CN105245878A

CN105245878A - New planar prediction mode

Info

Publication number: CN105245878A
Application number: CN201510511522.2A
Authority: CN
Inventors: 林宰显; 全柄文; 朴胜煜; 成宰源; 金郑善; 全勇俊; 朴俊永; 崔瑛喜
Original assignee: LG Electronics Inc
Current assignee: LG Electronics Inc
Priority date: 2010-05-25
Filing date: 2011-05-25
Publication date: 2016-01-13
Anticipated expiration: 2031-05-25
Also published as: CN105245879A; CN105245878B; CN105245901B; CN105245879B; CN105245901A; CN105245902A; CN105245902B

Abstract

New intra planar modes are introduced for predicting digital video data. As part of the new intra planar modes, various methods are offered for predicting a first sample within a prediction unit, where the first sample is needed for referencing to when processing the new intra planar modes. And once the first sample is successfully predicted, the new intra planar modes are able to predict a sample of video data within the prediction unit by processing a bi-linear interpolation of four previously reconstructed reference samples.

Description

New plane prediction mode

The application is the international filing date submitted on November 26th, 2012 be the application number on May 25th, 2011 is 201180026091.0 (PCT/KR2011/003828), and denomination of invention is the divisional application of " new plane prediction mode " patent application.

Technical field

The present invention relates to the method and apparatus that a kind of digital of digital video data for encoding to using frame inner plane mode-type to predict (intraplanarmodetypeprediction) performs frame inner plane mode-type prediction decoding.

Background technology

Digital video signal is made up of a series of digital video frame of the expression as original rgb video signal.As a part for analog to digital signal conversion, each frame of original rgb video signal is encoded into the digital video frame of the data comprising digital video signal.The object of coded treatment is the digital forecast calculating accurate as far as possible original rgb video signal, and the compression simultaneously also attempting the binary data of the numeral made as original rgb video signal maximizes.Although there is inter prediction (interprediction) method and infra-frame prediction (intraprediction) method that are used for encoded video signal, the present invention is only relevant with the intra-frame prediction method being also called as space predicting method.

In order to realize coded treatment, process in a part for original video frame is predicted it is encoded into digital of digital video data by coding unit.The digital of digital video data of the coding obtained is called as predicting unit.Multiple predicting unit will generally include tree-like piece (treeblock) of video data, multiple tree-like piece will generally include the band (slice) of video data and then multiple band generally includes the frame of digital of digital video data, although other configuration is possible.Particularly about the intra-frame prediction method being fixed against spatial prediction, the sample by referring to the spatially previous prediction of adjacent current prediction unit is predicted just processed current prediction unit.Once all digital video frame predicted and coding, digital video programs is just known as and is fully have compressed, and has carried out the preparation storing as digital of digital video data or signal or transmit.Together with the digital of digital video data of reality, coding unit also comprises the identifying information which predictive mode of instruction has been used to each predicting unit of prediction video data.

Then decoding unit is assigned with the decoding of combine digital video data/signal or decompression work.Decoding processes by the same predictive mode process of being applied by coding unit being applied in each predicting unit.By resolving identifying information and being defined as predicting each predicting unit of video data and the suitable predictive mode that identifies realizes this.By by suitable predicted application in each predicting unit of video data, decoding unit successfully can rebuild original video.Decoding unit thus be assigned with the task of digital video signal being reconstructed into the shown expression of raw video signal.According to the intra prediction mode for predicting unit of decoding, the sample from the previous reconstruction of the predicting unit of previously having rebuild is rebuild being referenced at present just by the sample of the current prediction unit of decoding unit process.

Among the many available intra prediction mode of the predicting unit for predicted figure video data, the present invention is relevant to frame inner plane predictive mode.Usually know frame inner plane model prediction be first by referring to by the adjacent block rebuild in advance to predict the single sample in current prediction unit.Then after predicting the first sample in current prediction unit, by referring to the first sample predicted in current prediction unit and from the reconstruction sample of the block of adjacent current prediction unit to predict the residue sample of current prediction unit.

Summary of the invention

Technical problem

For needing to realize higher efficiency according to the new method of frame inner plane predictive mode process infra-frame prediction.Also exist and the sum of the information bit of vision signal is encoded into reduce the needs of the total data size of vision signal to minimizing.

The solution of problem

The object of this invention is to provide the various new methods of the prediction for obtaining the first sample in current prediction unit, wherein, current prediction unit is just predicted by new frame inner plane pattern according to the present invention.The various methods being provided for the residue sample just predicted in current prediction unit once the predicted and reconstruction of the first sample in current prediction unit are also objects of the present invention.According to the present invention, these methods are by by referring to the first sample and predict residue sample in current prediction unit from the sample of the previous reconstruction of the block of adjacent current prediction unit.

The beneficial effect of the invention

New method for the treatment of frame inner plane predictive mode according to the present invention can realize the more high efficiency surpassing previous version.The present invention can also reduce the total data size of vision signal.

It being understood that aforementioned general execution mode of the present invention and following embodiment are exemplary and explanat, and aim to provide as claimed of the present invention further explanation.

Accompanying drawing explanation

Fig. 1 illustrates the first embodiment according to the first sample for obtaining predicting unit of the present invention;

Fig. 2 illustrates the second embodiment according to the first sample for obtaining predicting unit of the present invention;

Fig. 3 illustrates the 3rd embodiment according to the first sample for obtaining predicting unit of the present invention;

Fig. 4 illustrates the 4th embodiment according to the first sample for obtaining predicting unit of the present invention;

Fig. 5 illustrates according to the first embodiment for prediction residue sample after the first sample is predicted of the present invention;

Fig. 6 illustrates according to the second embodiment for prediction residue sample after the first sample is predicted of the present invention;

Fig. 7 illustrates the zoomed-in view of a part of getting from Fig. 6;

Fig. 8 illustrates according to the 3rd embodiment for prediction residue sample after the first sample is predicted of the present invention;

Fig. 9 illustrates the zoomed-in view of a part of getting from Fig. 8;

Figure 10 illustrates according to the 4th embodiment for prediction residue sample after the first sample is predicted of the present invention;

Figure 11 illustrates according to the 5th embodiment for prediction residue sample after the first sample is predicted of the present invention;

Figure 12 illustrates according to the 6th embodiment for prediction residue sample after the first sample is predicted of the present invention;

Figure 13 illustrates according to the method being confirmed as important specific reference sample for filtering (filter) of the present invention;

Figure 14 illustrates the example according to the first sample for the treatment of predicting unit of the present invention;

Figure 15 illustrates the example of the position according to the first sample for determining predicting unit of the present invention;

Figure 16 illustrates another example of the position according to the first sample for determining predicting unit of the present invention;

Figure 17 illustrates the example of the position according to the first sample for determining predicting unit of the present invention;

Figure 18 illustrates according to of the present invention for receiving the method for reprocessing predicting unit of the first sample of predicting unit;

Relation between Figure 19 illustrates according to the block of the video data of predicting unit of the present invention and this predicting unit adjacent;

Figure 20 illustrates according to decoding unit of the present invention, and

Figure 21 illustrates the close-up illustration of the predicting unit of the part as decoding unit illustrated in Figure 20.

Embodiment

Now carry out detailed reference to the preferred embodiments of the present invention, the example of preferred embodiment is illustrated in the accompanying drawings.The term used in the present description and claims or word are not construed as limited to general or dictionary meanings, and should by can suitably define based on inventor term concepts with most probable inventor described intention and being interpreted as there is the meaning and concept of mating with technological thought of the present invention.The disclosed embodiments and configuration shown are in the accompanying drawings exemplary in essence in the disclosure, and are not intended to comprise all in essence.Preferred embodiment does not represent all possible technique variation of the present invention.Therefore, should be understood that the modifications and variations of this invention are contained in the present invention, as long as they fall in the scope of claims and their equivalent when submitting the application to.

It should be noted that the object in order to illustrate as follows, all of adjacent block describe the block be understood to be about adjacent current prediction unit.Current prediction unit is understood to include the just current predictive sample of predicted process according to new frame inner plane pattern of the present invention.Distance H refers in the height of the current prediction unit of sample, and W refers in the width of the current prediction unit of sample.Similarly any gray area is illustrated as the sample representing and previously rebuild.The sample (reconstructedsample) rebuild is predicted and merge with the residual values be included in video signals (residualvalue), and then such reconstruction sample can be referenced for carrying out forecast sample according to the present invention.The sample (predictedsample) of prediction still not yet merges with residual values to be reconstructed, but predicted process and can being referenced for carrying out forecast sample according to the present invention.Any use of term " sample " is understood to exchange with known term " pixel ".

Fig. 1 illustrates for predicting that the first current predictive sample RB in current prediction unit 101 is for first embodiment of carrying out reference in new frame inner plane mode prediction method according to the present invention.See that current prediction unit 101 has and be positioned at contiguous and on the left side of current prediction unit 101 adjacent left part block 102.And see that current prediction unit 101 has and be positioned at contiguous and on the top of current prediction unit 101 adjacent upper block 103.Both adjacent left part block 102 and adjacent upper block 103 are depicted as gray area to indicate them by the composition of sample of previously having rebuild.Although only the first current predictive sample RB is illustrated particularly, current prediction unit 101 comprises the predicted and final multiple current predictive samples rebuild of needs in addition.In addition, the first current predictive sample RB sample that will be the right lower quadrant place being positioned at current prediction unit 101.

According to the first embodiment for predicting the first current predictive sample RB in current prediction unit 101, the first current predictive sample RB will be predicted based on DC model prediction in frame.The sample value of DC model prediction with reference to sample T and the sample value of sample L in frame, and determine the mean value of these two sample values.From adjacent upper block 103 with reference to reference sample T, and from adjacent left part block 102 with reference to reference sample L.Then according to DC model prediction in frame, the average sample value of reference sample T and L is using as the predicted value being used for the first current predictive sample RB.Utilize now DC model prediction value in this frame, the predicted value for RB can be referenced the residue sample predicting current prediction unit 101.Although Fig. 1 identifies the position of reference sample L and T particularly, but come with reference to reference sample L also within the scope of the invention from any one sample from adjacent left part block 102, and come with reference to reference sample T also within the scope of the invention from any one sample from adjacent upper block 103.

Fig. 2 illustrates for predicting that the first current predictive sample RB in current prediction unit 201 is for second embodiment of carrying out reference in new frame inner plane mode prediction method according to the present invention.According to this second embodiment, the mean value of the reference sample 2W based on the reference sample 2H previously rebuild and previously reconstruction is predicted the first current predictive sample RB.Depend on the size of the adjacent block on the left side in current prediction unit 201, reference sample 2H can be taken from adjacent left part block or adjacent lower left quarter block.Depend on the size of the adjacent block of the top in current prediction unit 201, reference sample 2W can be taken from adjacent upper block or adjacent upper right quarter block.According to this average sample value of this second embodiment reference sample 2H and 2W using as the predicted value being used for the first current predictive sample RB.Now by referring to this prediction for the first current predictive sample RB, the residue sample of current prediction unit 201 can be predicted.When reference sample 2H is not available, in order to the object of this second embodiment, be alternative in reference sample 2H, can with reference to the reference sample being positioned at distance H place in adjacent left part block.When reference sample 2W is not available, in order to the object of this second embodiment, be alternative in reference sample 2W, can with reference to the reference sample being positioned at distance W place in adjacent upper block.

Fig. 3 illustrates for predicting that the first current predictive sample RB in current prediction unit 301 is for the 3rd embodiment of carrying out reference in new frame inner plane mode prediction method according to the present invention.According to this 3rd embodiment, the mean value from four reference samples previously rebuild of adjacent block will be referenced to predict the first current predictive sample RB.Particularly, reference sample T and reference sample 2T will be referenced to obtain average sample value T '.Although from adjacent upper block with reference to reference sample T, depend on the size of adjacent upper block, can from same adjacent upper block or independent adjacent upper right quarter block reference reference sample 2T.Similarly, reference sample L and 2L will be referenced to obtain average sample value L '.Although reference sample L is obtained by from adjacent left part block, depend on the size of adjacent left part block, can from same adjacent left part block or independent lower left quarter adjacent block reference reference sample 2L.Then, according to the prediction of the 3rd embodiment first current predictive sample RB will be the mean value of T ' and L '.Now by referring to this prediction of the first current predictive sample RB, the residue sample of current prediction unit 301 can be predicted.

The function of the mean value for determining two reference samples is represented in figure 3 by shift function (shiftfunction).The mean value of such as reference sample L and 2L is by following expression:

L^{,} = (L + 2 L + 1) > > 1

According to above-mentioned shift function, value 1 is added to the summation of reference sample L and 2L, potential choice error (roundingerror) to be described, but abandons adding 1 also within the scope of the invention.And double-head arrow shift function >> represents by removing with two the mean value that summation gets summation.Utilize shift function, can be understood to carry out computing by similar mode for the residue calculating of RB with T ' average sample value illustrated in Fig. 3.

Fig. 4 illustrates for predicting that the first current predictive sample RB in current prediction unit 401 is for the 4th embodiment of carrying out reference in new frame inner plane mode prediction method according to the present invention.According to this 4th embodiment, obtain the predicted value of the first forecast sample RB by referring to two reference samples from adjacent block.Particularly, depend on the size of adjacent upper block, can come with reference to the first reference sample T ' from the adjacent block that can be adjacent upper block or adjacent upper right quarter block.The distinctive feature of the first reference sample T ' comes with reference to it between sample T and the 2T previously rebuild.Depend on the size of adjacent left part block, come with reference to the second reference sample L ' from the adjacent block that can be adjacent left part block or adjacent lower left quarter block.The distinctive feature of the second reference sample L ' comes with reference to it between sample L and the 2L previously rebuild.Then from the mean value T ' and L ' of reference sample, obtain the predicted value of the first current predictive sample RB.This is expressed as by the shift function in Fig. 4:

R B = (T^{,} + L^{,} + 1) > > 1

According to above-mentioned shift function, value 1 is added to the summation of reference sample T ' and L ', potential rounding error to be described, but abandons adding 1 also within the scope of the invention.And double-head arrow shift function >> represents by removing with two the mean value that summation gets summation.

After obtaining the predicted value of the first current predictive sample RB by the value of average reference sample T ' and L ', the first current predictive sample RB can be referenced the residue sample predicted for new intra-mode prediction according to the present invention in current prediction unit 401.

Fig. 5 illustrates the first embodiment for predicting the residue sample in current prediction unit 501 after the prediction first obtaining the first current predictive sample RB.Once any method having thus described the invention predicts the first current predictive sample RB, will be referenced to perform linear interpolation (linearinterpolation) from the sample L of adjacent left piece 502 with from the sample RB in current prediction unit 501.Result from the linear interpolation of reference sample L and reference sample RB is called as LI1 in Figure 5.The predicted value LI1 obtained from the linear interpolation of reference sample L and RB is then by the predicted value of flatly filling as all current predictive samples for the bottom line (bottomrow) along current prediction unit 501.These samples along the bottom line of current prediction unit 501 can be referenced now for the residue sample in prediction current prediction unit 501.

Similarly, from the sample T of adjacent upper block 503 with will be referenced to perform linear interpolation from the sample RB in current prediction unit 501.The predicted value obtained from the linear interpolation of reference sample T and reference sample RB is represented as LI2 in Figure 5.Then the predicted value LI2 obtained from the linear interpolation of reference sample T and RB is filled as vertically along the predicted value of all current predictive samples of most right part row (right-mostcolumn) of current prediction unit 501.These samples vertically along the most right part row of current prediction unit 501 can be referenced now for the residue sample in prediction current prediction unit 501.

Continue Fig. 5, by predicting the residue sample in not yet predicted current prediction unit 501 from the bilinear interpolation (bi-linearinterpolation) from four reference samples got among left part adjacent block 502, top adjacent block 503 and the sample from the previous reconstruction in current prediction unit 501.In order to exemplary purpose, Fig. 5 depicts the prediction processing for current predictive sample C.Therefore according to the first embodiment, the bilinear interpolation that the prediction for current predictive sample C will be reference sample Ly, LI2, Tx and LI1.Can see, reference sample Ly is the sample of the previous reconstruction from the adjacent left part block 502 adjacent with current prediction unit 501, and shares the y coordinate identical with current predictive sample C.Reference sample LI2 is the sample of the prediction in the current prediction unit 501 predicted from the linear interpolation by reference to sample L and RB.Can also see, reference sample Tx is the sample of the previous reconstruction from the adjacent upper block 503 adjacent with current prediction unit 501, and shares the x coordinate identical with current predictive sample C.Reference sample LI1 is the sample of the prediction in the current prediction unit 501 predicted from the linear interpolation by reference to sample L and RB.The bilinear interpolation of Ly, LI2, Tx and LI1 will side by side be processed to obtain the predicted value of current predictive sample C.By the remaining current predictive sample predicted according to the first embodiment by the mode similar with current predictive sample C in current prediction unit 501.

As an alternative, the bilinear interpolation of current predictive sample C can also be processed by average two independent linear interpolations.According to this substitute, the linear interpolation with reference sample Tx and LI1 is processed dividually the linear interpolation of reference sample Ly and LI2, and then the prediction of reference sample C by the mean value based on these two linear interpolations.Remaining current predictive sample is predicted by substituting according to this of the first embodiment by the mode similar with current predictive sample C.

Fig. 6 illustrates the second embodiment for predicting the residue sample in current prediction unit 601 after the prediction first obtaining the first current predictive sample RB.In figure 6, what see is that mean value by getting reference sample 2L and reference sample 2T predicts the first current predictive RB.After predicting the first current predictive sample RB, the remaining current predictive sample along the bottom line of current prediction unit 601 is copied by from adjacent block as depicted in figure 6.Depend on the size of adjacent left part block, the reference sample being referenced the bottom line copying current prediction unit 601 can from the adjacent left part block relevant to current prediction unit 601 or adjacent lower left quarter block.Similarly, the remaining current predictive sample along the most right part vertical row of current prediction unit 601 is copied by from adjacent block as depicted in figure 6.Depend on the size of adjacent upper block, the reference sample being referenced the most right part vertical row copied in current prediction unit 601 can from adjacent upper block or adjacent upper right quarter block.

Fig. 7 is the close-up illustration of the current prediction unit 601 of seeing in figure 6.By predicting from adjacent block reproduction copies as seen in Figure 6 along after the bottom line of current prediction unit 601 and the sample of most right part vertical row, predict the residue sample in current prediction unit 601 according to bilinear interpolation.By referring to four samples taking from left part adjacent block 602, top adjacent block 603 and take from current prediction unit 601 to process bilinear interpolation.In order to exemplary purpose, Fig. 7 depicts the prediction processing of current predictive sample C.Therefore according to this second embodiment, the bilinear interpolation that the prediction of current predictive sample C will be reference sample Ly, RC, Tx and BC.Reference sample Ly is the sample from the adjacent left part block 602 adjacent with current prediction unit 601, and on the y coordinate line identical with current predictive sample C.Reference sample RC is from the sample in current predictive sample 601, this current forecast sample 601 was as predicting by copying reference sample from adjacent block illustrated by Fig. 6, and reference sample RC is also on the y coordinate line identical with current predictive sample C.Reference sample Tx is the sample from the adjacent upper block 603 adjacent with current prediction unit 601, and on the x coordinate line identical with current predictive sample C.Reference sample BC is from the sample in current predictive sample 501, this current forecast sample 501 was as predicting by copying reference sample from adjacent block illustrated by Fig. 6, and reference sample BC is also on the x coordinate line identical with current predictive sample C.The bilinear interpolation of Ly, Rc, Tx and Bc will side by side be processed to obtain the predicted value of current predictive sample C.Remaining current predictive sample is predicted according to this second embodiment by by the mode similar with current predictive sample C.

As an alternative, the bilinear interpolation of current predictive sample C can also be processed by average two independent linear interpolations.According to this substitute, the linear interpolation with reference sample Tx and Bc is processed dividually the linear interpolation of reference sample Ly and Rc, and then the prediction of current reference sample C by the mean value based on these two linear interpolations.Remaining current predictive sample is predicted by substituting according to this of the second embodiment by similar mode.

Fig. 8 illustrates the 3rd embodiment for predicting the residue sample in current prediction unit 801 after the prediction first obtaining the first current predictive sample RB.After predicting the first current predictive sample RB according to any one in previously mentioned embodiment above, by utilizing the remaining current predictive sample of filling the bottom line belonging to current prediction unit 801 from the linear interpolation of two reference samples of two adjacent block references.In order to exemplary purpose, for predicting that the example of current predictive sample B 5 predicts the illustrative examples along the sample of the bottom line of current prediction unit 801 by being described for.Can see from Fig. 8, be referenced to predict that two reference samples of current predictive sample B 5 are placed along public linea angulata (commonangularline), and be from two independent adjacent blocks.Depend on the size of adjacent left part block, as the first reference sample of being indicated by arrow α can by from adjacent left part block or the reference of adjacent lower left quarter block.Depend on the size of adjacent upper block, as the second reference sample of being indicated by arrow 1-α can by from adjacent upper block or the reference of adjacent upper right quarter block.Therefore the first reference sample and the second reference sample by by interpolation linearly to obtain the predicted value being used for current predictive sample B 5.The remaining current predictive sample of the bottom line along current prediction unit 801 will be predicted in a similar manner by referring to two that place along corresponding linea angulata independent reference samples.Can then when predicting the residue sample in current prediction unit 801 with reference to the current predictive sample of prediction.

What similarly describe in fig. 8 is the Forecasting Methodology of the sample for the most right part vertical row in current prediction unit 801 according to this 3rd embodiment.In order to exemplary purpose, the illustrative examples of the sample of most right part vertical row that the example of prediction current predictive sample R2 will be described for predicting along current prediction unit 801.Can see from Fig. 8, be referenced to predict that two reference samples of current predictive sample R2 are placed along public linea angulata, and be from two independent adjacent blocks.Depend on the size of left part block, as the first reference sample of being indicated by arrow α can by from adjacent left part block or the reference of adjacent lower left quarter block.Depend on the size of upper mass, as the second reference sample of being indicated by arrow 1-α can by from adjacent upper block or the reference of adjacent upper right quarter block.Therefore the first reference sample and the second reference sample by by interpolation linearly to obtain the predicted value being used for current predictive sample R2.The residue sample of the most right part vertical row along current prediction unit 801 will be predicted in a similar manner by referring to two that place along corresponding linea angulata independent reference samples.Can then when predicting the residue sample in current prediction unit 801 with reference to the current predictive sample of prediction.

Fig. 9 is the close-up illustration of the current prediction unit 801 of seeing in fig. 8, wherein along the sample of bottom line and the sample along most right part vertical row predicted.By predicting the residue sample in not yet predicted current prediction unit 801 from the bilinear interpolation from four reference samples got among left part adjacent block 802, top adjacent block 803 and the sample from the previous reconstruction in current prediction unit 801.In order to exemplary purpose, Fig. 9 depicts the example of the prediction processing of current predictive sample C.Therefore according to this 3rd embodiment, the bilinear interpolation that the prediction of current predictive sample C will be reference sample Ly, RC, Tx and BC.Reference sample Ly is the sample from the adjacent left part block 802 adjacent with current prediction unit 801, and on the y coordinate line identical with current predictive sample C.As illustrating above with reference to Fig. 8, reference sample RC is from the sample in current predictive sample 801, this current forecast sample 801 based on two samples from adjacent block placed along public linea angulata linear interpolation and predicted, and reference sample RC is also on the y coordinate line identical with current predictive sample C.Reference sample Tx is the sample from the adjacent left part block 803 adjacent with current prediction unit 801, and on the x coordinate line identical with current predictive sample C.As illustrating above with reference to Fig. 8, reference sample BC is from the sample in current predictive sample 801, this current forecast sample 801 based on two samples from adjacent block placed along public linea angulata linear interpolation and predicted, and reference sample RC is also on the x coordinate line identical with current predictive sample C.The bilinear interpolation of Ly, Rc, Tx and Bc will side by side be processed to obtain the predicted value of current predictive sample C.Remaining current predictive sample is predicted according to this 3rd embodiment by by the mode similar with current predictive sample C.

As an alternative, the bilinear interpolation of current predictive sample C can also by getting the linear interpolation of reference sample Ly and RC and getting the linear interpolation of reference sample Tx and BC individually, and the mean value then getting these two linear interpolations processes.Remaining current predictive sample is predicted by substituting according to this of the 3rd embodiment by similar mode.

Figure 10 illustrates the 4th embodiment for predicting the residue sample in current prediction unit 1001 after the prediction first obtaining the first current predictive sample RB.After predicting the first current predictive sample RB according to any one in previously mentioned embodiment above, fill/predict from the bilinear interpolation of four reference samples of adjacent block institute reference all remaining current predictive sample belonging to current prediction unit 1001 by utilizing.Therefore without the need to calculating the prediction of the current predictive sample that is used for placing along the bottom line of current prediction unit 1001 and the current predictive sample along the most right part vertical row placement of current prediction unit 1001 individually.In order to exemplary purpose, Figure 10 depicts the prediction processing of the current predictive sample C according to this 4th embodiment.

In order to process the prediction of current predictive sample C, the bilinear interpolation of four reference samples will be made: Ly, 2T, Tx and 2L.Reference sample Ly is the sample from the adjacent left part block adjacent with current prediction unit 1001, and on the y coordinate line identical with current predictive sample C.Reference sample 2T is only copying of the reference sample 2T of the previous reconstruction of taking from adjacent upper block or adjacent upper right quarter block.Reference sample Tx is the sample from the adjacent upper block adjacent with current prediction unit 1001, and on the x coordinate line identical with current predictive sample C.Reference sample 2L is only copying of the reference sample 2L of the previous reconstruction of taking from adjacent left part block or adjacent lower left quarter block.Then the bilinear interpolation of Ly, 2T, Tx and 2L will side by side be processed to obtain the predicted value being used for current predictive sample C.Remaining current predictive sample is predicted according to this 4th embodiment by the mode similar with current predictive sample C.

It should be noted that according to this 4th embodiment, reference sample 2L and 2T is by for being the current predictive sample in prediction the current prediction unit 1001 and each bilinear interpolation made keeps constant.But reference sample Ly and Tx will change x and the y coordinate to correspond to just predicted current predictive sample.

Substitute as first of this 4th embodiment, bilinear interpolation can also be treated to the mean value of two independent linear interpolations.According to this first substitute, get the first linear interpolation by referring to sample Ly and 2T.And take the second linear interpolation by referring to sample Tx and 2L.The prediction of current predictive sample C will be the mean value obtained from first linear interpolation of sample Ly and 2T and second linear interpolation of sample Tx and 2L.Remaining current predictive sample is predicted by substituting according to this of the 4th embodiment by similar mode.

In substituting for second of this 4th embodiment, after predicting the first current predictive sample RB and before prediction current predictive sample C, the current predictive sample be placed in the bottom line of current prediction unit 1001 can utilize 2L copy fill.And similarly, the current predictive sample be placed in the most right part row of current prediction unit 1001 can utilize 2T copy fill.

Figure 11 illustrates the 5th embodiment for predicting the residue sample in current prediction unit 1101 after the prediction first obtaining the first current predictive sample RB.After any one method according to the present invention predicts the first current predictive sample RB, all remaining current predictive sample in current prediction unit 1101 is predicted in the second linear interpolation of the first linear interpolation by average two reference samples previously rebuild and two reference samples previously rebuild from adjacent block.

In fig. 11, by referring to be positioned at the y coordinate place identical with current predictive sample C, from the reference sample Ly of the previous reconstruction of adjacent left part block reference, and with reference to from current prediction unit 1101, make the first linear interpolation as the sample T copied of the previous reconstruction sample T from the adjacent upper block adjoined with upper right quarter sample.By referring to be positioned at the x coordinate place identical with current predictive sample C, from the reference sample Tx of the previous reconstruction of adjacent upper block, and with reference to from current prediction unit 1101, make the second linear interpolation as the sample L copied of the previous reconstruction sample L from the adjacent left part block adjoined with lower left quarter sample.

First linear interpolation is added to the second linear interpolation, and then average, to obtain the prediction of current predictive sample C.Average function passes through graphical presentation by dextroposition function G reatT.GreaT.GT> in fig. 11.Then the prediction of current predictive sample C is the average of the first linear interpolation and the second linear interpolation.Residue sample will be predicted in a similar manner.First linear interpolation by always with reference to from the reproduction copies T of adjacent upper block, and then alternatively with reference to have the y coordinate identical with current predictive sample C, from the sample Ly of the previous reconstruction of adjacent left part block.And second linear interpolation by always with reference to from the reproduction copies L of adjacent left part block, and then alternatively with reference to having the x coordinate identical with current predictive sample C, sample Tx from the previous reconstruction of adjacent upper block.

In the substituting of this 5th embodiment, after predicting the first current predictive sample RB, before prediction current predictive sample C, the current predictive sample copied to fill in the bottom line being placed on current prediction unit 1101 of reference sample L can be utilized.And similarly, can utilize that reference sample T's copy to fill the current predictive sample be placed in the most right part row of current prediction unit 1101.

Figure 12 illustrates the 6th embodiment for predicting the residue sample in current prediction unit 1201 after the prediction first obtaining the first current predictive sample RB.After any one method according to the present invention predicts the first current predictive sample RB, the combination of the linear interpolation based on two weightings is predicted by all remaining current predictive sample in current prediction unit 1201.

In fig. 12, by referring to be positioned at the y coordinate place identical with current predictive sample C, from the reference sample Ly of the previous reconstruction of adjacent left part block institute reference, and make the first linear interpolation with reference to the sample T copied as the previous reconstruction sample T of the adjacent upper block adjoined from the upper right quarter sample with current prediction unit 1201.By referring to be positioned at the x coordinate place identical with current predictive sample C, from the reference sample Tx of the previous reconstruction of adjacent upper block institute reference, and make the second linear interpolation with reference to the sample L copied as the previous reconstruction sample L of the adjacent left part block adjoined from the lower left quarter sample with current prediction unit 1201.

Each weighted value that will be assigned with them in present first linear interpolation and the second linear interpolation.Figure 12 illustrates the first linear interpolation will carry out weighting by the first weighted value WH, and the second linear interpolation will carry out weighting by the second weighted value WV.Therefore, the prediction Pc of current predictive sample C will be summation:

Pc=WHx (the first linear interpolation)+WVx (the second linear interpolation)

Assuming that current predictive sample C has coordinate (x, y) in current prediction unit 1201, then the prediction PC (x, y) of current predictive sample C will follow following equation:

PC (x, y)=(y+1)/(x+y+2) (the first linear interpolation)+(x+1)/(x+y+2) (the second linear interpolation)

According to above-mentioned equation, be (y+1)/(x+y+2) by the value being applied to the WH of the first linear interpolation.

And be (x+1)/(x+y+2) by the value being applied to the WV of the second linear interpolation.

In the substituting of this 6th embodiment, after predicting the first current predictive sample RB, before prediction current predictive sample C, the current predictive sample copied to fill in the bottom line being placed on current prediction unit 1201 of reference sample L can be utilized.And similarly, can utilize that reference sample T's copy to fill the current predictive sample be placed in the most right part row of current prediction unit 1201.

Figure 13 illustrates the new method of the specific sample for filtering adjacent current prediction unit, wherein, and can with reference to the sample being selected for filtration when the frame inner plane model prediction new according to process of the present invention.The adjacent sample T that upper right quarter sample that is usual and current prediction unit 1301 adjoins will be by one of sample of importantly reference during new frame inner plane model prediction.Similarly, the adjacent sample L that right lower quadrant sample that is usual and current prediction unit 1301 adjoins will be by another sample of importantly reference during new frame inner plane model prediction.Therefore, there is the expectation making two reference sample T and L priorization.Therefore as the method making reference sample T and L priorization, the invention provides the solution being processed reference sample T and L by filtration treatment, to improve the flatness of the forecast sample of the gained in current prediction unit 1301, and improve the efficiency of overall video Signal Compression.

Therefore, Figure 13 is the diagrammatic illustration being processed reference sample T and L by 1:2:1 filtration treatment.In order to realize according to filtration treatment of the present invention, also the sample of the previous reconstruction adjoined with reference sample T and L will be utilized.Such as, the sample T-1 adjoined with the left part of reference sample T and the T+1 adjacent with the right part of reference sample T has been highlighted in fig. 13.Similarly, with the sample L-1 of the upper adjacent of reference sample L and be highlighted in fig. 13 with the bottom contiguous L+1 of reference sample L.The reference sample produced by the filtration treatment of reference sample T and L will be hereinafter referred to as T ' and L '.Then predict in new frame inner plane pattern according to the present invention that in fact sample T ' and the L ' of these filtrations during the sample in current prediction unit 1301 will be referenced.

The 1:2:1 filtration treatment of reference sample T is applied to according to following equation process:

T^{''} = \frac{(1 \times T_{- 1}) + (2 \times T) + (1 \times T_{+ 1}) + 2}{4}

Because in the most main level other places of numerical data, actual sample value is represented by a string binary bits, so above-mentioned equation can be write by shift function with the binary bits of the value representing reference sample T.This equation write by shift function can be write as follows:

T^{''} = {T_{- 1} + (T < < 1) + T_{+ 1} + 2} > > 2

With reference to the above-mentioned equation write by shift function, (<<1) shifts left the single displacement represented left, and this is mathematically understood to be equivalent to two times or be multiplied by two.(>>2) dextroposition represents two displacements to the right, and this is mathematically understood to be equivalent to divided by four.

Similarly, the 1:2:1 filtration treatment of reference sample L is applied to according to following equation process:

L^{''} = \frac{(1 \times L_{- 1}) + (2 \times L) + (1 \times L_{+ 1}) + 2}{4}

This equation can also be represented by shift function by following:

L^{''} = {L_{- 1} + (L < < 1) + L_{+ 1} + 2} > > 2

After the filtration having processed reference sample L and T, the value L ' after new filtration and T ' will replace original reference sample L and T.By doing like this, with reference to the reference sample L ' after new filtration and T ' when can process new frame inner plane model prediction in current prediction unit 1301.

But if one in the sample adjoined with reference sample T and L is not available, then the value for this reference sample will need by weighting to a greater degree.If the sample T+1 such as adjoined with the right part of reference sample T is not available, then reference sample T can experience the 1:3 filtration treatment processed according to following equation:

T^{''} = \frac{(1 \times T_{- 1}) + (3 \times T) + 2}{4}

Be apparent that from the filtration equation of revision, relative with original twice, the value for reference sample T is weighted three times, to compensate as disabled sample T+1.Can be write as follows by this new filtration treatment of shift function:

T^{''} = {T_{- 1} + (T < < 1) + T + 2} > > 2

Similarly, if the sample T-1 adjoined with the left part of reference sample T is not available, then reference sample T can experience the 3:1 filtration treatment processed according to following equation:

T^{''} = \frac{(3 \times T) + (1 \times T_{+ 1}) + 2}{4}

Can be write as follows by this new filtration treatment of shift function:

T^{''} = {(T < < 1) + T + T_{+ 1} + 2} > > 2

When any one in adjacent sample L+1 or L-1 is not available, the compensation filtration treatment of same type goes for filtering reference sample L.

Produce more level and smooth prediction with reference to the reference sample L ' after new filtration and T ' by for the sample in current prediction unit 1301, and improve the compression efficiency of digital video signal.

All embodiments described till now require after decoded unit receives, and the right lower quadrant sample in current prediction unit is the first forecast sample to be predicted.But, according to substituting aspect of the present invention, a sample in the predicting unit that new frame inner plane pattern according to the present invention can be predicted by the coding unit at first raw video signal being encoded into the predicting unit of video data remains in reconstruction state.Be in the first forecast sample of reconstruction state by then transmitting, this by from perform prediction with obtain during new frame inner plane model prediction, be used for this first forecast sample of reference task in the middle of liberate decoding unit.This first forecast sample has been described to for the interpolation process required by new frame inner plane model prediction according to the present invention be above required.

In addition, this substituting aspect of the present invention also allows any some place being positioned at predicting unit by this first predicted value transmitted as reconstruction sample.Although previous example this first forecast sample contemplated in predicting unit is positioned at the right lower quadrant corner of predicting unit always, this substituting aspect of the present invention allows this first forecast sample to be positioned at any position.

In order to exemplary purpose, Figure 14 illustrates the first current predictive sample C being in reconstruction state, this first current predictive sample C drawn gray shade with indicate it decoded unit receive.In addition, from adjacent left part block 1402 and adjacent upper block sample by picture gray shade with indicate they previous decoded unit rebuild.The all right Adjoint Coordinates information of the first current predictive sample C transmitted with state of rebuilding, the position of the first current predictive sample C in this coordinate information instruction current prediction unit 1401.If do not comprised in video signals by the part as identifying information by the coordinate of the first current predictive sample C transmitted using reconstruction state, then decoding unit has the various options for determining coordinate.

Usually, the coding unit side of predicting unit is encoded at raw video signal, if for the position that identifies first forecast sample that will transmit with reconstruction state information not by with together with the vision signal that transmits is included in, then the position of the first forecast sample and the edge along adjacent block are coincided (coincide) by coding unit.This is desired practice, because exist for receiving predicting unit to detect the various methods of the decoding unit at these edges.The position being used for the first forecast sample C can be selected to see that this puts into practice the diagram of the Figure 15 coincided with edge 1504 and 1505 from coding unit.Edge is the edge of the adjacent block adjacent with predicting unit 1501 substantially.Therefore edge 1504 is edges of adjacent left part block 1502, and edge 1505 is the edges from adjacent upper block 1503.These edges of segmentation due to adjacent block are formed to be less than the size of predicting unit 1501.If therefore coding unit does not transmit for identifying by the information of the position of the forecast sample C transmitted with reconstruction state significantly in vision signal, then coding unit must select the position in the predicting unit that easily can be identified by decoding unit.The edge of adjacent block provides such gauge point that easily can be identified by receipt decoding unit.Therefore coding unit will do basis to correspond to these neighboring edges, because a block end and the decoding unit that each marginal point at next BOB(beginning of block) place can be received the vision signal comprising predicting unit 1501 easily detect with the coordinate of the first forecast sample.

Figure 16 illustrates decoding unit how can a method of Edge detected.In Figure 16 illustrated method can by the sample of the previous reconstruction finding two vicinities have in sample value maximum difference place point locate marginal position.The sample of each reconstruction has its corresponding sample value.Then because each adjacent block can be assumed that and predicted uniquely and rebuild, although so can suppose the sample value of the reconstruction sample share similar in same, the sample belonging to the block of separation is by the sample value of not share similar.Therefore infer the marginal point place becoming the beginning of next contiguous block in the end of a block, compare two adjacent sample, one of them sample belongs to first piece and the second sample belongs to second piece, will produce maximum difference in sample value.Therefore in figure 16, decoding unit can calculate difference for all adjacent sample, and this all adjacent sample composition just in time lines up the row of the sample of line (lineup) on the top of current prediction unit 1601.Then, find in the sample value of decoding unit between two adjacent sample maximum poor time, this can be considered to be in the point that there is edge between two contiguous blocks.And correspondingly, this is the point that decoding unit can think the x coordinate for the first forecast sample C received with reconstruction state.Similarly, decoding unit will calculate difference for all adjacent sample, and this all adjacent sample composition just in time lines up the row of the sample of line at the left part of current prediction unit 1601.Then, in the sample value of decoding unit between two adjacent sample, find the place of maximum difference, this can be considered to be in the point that there is edge between two contiguous blocks.And correspondingly, this is the point that decoding unit can think the y coordinate for the first forecast sample C received with reconstruction state.

The other method of the coordinate of the first forecast sample that the mode for determining the state of rebuilding transmits is that decoding unit receives predicting unit to resolve the carve information for adjacent block.Usually, will be made up of video data and identifying information the vision signal of encoding unit encodes.Predicting unit can be considered to a part for video data, and carve information can be considered to a part for identifying information.How each piece of carve information identification video data be divided into the more fritter of video data.Tree-like piece (treeblock) of such as data can be divided into multiple encoding block, and then each encoding block can be divided into multiple prediction block/unit, and then each prediction block/unit can be divided into multiple transform block/unit etc.Partitioning video data is become the non-square region of sample data also within the scope of the invention, geometry block comminute information can also be included as a part for identifying information in this case.

Under any circumstance, such carve information is transmitted together with the predicting unit comprising video data in video signals by the part as identifying information.Therefore, after receiving the vision signal comprising predicting unit and carve information, decoding unit can resolve this carve information to determine the cut size of each predicting unit of adjacent current prediction unit.This is described in fig. 17.

In fig. 17, if decoding unit receives current prediction unit 1701 and new frame inner plane pattern just according to the present invention processes prediction in current prediction unit 1701, then decoding unit also will receive and has resolved the carve information relevant to the block of adjacent current prediction unit 1701.Therefore carve information will identify the size being contiguous to each block of the video data on the top of current prediction unit 1701.And from this carve information, decoding unit can identify and where be partitioned into now between adjacent upper block NB1 and NB2.And in from then on determining, decoding unit can follow obtained edge line to determine the x coordinate of the first current predictive sample C.Similarly, carve information will identify the size being contiguous to each block of the video data of the left part of current prediction unit 1701.And from this carve information, decoding unit can identify and where be partitioned into now between adjacent left part block NB3 and NB4.Then where be partitioned into now by being aware of between adjacent left part block NB3 and NB4, decoding unit can follow obtained edge line to determine the y coordinate of the first current predictive sample C.

Therefore when be selected as by the position of the first predicting unit of transmitting with reconstruction state with the segmentation from adjacent block the edge line that obtains coincide time, decoding unit successfully can identify that these edge lines are to determine by the position of the first predicting unit transmitted with reconstruction state.

Receiving the predicting unit comprising the first forecast sample being in reconstruction state now, and after the position then determining the first forecast sample, the prediction for the residue sample in predicting unit according to new frame inner plane predictive mode of the present invention can be processed.Therefore in figure 18, see that the first current predictive sample C is received by with reconstruction state, and the position of the first current predictive sample C has been confirmed as having corresponding x coordinate and y coordinate.To be predicted by any one embodiment for new frame inner plane pattern of the present invention now by the block making the first current predictive sample C define, be labeled as 1 as its right lower quadrant sample in current prediction unit 1801.After the sample predicting block 1, the remaining current predictive sample then in current prediction unit 1801 is marked as block 2,3 and 4.These residue samples can be predicted by any one embodiment of new frame inner plane pattern of the present invention.Can by first rebuilding right lower quadrant sample in each piece 2,3 and 4 and with reference to this right lower quadrant sample for predicting that each independent block 2,3 and 4 realizes this individually.Or the residue sample in block 2,3 and 4 can integrally be predicted.Residue sample under any circumstance in block 2,3 and 4 can be predicted according to any one embodiment of new frame inner plane pattern according to the present invention.

Or as an alternative, can by copying the residue sample predicted by the reconstruction sample in any adjacent block (such as block 1) of previously having rebuild in block 2,3 and 4.

Or substitute as another, can by referring to from being confirmed as the sample of adjacent block of advantage block (dominantblock) to predict rest block.Advantage block has the sample properties the most similar to the current block of just predicted process.This is illustrated in Figure 19, and wherein current prediction block 1901 is just predicted at present.In order to exemplary purpose, each adjacent block is depicted as the various shades with grey, the advantage that wherein darker shade instruction is higher.See that adjacent upper left quarter block has the highest advantage, and therefore adjacent upper left quarter block will be referenced to predict current prediction block 1901.By referring to adjacent upper left quarter block, can by copying from the sample of adjacent upper left quarter block simply or predicting current prediction block by referring to the sample in adjacent upper left quarter block, with according to any one can intra prediction mode perform prediction processing.

In order to identify the most advantage adjacent block for current prediction block 1901, advantage block message will be comprised by the part as identifying information, this identifying information by the part as vision signal by coding unit transmission.Then, when decoding unit receives the vision signal comprising this advantage block message and current prediction block 1901, decoding unit can resolve this advantage block message to determine which adjacent block will be referenced for predicting current prediction block 1901.

Figure 20 can be used to perform the schematic block diagram according to the video signal decoding unit of new frame inner plane model prediction of the present invention.

With reference to Figure 20, decoding unit according to the present invention comprises entropy decoding unit 2010, inverse quantization unit 2020, inverse transformation block 2025, removes block filter element (deblockingfilteringunit) 2030, decodes/reconstruction image store 2040, inter prediction unit 2050 and intraprediction unit 2060.

Entropy decoding unit 2010 is by performing the conversion coefficient, motion vector, reference picture index etc. that each piece of video data is extracted in entropy decoding to the video signal bitstream of being encoded by coding unit (description).The decoded conversion coefficient of inverse quantization unit 2020 re-quantization entropy, and inverse transformation block 2025 then use re-quantization after variation coefficient recover original sample value.Block filter element 2030 is gone to be applied to the block of each coding of video data to reduce block distortion.Picture through filtering is stored in decoding picture memory cell 2040 to be output or to be used as reference picture.Inter prediction unit 2050 uses the reference picture be stored in decoding picture memory cell 2040 and the intraframe prediction information (such as reference picture index, motion vector etc.) of sending from entropy decoding unit 2010 to predict photo current.Especially, the motion vector being close to the block (i.e. adjacent block) of current block is extracted by from vision signal.The motion vector of the prediction of current block can be obtained from adjacent block.Adjacent block can comprise be positioned at current block left side, upside or upper right side place block.Such as, the horizontal component of the motion vector of adjacent block and the average of vertical component can be used to obtain the motion vector of the prediction of current block.Or, when the left part block of current block has at least one the prediction block by coded in inter mode, the motion vector of the prediction block at the upside place being positioned at current block can be used to obtain the motion vector of the prediction of current block.When the upper mass of current block has at least one the prediction block by coded in inter mode, the motion vector of the prediction block being positioned at leftmost side place can be used to obtain the motion vector of the prediction of current block.The block at upside among adjacent block, that be positioned at current block and right side place is positioned at beyond the border of picture or band, the motion vector of the prediction of current block can be set to the motion vector of left part block.If there is a block with the same reference picture indices of current block among adjacent block, then the motion vector of this block can be used to motion prediction.

Intraprediction unit 2060 performs infra-frame prediction by referring to the sample from the previous reconstruction in photo current.Reconstruction sample in photo current can comprise block and filter the sample be not applied to.Then by by the photo current of prediction be added together from the residual error that inverse transformation block 2025 exports and rebuild original image.For each predicting unit of video data, each current predictive sample of current prediction unit will be processed by intraprediction unit 2060 new frame inner plane model prediction according to the present invention.Then predicted current predictive sample is rebuild by being merged mutually with the residual error exported from inverse transformation block 2025 by the sample of prediction.

Figure 21 is the block diagram of the substituting view of decoding unit illustrated in Figure 20.Figure comprises block type determining unit 2100 and reconstruction unit 2170 in addition.Block type determining unit 2100 determines that current prediction unit is inter prediction type unit or infra-frame prediction type unit.If block type determining units determination current prediction unit is inter prediction type unit, then current prediction unit will be sent to inter prediction unit 2150.And if block type determining units determination current prediction unit is infra-frame prediction type unit, then current prediction unit will be sent to intraprediction unit 2160.

Figure 21 also illustrates intraprediction unit 2160 and is made up of predicted size determining unit 2161 and predictive mode acquisition unit 2162.Predicted size determining unit 2161 can become vision signal by encoding unit encodes and the identifying information received by decoding unit determines the size of the current prediction unit just predicted by intraprediction unit 2160 by resolving, or determines its size by directly processing current prediction unit.Therefore according to the first method, the coding unit of encoded video signal and adjoint identifying information will comprise the dimension information of each predicting unit of the video data being encoded into vision signal.Then decoding unit only needs from video information, to resolve identifying information to determine the size of each predicting unit that it receives.According to the second method, coding unit does not comprise the dimension information of each predicting unit of the video data being encoded into vision signal significantly.Alternatively, the predicted size determining unit 2161 of decoding unit has been assigned each predicting unit of process to determine the work of the size of each predicting unit.It should be noted according to the first method, for determining that the actual resolved of the identifying information of each predicting unit can by visible predicted size determining unit 2161 or entropy decoding unit 2010 process in fig. 20.

Predictive mode obtains unit 2162 and has been assigned the identifying information of resolving and being included in vision signal to determine suitable intra prediction mode to be applied to the work of each current prediction unit just predicted by intraprediction unit 2160.Therefore according to the present invention, predictive mode obtains unit 2162 by the signaling information of process from identifying information included in vision signal, and should be applied to current prediction unit from the new frame inner plane pattern that this signaling information determines predicting.

And once correctly predict current prediction unit according to the suitable intra prediction mode identified by predictive mode determining unit 2162 by intraprediction unit 2160, just the forecast sample of current prediction unit will be rebuild by reconstruction unit 2170.Reconstruction unit 2170 can by merging them and the residual values obtained from inverse transformation block 2125 sample rebuilding prediction mutually.

Industrial applicibility

Although with reference to the preferred embodiments of the present invention, invention has been described and illustrate in this article, but it is evident that when not deviating from the spirit and scope of the present invention to those skilled in the art, various modifications and variations can be made in the present invention.Therefore, be intended that, the present invention contain fall into claims and they equivalent scope in this invention modifications and variations.

Although with reference to the preferred embodiments of the present invention, invention has been described and illustrate in this article, but it is evident that when not deviating from the spirit and scope of the present invention to those skilled in the art, various modifications and variations can be made in the present invention.Therefore, be intended that, the present invention contain fall into claims and they equivalent scope in this invention modifications and variations.Can encode, transmit, decoding and fast acquisition of digital video data time apply the present invention.

Claims

1., by a method for decoding device process vision signal, described method comprises:

Obtained the prediction type of current block by described decoding device, described prediction type instruction uses infra-frame prediction or inter prediction to encode described current block;

When the described infra-frame prediction of described prediction type instruction use encodes described current block, obtained the intra prediction mode of described current block by described decoding device;

Based on obtained intra prediction mode, predicted the current sample be included in described current block by described decoding device,

Wherein, predict that described current sample comprises:

Perform the first linear interpolation by described decoding device in the horizontal direction of the first reference sample and the 3rd reference sample, and perform the second linear interpolation in the vertical direction of the second reference sample and the 4th reference sample,

Wherein, described first reference sample is the left part adjacent sample adjacent with the current block with the Y-coordinate identical with described current sample, described second reference sample is the top adjacent sample adjacent with the current block with the X-coordinate identical with described current sample, described 3rd reference sample is from being positioned in described current block the rightest sample with the Y-coordinate identical with described current sample with the sample that the upper right quarter block that described current block adjoins obtains, and described 4th reference sample be from the sample that the lower left quarter block that described current block adjoins obtains and be positioned in described current block there is the X-coordinate identical with described current sample play sample most, and

By described decoding device on average from the value of described first linear interpolation generation and the value from described second linear interpolation generation; And

The sample of described prediction is used to rebuild described current block by described decoding device,

Wherein, described upper right quarter block is the right part adjacent block of the top adjacent block of described current block, and described lower left quarter block is the bottom adjacent block of the left part adjacent block of described current block.

2. method according to claim 1, wherein, the value of described 3rd reference sample is set to the value of the lower left corner sample in described upper right quarter block.

3. method according to claim 1, wherein, the value of described 4th reference sample is set to the value of the upper right corner sample in described lower left quarter block.

4., by a method for decoding device process vision signal, described method comprises:

Wherein, predict that described current sample comprises:

The bilinear interpolation of the first sample, the second sample, the 3rd sample and the 4th sample is performed by described decoding device,

Wherein, described first sample is the left part adjacent sample adjacent with the current block with the Y-coordinate identical with described current sample, described second sample is the top adjacent sample adjacent with the current block with the X-coordinate identical with described current sample, described 3rd sample is from being positioned in described current block the rightest sample with the Y-coordinate identical with described current sample with the sample that the upper right quarter block that described current block adjoins obtains, and described 4th sample be from the sample that the lower left quarter block that described current block adjoins obtains and be positioned in described current block there is the X-coordinate identical with described current sample play sample most, and

5. method according to claim 4, wherein, the value of described 3rd sample is set to the value of the lower left corner sample in described upper right quarter block.

6. method according to claim 4, wherein, the value of described 4th sample is set to the value of the upper right corner sample in described lower left quarter block.

7. method according to claim 5, wherein performs described bilinear interpolation and comprises:

Perform the first linear interpolation by described decoding device in the horizontal direction of described first sample and described 3rd sample, and perform the second linear interpolation in the vertical direction of described second sample and described 4th sample, and

By described decoding device on average from the value of described first linear interpolation generation and the value from described second linear interpolation generation.

8. process an equipment for vision signal, described equipment comprises:

First obtaining means, described first obtaining means is for obtaining the prediction type of current block, and the instruction of described prediction type uses infra-frame prediction or inter prediction to encode described current block;

Second obtaining means, described second obtaining means is used for, when the described infra-frame prediction of described prediction type instruction use encodes described current block, obtaining the intra prediction mode of current block;

Prediction unit, described prediction unit is used for predicting based on obtained intra prediction mode the current sample be included in described current block,

Wherein, predict that described current sample comprises:

Perform the first linear interpolation in the horizontal direction of the first reference sample and the 3rd reference sample, and perform the second linear interpolation in the vertical direction of the second reference sample and the 4th reference sample,

On average from the value of described first linear interpolation generation and the value from described second linear interpolation generation; And

Reconstructing device, described reconstructing device rebuilds described current block for using the sample of described prediction,

9. equipment according to claim 8, wherein, the value of described 3rd reference sample is set to the value of the lower left corner sample in described upper right quarter block.

10. equipment according to claim 8, wherein, the value of described 4th reference sample is set to the value of the upper right corner sample in described lower left quarter block.

11. 1 kinds of equipment by decoding device process vision signal, described equipment comprises:

Second obtaining means, described second obtaining means is used for, when the described infra-frame prediction of described prediction type instruction use encodes described current block, obtaining the intra prediction mode of described current block;

Prediction unit, described prediction unit is used for predicting based on obtained intra prediction mode the current sample be included in described current block;

Wherein, predict that described current sample comprises:

12. equipment according to claim 11, wherein, the value of described 3rd sample is set to the value of the lower left corner sample in described upper right quarter block.

13. equipment according to claim 11, wherein, the value of described 4th sample is set to the value of the upper right corner sample in described lower left quarter block.

14. equipment according to claim 11, wherein said execution bilinear interpolation comprises: