CN101340584B - Video encoding method and device - Google Patents
Video encoding method and device Download PDFInfo
- Publication number
- CN101340584B CN101340584B CN 200810118099 CN200810118099A CN101340584B CN 101340584 B CN101340584 B CN 101340584B CN 200810118099 CN200810118099 CN 200810118099 CN 200810118099 A CN200810118099 A CN 200810118099A CN 101340584 B CN101340584 B CN 101340584B
- Authority
- CN
- China
- Prior art keywords
- frame
- macro block
- loop filtering
- video
- rank
- 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.)
- Active
Links
Images
Abstract
The invention provides a video decoding method which includes the following steps: 1) according to a preset time percentage K which is required for reducing, a processing level of loop wave filtering is determined; 2) a residuals matrix of a video macroblock is calculated; 3) according to the video macroblock after decoding, a predicted value matrix is calculated; 4) according to the processing level of the loop wave filtering, the loop wave filtering is carried out on the sum of the residuals matrix and the predicted value matrix. By adopting the invention, a percentage K is given and the video decoding method can ensure that the calculating time is reduced by K of the former time and image quality which is maintained can be accepted.
Description
Technical field
The present invention relates to the processing of digital video signal, specifically, the present invention relates to a kind of video encoding/decoding method and device.
Background technology
Mobile communication technology is flourishing day by day, and the technology of mobile terminal devices such as mobile phone, PDA aspect Video Applications is more and more advanced.After video is sent out or downloads to portable terminal, play in the portable terminal decoding.In the time of the raising of video quality, the computation complexity in the video decoding process also increases greatly, has strengthened the computational load of the processor of portable terminal, and the power that consumes in the decode procedure also increases thereupon.The battery technology of mobile device is in continuous development, but because the speed that battery electric quantity increases can't be offset the energy consumption that the complicated day by day abundant application of mobile device brings, add that mobile device can't charge whenever and wherever possible, energy consumption control is one of sixty-four dollar question for mobile device always.Along with the continuous growth of 3G subscription quantity, Downloadable video content is more and more abundanter, and the consumer is also growing to the demand of mobile device high-quality visual experience.The proportion that the energy that displaying video consumed accounts for mobile device consumption gross energy also will constantly increase.Therefore, during video playback, the control of decoder power is become an important problem.
If can reduce the computation complexity of decoder, just can shorten the time of decoding, and the speed of displaying video is certain, the processor of mobile device just can have the more free time like this, thereby reaches the effect that reduces power.The minimizing of computation complexity need be introduced some approximate calculation, and picture quality can descend to some extent like this.Therefore, need a kind of like this technology, promptly can by approximate calculation, reduce the computation complexity in the decode procedure according to demand, can keep picture quality again in the acceptable scope of user.When electric weight abundance or the high-quality image of user's needs, decoder can not carry out any approximate calculation, solves high-quality image, when electric weight deficiency or user wish power saving, decoder can suitably reduce picture quality, to satisfy the needs of power reduction.
Application number is 02813469.9 and is called that the Chinese patent application of " resource scalable decoding " discloses gradable decoder system of a kind of complexity and method.This system comprises: have the decoder of a plurality of functional blocks, be used for compressed video data is decoded.Decoder also comprises a plurality of grading functions, can be reduced by at least the complexity of a functional block selectively.The local resource controller is used to decoder to generate the classification rule, wherein, for the response select a classification rule from the complexity of system resource management device, moreover, the classification rule be from the local resource controller can with a plurality of predetermined rule select.In question blank, search predetermined classification rule, and question blank designs under off-line state.In this invention, the classification rule does not relate to computation complexity is accurately controlled from the look-up table that designs under the off-line state in sum.
Application number is 200580010306.4 and is called the Chinese patent application of " method and apparatus that is used for complexity scalable video decoder ", has adopted different image resolution ratios, needs encoder that the bit stream of base layer image and enhancement layer image is provided.This invention does not relate to the accurate control to computation complexity equally.
