first parameter is included in the signal of giving demoder to be transmitted; And
Unit the 4th is used for based on described first parameter second parameter of deriving; Wherein all described reference pictures are by described second parameter recognition.
17. module according to claim 16, wherein said module is included in the wireless device.
18. one kind is used for module that the image encoded sequence is decoded, described image comprises as one or more image with reference to image, utilizes first parameter that described reference picture is carried out mark; Described module comprises:
First module is used for obtaining described first parameter from described image encoded;
The parametric image manager; And
Unit second is used for deriving second parameter based on described first parameter and is used to discern all described reference pictures.
19. module according to claim 18, wherein said module is included in the wireless device.
20. a system comprises:
Encoding device is used for image sequence is encoded, and comprising:
First module is used to select one or more image as reference picture;
Unit second is used to utilize first parameter that described reference picture is carried out mark;
Unit the 3rd is used for described first parameter is included in the signal of giving demoder to be transmitted;
Unit the 4th is used for based on described first parameter second parameter of deriving; Wherein all described reference pictures are discerned by described second parameter; Decoding device is used for described signal is decoded, and described decoding device comprises
Unit the 5th is used for obtaining described first parameter from described image encoded; The reference picture management device; And
Unit the 6th is used for deriving second parameter with identification institute based on described first parameter
Described reference picture is arranged.
21. a computer program comprises being used for software that image sequence is encoded, described software comprises the machine executable code of being carried out by processor that is stored on the computer-readable recording medium, and described machine executable code is used for:
Use one or more image as the reference image;
Utilize first parameter that described reference picture is carried out mark;
Described first parameter is included in the signal waiting for transmission; And
Based on described first parameter second parameter of deriving; Wherein all described reference pictures are by described second parameter recognition.
22. a computer program comprises being used for software that image sequence is decoded, described software comprises the machine executable code of being carried out by processor that is stored on the computer-readable recording medium, and described machine executable code is used for:
Use one or more image as the reference image, utilize first parameter that described reference picture is carried out mark;
Obtain described first parameter from described image encoded;
Use reference picture management; And
Based on described first parameter second parameter of deriving; And
By all described reference pictures of described second parameter recognition.
Applications Claiming Priority (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US61897404P | 2004-10-14 | 2004-10-14 | |
US60/618,974 | 2004-10-14 | ||
US11/116,109 | 2005-04-26 |
Publications (1)
Publication Number | Publication Date |
---|---|
CN101065780A true CN101065780A (en) | 2007-10-31 |
Family
ID=38965735
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN 200580040403 Pending CN101065780A (en) | 2004-10-14 | 2005-10-13 | Reference picture management in video coding |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN101065780A (en) |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103814575A (en) * | 2011-09-23 | 2014-05-21 | 高通股份有限公司 | Coding reference pictures for a reference picture set |
CN103916666A (en) * | 2013-01-09 | 2014-07-09 | 北京大学 | Video coding method, video decoding method, coder and decoder |
CN104025596A (en) * | 2011-12-02 | 2014-09-03 | 高通股份有限公司 | Coding least signficant bits of picture order count values identifying long-term reference pictures |
CN104365106A (en) * | 2012-06-07 | 2015-02-18 | 高通股份有限公司 | Signaling data for long term reference pictures for video coding |
CN105325003A (en) * | 2013-04-17 | 2016-02-10 | 诺基亚技术有限公司 | An apparatus, a method and a computer program for video coding and decoding |
CN106454369A (en) * | 2011-10-18 | 2017-02-22 | 株式会社Ntt都科摩 | Motion video predicting decoding method and motion video predicting decoding device |
CN106717005A (en) * | 2014-09-23 | 2017-05-24 | 三星电子株式会社 | Video encoding/decoding method and device for controlling reference image data according to reference frequency |
CN107770548A (en) * | 2011-10-19 | 2018-03-06 | 太阳专利托管公司 | Coding/decoding method, decoding apparatus |
-
2005
- 2005-10-13 CN CN 200580040403 patent/CN101065780A/en active Pending
Cited By (29)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107770523A (en) * | 2011-09-23 | 2018-03-06 | 维洛媒体国际有限公司 | Reference picture list for video coding constructs |
US11490119B2 (en) | 2011-09-23 | 2022-11-01 | Qualcomm Incorporated | Decoded picture buffer management |
CN108174202B (en) * | 2011-09-23 | 2021-06-01 | 维洛媒体国际有限公司 | Method for encoding video data and video encoder |
CN108184123B (en) * | 2011-09-23 | 2021-06-01 | 维洛媒体国际有限公司 | Method for decoding video data and video decoder |
CN107770523B (en) * | 2011-09-23 | 2020-07-17 | 维洛媒体国际有限公司 | Method and apparatus for decoding video data |
CN108174202A (en) * | 2011-09-23 | 2018-06-15 | 维洛媒体国际有限公司 | It is constructed for the reference picture list of video coding |
CN107682695B (en) * | 2011-09-23 | 2020-06-30 | 维洛媒体国际有限公司 | Method and apparatus for encoding video data |
CN103814575A (en) * | 2011-09-23 | 2014-05-21 | 高通股份有限公司 | Coding reference pictures for a reference picture set |
CN108184123A (en) * | 2011-09-23 | 2018-06-19 | 维洛媒体国际有限公司 | It is constructed for the reference picture list of video coding |
US9998757B2 (en) | 2011-09-23 | 2018-06-12 | Velos Media, Llc | Reference picture signaling and decoded picture buffer management |
CN107682695A (en) * | 2011-09-23 | 2018-02-09 | 维洛媒体国际有限公司 | Reference picture list for video coding constructs |
CN106454369B (en) * | 2011-10-18 | 2019-07-23 | 株式会社Ntt都科摩 | Dynamic image prediction decoding method, dynamic image prediction decoding device |
CN103858429B (en) * | 2011-10-18 | 2018-11-27 | 株式会社Ntt都科摩 | Dynamic image predicts decoding method and device |
CN107105265B (en) * | 2011-10-18 | 2018-06-26 | 株式会社Ntt都科摩 | Dynamic image prediction decoding method, dynamic image prediction decoding device |
CN107105265A (en) * | 2011-10-18 | 2017-08-29 | 株式会社Ntt都科摩 | Dynamic image prediction decoding method, dynamic image prediction decoding device |
CN106454369A (en) * | 2011-10-18 | 2017-02-22 | 株式会社Ntt都科摩 | Motion video predicting decoding method and motion video predicting decoding device |
CN107770548B (en) * | 2011-10-19 | 2020-04-14 | 太阳专利托管公司 | Decoding method and decoding device |
CN107770548A (en) * | 2011-10-19 | 2018-03-06 | 太阳专利托管公司 | Coding/decoding method, decoding apparatus |
US9648321B2 (en) | 2011-12-02 | 2017-05-09 | Qualcomm Incorporated | Coding picture order count values identifying long-term reference frames |
CN104025596A (en) * | 2011-12-02 | 2014-09-03 | 高通股份有限公司 | Coding least signficant bits of picture order count values identifying long-term reference pictures |
CN104365106B (en) * | 2012-06-07 | 2018-04-17 | 高通股份有限公司 | Transmit the data of the long-term reference picture for video coding |
CN104365106A (en) * | 2012-06-07 | 2015-02-18 | 高通股份有限公司 | Signaling data for long term reference pictures for video coding |
CN103916666B (en) * | 2013-01-09 | 2017-07-14 | 北京大学 | Method for video coding, coding/decoding method, encoder and decoder |
CN103916666A (en) * | 2013-01-09 | 2014-07-09 | 北京大学 | Video coding method, video decoding method, coder and decoder |
CN105325003B (en) * | 2013-04-17 | 2019-05-28 | 诺基亚技术有限公司 | For Video coding and decoded device, method |
CN105325003A (en) * | 2013-04-17 | 2016-02-10 | 诺基亚技术有限公司 | An apparatus, a method and a computer program for video coding and decoding |
CN106717005A (en) * | 2014-09-23 | 2017-05-24 | 三星电子株式会社 | Video encoding/decoding method and device for controlling reference image data according to reference frequency |
CN106717005B (en) * | 2014-09-23 | 2020-08-18 | 三星电子株式会社 | Video encoding/decoding method and apparatus for controlling reference picture data according to reference frequency |
US11032554B2 (en) | 2014-09-23 | 2021-06-08 | Samsung Electronics Co., Ltd. | Video encoding/decoding method and device for controlling reference image data according to reference frequency |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN1178515C (en) | Image encoding method and device thereof | |
CN101065780A (en) | Reference picture management in video coding | |
CN1264333C (en) | Video coding method, video decoding method, video coding equipment, video decoding equipment, video coding program and video decoding program | |
CN1305311C (en) | Image encoder, image decoder, image encoding method, and image decoding method | |
CN101189881B (en) | Coding of frame number in scalable video coding | |
CN104412588B (en) | For signaling the equipment of long-term reference picture in parameter set | |
CN1134992C (en) | Picture predictive decoding/coding method and device, and data memory medium | |
US10021418B2 (en) | Method and apparatus of candidate generation for single sample mode in video coding | |
EP1784986B1 (en) | Parameter set and picture header in video coding | |
CN1229999C (en) | Method and device for code and decode motion vector | |
US20060083298A1 (en) | Reference picture management in video coding | |
CN1787640A (en) | Video coding, decoding and hypothetical reference decoder | |
CN1771734A (en) | Method, medium, and apparatus for 3-dimensional encoding and/or decoding of video | |
CN103299622A (en) | Method and apparatus of improved intra luma prediction mode coding | |
CN1126374C (en) | Image communication system and method | |
CN1825978A (en) | Frame compression using cardinal approximation or differential code and escape code | |
US20140098878A1 (en) | Method and device for motion information prediction in multi-view video coding | |
CN1529988A (en) | Image encoding method, image decoding method, image encoding device, image decoding device, program, computer dato signal and image transmission system | |
CN1299510C (en) | Moving-picture information compressing method and system | |
WO2020255668A1 (en) | Image coding device, image decoding device, image coding method, and image decoding method | |
CN101035284A (en) | Stream-type video pixel domain code conversion method | |
CN1225128C (en) | Method of determing reference image block under direct coding mode | |
CN101466037A (en) | Method for implementing video decoder combining software and hardware | |
WO2020255667A1 (en) | Image coding device, image decoding device, image coding method, and image decoding method | |
US20140362920A1 (en) | Image processing device and imaging processing method, program, and imaging apparatus |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
REG | Reference to a national code |
Ref country code: HK Ref legal event code: DE Ref document number: 1107169 Country of ref document: HK |
|
C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20071031 |
|
REG | Reference to a national code |
Ref country code: HK Ref legal event code: WD Ref document number: 1107169 Country of ref document: HK |