Summary of the invention
The present invention will separate the technical problem that must not accurately control the computation complexity of decoder, thereby provide a kind of video decoder and method, certain percentage in the time of can accurately controlling decode time and reduce to not classified handle keeps picture quality to accept simultaneously.
In order to realize the foregoing invention purpose, according to an aspect of the present invention, provide a kind of video encoding/decoding method, comprise the following steps:
1) estimation B frame loop filtering accounts for the percentage LFB of whole decode time, determines the processing rank of loop filtering according to the relation of percentage of time K that presets required minimizing and LFB:
11) for K≤LFB, determine that the processing rank of described loop filtering is: omit to partly or entirely the residual matrix and the predicted value matrix sum of macro block are done loop filtering in the video B frame;
12) for K＞LFB, the processing rank of determining described loop filtering is: omit the residual matrix and the predicted value matrix sum of the whole macro blocks of video B frame are done loop filtering, and omit partly or entirely the residual matrix and the predicted value matrix sum of macro block are done loop filtering in the video P frame;
2) residual matrix of calculating video macro block;
3) calculate the predicted value matrix according to the video macro block of having decoded;
4) to residual matrix and predicted value matrix sum, carry out loop filtering according to the processing rank of described loop filtering.
In the said method, the part macro block is evenly distributed in each B frame in the frame of video B described in the described step 11).
In the said method, the part macro block is distributed in the zone near the image border in the frame of video B described in the described step 11).
In the said method, the part macro block is evenly distributed in each P frame in the frame of video P described in the described step 12).
In the said method, the part macro block is distributed in the zone near the image border in the frame of video P described in the described step 12).
In the said method, described step 1) comprises: 13) the percentage LFsaved that the time of saving accounts for whole decode time is handled in estimated loop filtering classification;
Described step 3) comprises: 31) for K＞LFsaved, determine the processing rank of the brightness interpolating of non-infra-frame prediction macro block in the B frame, processing rank according to the brightness interpolating of non-infra-frame prediction macro block in the described B frame is carried out inter prediction, thereby calculates described predicted value matrix.
In the said method, the processing rank of the brightness interpolating in the described step 3) comprises: a) keep raw interpolation to handle; B) only carrying out 1/2 interpolation or whole pixel handles; C) only putting in order pixel handles.
In the said method, described step 31) comprising:
311) handle rank, first non-infra-frame prediction macro block of first B frame of each processing unit of inter prediction according to default brightness interpolating;
312) handle the time of being saved according to the brightness interpolating classification of non-infra-frame prediction macro block in the decoded B frame in the accumulative total decode time of the video macro block of having decoded in described K, described LFsaved, the described processing unit and the described processing unit, determine that the brightness interpolating of next non-infra-frame prediction macro block is handled rank.
In the said method, described step 311) to handle rank be to keep raw interpolation to handle for described default brightness interpolating.
In the said method, described step 3) comprises according to the brightness interpolating classification of non-infra-frame prediction macro block in the decoded B frame in the accumulative total decode time of the video macro block of having decoded in described K, described LFsaved, the processing unit and the described processing unit handles the time of being saved, and determines to omit in the next processing unit number of the B frame of decode operation.
According to a further aspect in the invention, also provide a kind of video decoder, having comprised:
Sorter controller is used to estimate that B frame loop filtering accounts for the percentage LFB of whole decode time, and determines the processing rank of loop filtering according to the relation of percentage of time K that presets required minimizing and LFB:
For K≤LFB, be used for determining that the processing rank of described loop filtering is: omit to partly or entirely the residual matrix and the predicted value matrix sum of macro block are done loop filtering in the video B frame;
For K＞LFB, be used for determining that the processing rank of described loop filtering is: omit the residual matrix and the predicted value matrix sum of the whole macro blocks of video B frame are done loop filtering, and omit to partly or entirely the residual matrix and the predicted value matrix sum of macro block are done loop filtering in the video P frame;
Calculate the residual matrix unit, be used to calculate the residual matrix of video macro block;
The frame predicting unit is used for calculating the predicted value matrix according to the video macro block of having decoded;
The loop filtering unit is used for residual matrix and predicted value matrix sum are carried out loop filtering according to the processing rank of described loop filtering.
In the said apparatus, described sorter controller, be used to estimate that described loop filtering classification handles the percentage LFsaved that the time of saving accounts for whole decode time,, determine the processing rank of the brightness interpolating of non-infra-frame prediction macro block in the B frame according to described K, described LFsaved;
Described frame predicting unit is used for carrying out inter prediction according to the processing rank of the brightness interpolating of non-infra-frame prediction macro block in the described B frame.
In the said apparatus, described sorter controller, also be used for handling the time of being saved, determine to omit the number x of the B frame in the next processing unit of decoding according to the brightness interpolating classification of decoded B intra-frame macro block in the accumulative total decode time of the macro block of having decoded in described K, described LFsaved, the processing unit and the described processing unit;
Described calculating residual matrix unit and described frame predicting unit are used for according to described x, omit the decode operation of the B frame in the next processing unit.
Effect of the present invention is, a given percentage K, and scalable decoder of the present invention can make reduce original K computing time, keeps picture quality to accept simultaneously.
Description of drawings
Fig. 1 is the structure chart of known Video Decoder;
Fig. 2 is the structure chart of Video Decoder according to an embodiment of the invention;
Fig. 3 is the brightness interpolating schematic diagram;
Fig. 4 is a parameter schematic diagram according to an embodiment of the invention.
Embodiment
In order to make purpose of the present invention, technical scheme and advantage clearer,, Video Decoder is according to an embodiment of the invention further described below in conjunction with accompanying drawing.Should be appreciated that specific implementation process described herein only in order to explanation the present invention, and be not used in qualification the present invention.
The present invention is applicable to and adopts the various video standards that mix differential pulse coding modulation/discrete cosine transform (hybridDPCM/DCT), comprises H.263, H.264, MPEG-4, AVS etc.In the specific embodiment that the present invention introduces, be example with the AVS standard.
A decoding order of AVS standard is for example:
I P B B P B B I B B P B B
As implied above, frame is divided into 3 classes, decoded picture (B frame) between decoded picture (P frame), bidirectional frame between intraframe decoder image (I frame), forward frame.The I frame adopts infra-frame prediction, and the P frame can be with reference to 1 or 2 frames of front.The B frame has only I frame and P frame to can be used as reference frame with reference to each 1 frame of front and back.Every some frames, an I frame will appear, and this is determined by encoder at interval.The number of continuous B frame is that span is 0 to 2 by a definite value of encoder decision.For other video encoding and decoding standards, its video sequence structure is similar.
Fig. 1 is the structural representation of existing universal decoder.The decoding general process is with macro block, and promptly 16 * 16 pixel region is as the computing elementary cell.Decoder is received after the video code flow, at first calculates residual matrix by calculating the residual matrix unit.Comprise: entropy decoding (VLD), the conversion coefficient that entropy solves obtains residual matrix through inverse quantization and inverse transformation.On the other hand, if carry out infra-frame prediction, the macro block that uses present frame to decode calculates the predicted value of current macro.If carry out inter prediction, will be delivered to motion compensating module 101 from the motion vector that decoder output obtains, calculate predicted value by motion vector and reference frame, wherein reference frame is 1 frame or 2 frames of early decoding.Predicted value matrix and residual matrix addition are passed through loop filtering (Loop Filter is designated as LF) 100 again and are removed blocking effect, improve picture quality, thereby finish the decoding of present frame.The frame output that decoder is finished decoding, simultaneously for example between intraframe decoder image (I frame) and forward frame decoded picture (P frame) wait need be as the reference frame of the frame of back, quilt is deposited in frame buffer.
Thought of the present invention is, utilize system clock to obtain the processing time of macro block, and calculate each module classification of decoder and handle the time of being saved, estimate or calculate the ratio that each module time accounts for total decode time, according to the comparative result of these ratios and demand, dynamically adjust the rank that each module classification is handled.According to one embodiment of the invention, at first obtain the amount of calculation K that needs minimizing, also promptly need decoding operation time of reducing,, handle as the classification of loop filtering 100, motion compensation 101 and to realize that amount of calculation or decoding reduce K operation time by several main modular to decoder.Wherein the classification processing of the brightness interpolating (Luminance Interpolation, note is made LIP) that is presented as inter prediction is handled in the classification of motion compensation 101.The processing of loop filtering 100, motion compensating module 101 is divided into several different ranks, the computation complexity that each rank is corresponding different, the calculating of minimizing is many more, and the distortion of image is generally also big more.The basic thought of classification processing is the calculating of simplifying a part according to demand, and keeps picture quality as far as possible, and this will greatly still influence little operation to picture quality by refinement letter computation complexity.Because the B frame as with reference to frame, therefore, does not cause decrease in image quality can not cause the mistake transmission because classification is handled in the B frame, can carry out in the B frame being similar to calculating, even skip some B frame, does not promptly decode.Opposite, the picture quality of I frame and P frame must be guaranteed, in the described decoder of present embodiment, do not carry out any approximate calculation and classification for the I frame and handle, and only carries out the classification of loop filtering for the P frame and handle.This is because the loop filtering computation complexity is big, and is minimum in each module for the influence of picture quality.
Fig. 2 is a scalable decoder structural representation according to an embodiment of the invention.Before the video sequence decoding, sorter controller is determined the classification processing rank of loop filtering according to the size of K.In decode procedure, before the non-infra-frame prediction macro block decoding of each of B frame, it is foundation that sorter controller is handled the percentage of time LFsaved that is saved with the loop filtering classification, determines the rank of the brightness interpolating classification processing of motion compensating module 101.In addition, sorter controller is also handled the time of being saved according to the accumulative total decode time of the macro block of having decoded and the brightness interpolating classification of decoded B intra-frame macro block, determines whether skip the B frame.When the B frame is skipped, its corresponding code stream will not carry out entropy decoding and all operations subsequently.
If continuous B frame number is 0 in the sequence of frames of video, so only to carry out the classification of P frame loop filtering and handle, this situation occurs considerably less in actual applications.According to one embodiment of present invention, every group of continuous B frame number is 2, is one " processing unit " with P frame B frame B frame or I frame B frame B frame.Since the 2nd frame, above-mentioned processing unit circulation occurs up to EOS according to decoding order.According to these characteristics of sequence of frames of video itself, decoding according to processing unit can cycling.In addition, because comprise the step of prediction and estimation in the method for the present invention, and each processing unit is independently, so can avoid owing to the inaccurate problem that has influence on overall situation decoding of individual elements control.It should be appreciated by those skilled in the art that if the B frame number is 1 continuously, is one " processing unit " with P frame B frame or I frame B frame then.Can certainly not divide processing unit, to all P frames and the B frame processed in sequence of whole video.To describe in detail below according to processing unit and carry out the method that classification is handled.
The percentage K of the computing time of Jie Shenging determines the processing mode of loop filtering module at first as required.The classification processing mode of loop filtering module is specially, and considers earlier not do the loop filtering of B frame, if can not satisfy the requirement that computation complexity reduces, then considers not do the loop filtering of P frame.Loop filtering is that unit carries out with the macro block, when doing the classification processing, once saves the loop filtering operation of a macro block.
The video of different content different characteristics, also have nothing in common with each other the average time of each macro-block loop filtering in B frame and the P frame.For the actual video sequence, may comprise different characteristics in different time sections, be a relatively-stationary value but B frame and P frame loop filtering account for the ratio of whole decode time.These two ratios also can obtain by a large amount of decoding statistics, and meter is made LFB, LFP.If K≤LFB, illustrate that only need the part or all of macro block of B frame not being carried out loop filtering just can satisfy the demand that computation complexity reduces, select these not make the macro block of loop filtering, they are evenly distributed in each B frame, and in each B frame, are distributed in the zone of not paying close attention near the edge of image human eye as far as possible.If K＞LFB, need all macro blocks of B frame not made loop filtering, then also needing the percentage of computing time of saving is M=K-LFB, relatively M and LFP are if M≤LFP illustrates that some macro block to the P frame does not carry out LF and just can satisfy the demand that computation complexity reduces, select these not make the macro block of loop filtering, they are evenly distributed in each P frame, and in each P frame, are distributed in the zone of not paying close attention near the edge of image human eye as far as possible.If M＞LFP, then all macro blocks to B frame and P frame do not carry out loop filtering, but still can not satisfy the demand that computation complexity reduces like this, need carry out the classification of B frame brightness interpolating and handle or skip the B frame.
When all macro blocks to B frame and P frame all do not carry out loop filtering, then the percentage IFsaved=LFB+LFP that is saved is handled in the loop filtering classification, also need to save computing time during percentage N=M-LFP, the brightness interpolating of the non-infra-frame prediction macro block of B frame is carried out classification handle or skip the B frame.Describe the classification of brightness interpolating below in detail and handle and skip determining of B frame number.
Fig. 3 is the brightness interpolating schematic diagram of AVS standard, wherein shows the position of integer pixel point, 1/2 sample point and 1/4 sample point.Wherein capitalization is represented the integer pixel point, and b, h and j represent 1/2 sample point among the figure, and the each point except that whole pixel and 1/2 sample point is 1/4 sample point.When entropy is decoded the motion vector points non-integer pixel point that obtains,, need carry out brightness interpolating to reference frame in order to obtain the luma prediction value matrix.All adopt 1/4 interpolation in the most standard, for being the square area of 16 sample points of left upper apex with the D point, motion vector may point among D, a, b, c, d, e, f, g, h, i, j, k, n, p, q and the r more arbitrarily, and complexities of the corresponding interpolation calculation of these points are inconsistent.The D point is the integer pixel point, and obtaining with the D point is the luma prediction matrix of left upper apex, do not need interpolation, so the computation complexity that D is ordered is minimum; All the other each point correspondences interpolation operation separately, computation complexity is different, all things considered, the computation complexity of 1/2 interpolation is less than 1/4 interpolation.Only the classification of the brightness interpolating that carries out in the B frame is handled, and being divided into is 3 ranks: 1) keep raw interpolation to handle, promptly interpolation is not done to simplify and handle; 2) only carrying out 1/2 interpolation or whole pixel and handle, is example with Fig. 3, i.e. a, and d, the interpolation of e transfers the processing of D to, c, f, the interpolation of g transfers the interpolation of b to, i, n, p transfer the h point interpolation to, k, q, r transfer the j point interpolation to; 3) with a, the b...r point interpolation all transfers the processing of whole pixel D to.For these 16 points of D, a, b, c, d, e, f, g, h, i, j, k, n, p, q and r, the computation complexity of each point is certain, decoder can obtain by a large amount of decoding statistics at the clock number that each point carries out brightness interpolating, is kept in the decoder as constant.
The method that the brightness interpolating classification is handled is as follows: in each " processing unit " of video sequence, and during the non-infra-frame prediction macro block of first of first B frame of decoding, the rank of an at first given one default brightness interpolating, preferred, keep raw interpolation to handle.In each processing unit, if next macro block is the non-infra-frame prediction macro block of B frame, then according under not doing loop filtering and having carried out situation that the classification of brightness interpolating handles, record is as the decode time Ta of pretreatment unit first frame (I frame or P frame), time tb when the accumulative total decoding of pretreatment unit from first macro block of B frame to current macro, with the time ts that is saved when the brightness interpolating classification processing of pretreatment unit from first macro block of B frame to current macro, determine the brightness interpolating rank of next macro block, up to finishing when pretreatment unit.Under personal computer (PC), digital signal processor (DSP), programmable gate array platforms such as (FPGA), the method that obtains the system clock number is all arranged, so Ta and tb obtain easily.In motion compensating module, determine respectively which two point among D, a, b, c, d, e, f, g, h, i, j, k, n, p, q and the r are handled before and after the classification, the difference of the two required clock number is exactly the time that brightness interpolating is saved, the time that the each brightness interpolating of accumulative total is saved, obtain ts.Before each non-infra-frame prediction macro block brightness interpolation rank of definite B frame,, can estimate the time T s that can save as pretreatment unit B frame decoding time T b with when the pretreatment unit brightness interpolating by current tb, ts.
BMB wherein
CurrentAnd BMB
TotalRepresent macroblock number of having decoded and the sum of working as pretreatment unit B frame macro block respectively when pretreatment unit.
The ratio of decoding total time when not carrying out classification and handle time that the expression brightness interpolating is saved, the relation of each parameter as shown in Figure 4 in the formula, the size that compares IPSaved and N, if IPSaved than the big amount of N greater than certain threshold value, the approximate calculation degree that brightness interpolating is described is too big, can reduce the rank of brightness interpolating, do meticulousr calculating.If IPSaved greater than certain threshold value, illustrates that the approximate calculation degree of brightness interpolating is too little than the little amount of N, can improve the rank of brightness interpolating, do simpler calculating.Above-mentioned threshold value is approaching or equal 0, and preferred span is [0,0.001].
After each processing unit decoding finished, the branch level processor also will be according to the Ta of the each processing unit of having decoded, Tb, and Ts and K judge the number of skipping B frame in the next processing unit.The clock number that brightness interpolating can reduce is limited, when all interpolation calculation all are converted into whole pixel when handling, the spent clock number of brightness interpolating reaches minimum value, if find that the brightness interpolating classification of decoding unit can not meet the demands, then skips the B frame of next processing unit.B frame for first processing unit is not skipped.The method of determining to skip B frame number is specific as follows: represent the clock number that the part except loop filtering and brightness interpolating should reduce with S, what CB represented to estimate skips the total time that the B frame is saved.The initial value of S and CB all is 0.When the decoding of processing unit finishes, calculate this processing unit
Renewal is corresponding to the S of whole video sequence
Total, make it equal the current each processing unit S that has decoded and.(S
Total-CB) expression has been carried out the classification processing to loop filtering and brightness interpolating, has adopted after the method for skipping the B frame current should the minimizing but still unbated computation complexity.If S
Total-CB＞0, then order
Wherein, here
Expression rounds downwards, and Tb is the average time of the B frame decoding after classification is handled, and x2 represents the number of continuous B frame, and x is x1, the x2 minimum value in the two.Calculate after the x, skip preceding x B frame of next processing unit, and upgrade the CB value, make it equal original CB and add Tb * x.If x=0, expression should reduce but still unbated computation complexity is littler than the average time of B frame decoding, does not then skip the B frame.
Loop filtering has been described above, algorithm that the brightness interpolating classification is handled and the method for skipping the B frame.After all the B frames the B frame in first processing unit are all skipped, amount of calculation can't continue to reduce, at this moment, video only has the I frame, the B frame of the loop filtering of having removed the P frame of loop filtering and the removal in first processing unit and whole pixel brightness interpolating, the most basic picture quality in the coding/decoding system that corresponding the present invention proposes.The K of this moment is exactly the largest percentage that can reduce amount of calculation or computing time, and note is made Kmax.In order to guarantee the most basic picture quality, when K＞Kmax, no longer carry out operations such as more approximate calculation or classification processing.Kmax is by the decision of the content of video sequence, is 2 situation for continuous B frame number, generally more than 70%.When K＜Kmax, can reach the gradable requirement of computation complexity in the practical application fully according to embodiments of the invention.
Should be noted that and understand, under the situation that does not break away from the desired the spirit and scope of the present invention of accompanying Claim, can make various modifications and improvement the present invention of foregoing detailed description.Therefore, the scope of claimed technical scheme is not subjected to the restriction of given any specific exemplary teachings.
Claims (13)
1. a video encoding/decoding method comprises the following steps:
1) estimation B frame loop filtering accounts for the percentage LFB of whole decode time, determines the processing rank of loop filtering according to the relation of percentage of time K that presets required minimizing and LFB:
11) for K≤LFB, determine that the processing rank of described loop filtering is: omit to partly or entirely the residual matrix and the predicted value matrix sum of macro block are done loop filtering in the video B frame;
12) for K＞LFB, the processing rank of determining described loop filtering is: omit the residual matrix and the predicted value matrix sum of the whole macro blocks of video B frame are done loop filtering, and omit partly or entirely the residual matrix and the predicted value matrix sum of macro block are done loop filtering in the video P frame;
2) residual matrix of calculating video macro block;
3) calculate the predicted value matrix according to the video macro block of having decoded;
4) to residual matrix and predicted value matrix sum, carry out loop filtering according to the processing rank of described loop filtering.
2. method according to claim 1 is characterized in that, the part macro block is evenly distributed in each B frame in the frame of video B described in the described step 11).
3. method according to claim 1 and 2 is characterized in that, the part macro block is distributed in the zone near the image border in the frame of video B described in the described step 11).
4. method according to claim 1 is characterized in that, the part macro block is evenly distributed in each P frame in the frame of video P described in the described step 12).
5. according to claim 1 or 4 described methods, it is characterized in that the part macro block is distributed in the zone near the image border in the frame of video P described in the described step 12).
6. method according to claim 1 is characterized in that,
Described step 1) comprises: 13) the percentage LFsaved that the time of saving accounts for whole decode time is handled in estimated loop filtering classification;
Described step 3) comprises: 31) for K＞LFsaved, determine the processing rank of the brightness interpolating of non-infra-frame prediction macro block in the B frame, processing rank according to the brightness interpolating of non-infra-frame prediction macro block in the described B frame is carried out inter prediction, thereby calculates described predicted value matrix.
7. method according to claim 6 is characterized in that, the processing rank of the brightness interpolating in the described step 3) comprises: a) keep raw interpolation to handle; B) only carrying out 1/2 interpolation or whole pixel handles; C) only putting in order pixel handles.
8. according to claim 6 or 7 described methods, it is characterized in that described step 31) comprising:
311) handle rank, first non-infra-frame prediction macro block of first B frame of each processing unit of inter prediction according to default brightness interpolating;
312) handle the time of being saved according to the brightness interpolating classification of non-infra-frame prediction macro block in the decoded B frame in the accumulative total decode time of the video macro block of having decoded in described K, described LFsaved, the described processing unit and the described processing unit, determine that the brightness interpolating of next non-infra-frame prediction macro block is handled rank.
9. method according to claim 8 is characterized in that, described step 311) to handle rank be to keep raw interpolation to handle for described default brightness interpolating.
10. method according to claim 6, it is characterized in that, described step 3) comprises according to the brightness interpolating classification of non-infra-frame prediction macro block in the decoded B frame in the accumulative total decode time of the video macro block of having decoded in described K, described LFsaved, the processing unit and the described processing unit handles the time of being saved, and determines to omit in the next processing unit number of the B frame of decode operation.
11. a video decoder comprises:
Sorter controller is used to estimate that B frame loop filtering accounts for the percentage LFB of whole decode time, and determines the processing rank of loop filtering according to the relation of percentage of time K that presets required minimizing and LFB:
For K≤LFB, be used for determining that the processing rank of described loop filtering is: omit to partly or entirely the residual matrix and the predicted value matrix sum of macro block are done loop filtering in the video B frame;
For K＞LFB, be used for determining that the processing rank of described loop filtering is: omit the residual matrix and the predicted value matrix sum of the whole macro blocks of video B frame are done loop filtering, and omit to partly or entirely the residual matrix and the predicted value matrix sum of macro block are done loop filtering in the video P frame;
Calculate the residual matrix unit, be used to calculate the residual matrix of video macro block;
The frame predicting unit is used for calculating the predicted value matrix according to the video macro block of having decoded;
The loop filtering unit is used for residual matrix and predicted value matrix sum are carried out loop filtering according to the processing rank of described loop filtering.
12. device according to claim 11 is characterized in that,
Described sorter controller is used to estimate that described loop filtering classification handles the percentage LFsaved that the time of saving accounts for whole decode time, according to described K, described LFsaved, determines the processing rank of the brightness interpolating of non-infra-frame prediction macro block in the B frame;
Described frame predicting unit is used for carrying out inter prediction according to the processing rank of the brightness interpolating of non-infra-frame prediction macro block in the described B frame.
13. device according to claim 12 is characterized in that,
Described sorter controller, also be used for handling the time of being saved, determine to omit the number x of the B frame in the next processing unit of decoding according to the brightness interpolating classification of decoded B intra-frame macro block in the accumulative total decode time of the macro block of having decoded in described K, described LFsaved, the processing unit and the described processing unit;
Described calculating residual matrix unit and described frame predicting unit are used for according to described x, omit the decode operation of the B frame in the next processing unit.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200810118099 CN101340584B (en) | 2008-08-11 | 2008-08-11 | Video encoding method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200810118099 CN101340584B (en) | 2008-08-11 | 2008-08-11 | Video encoding method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101340584A CN101340584A (en) | 2009-01-07 |
| CN101340584B true CN101340584B (en) | 2010-09-29 |
Family
ID=40214515
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 200810118099 Active CN101340584B (en) | 2008-08-11 | 2008-08-11 | Video encoding method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101340584B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| MX353197B (en) * | 2010-11-26 | 2018-01-05 | Nec Corp | Image encoding device, image decoding device, image encoding method, image decoding method, and program. |
| EP2805493A1 (en) * | 2012-01-17 | 2014-11-26 | Futurewei Technologies, Inc. | In-loop filtering for lossless coding mode in high efficiency video coding |
| CN110213577B (en) * | 2019-06-20 | 2023-05-09 | 深圳市网心科技有限公司 | Reconstruction frame construction method, system, electronic equipment and storage medium |
-
2008
- 2008-08-11 CN CN 200810118099 patent/CN101340584B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN101340584A (en) | 2009-01-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102057676B (en) | Method and system for video coder and decoder joint optimization | |
| CN101233757A (en) | Method, module, device and system for rate control provision for video encoders capable of variable bit rate encoding | |
| CN102271257B (en) | Image processing device, method, and program | |
| US10284849B2 (en) | Quantization parameter (QP) calculation for display stream compression (DSC) based on complexity measure | |
| CN100581255C (en) | Pixel loop filtering method and filter | |
| CN109905711A (en) | A kind of processing method of image, system and terminal device | |
| CN108702513B (en) | Apparatus and method for adaptive computation of quantization parameters in display stream compression | |
| CN100574447C (en) | Fast intraframe predicting mode selecting method based on the AVS video coding | |
| TWI721042B (en) | System and methods for fixed-point approximations in display stream compression (dsc) | |
| CN103327325A (en) | Intra-frame prediction mode rapid self-adaptation selection method based on HEVC standard | |
| US8270747B2 (en) | Image encoding device, image decoding device, and integrated circuit | |
| US20240040127A1 (en) | Video encoding method and apparatus and electronic device | |
| CN1399472A (en) | Multichannel image coding device and method | |
| CN101340584B (en) | Video encoding method and device | |
| US20050249280A1 (en) | Image data compression apparatus, enencoder electronic equipment, and image data compression method | |
| US8559738B2 (en) | Predictive coding method for coding texture of image | |
| CN101742323B (en) | Method and device for coding and decoding re-loss-free video | |
| EP1947863A1 (en) | Image decoding apparatus and method, and image encoding apparatus | |
| CN101287124A (en) | Video decoding device and method | |
| CN110012292A (en) | Method and apparatus for compressed video data | |
| KR20180136618A (en) | Method of compressing image and display apparatus for performing the same | |
| KR20120118283A (en) | Method for coding image using adaptive coding structure selection and system for coding image using the method | |
| CN106231340A (en) | A kind of infra-frame prediction coding/decoding method based on HEVC and device | |
| Cui et al. | Perceptual Based Fast CU Partition Algorithm for VVC Intra Coding | |
| CN116320446A (en) | Video encoding method and device, electronic equipment and computer medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |