CN105915918B

CN105915918B - Method and apparatus across planar prediction

Info

Publication number: CN105915918B
Application number: CN201610412834.2A
Authority: CN
Inventors: 海纳·基希霍弗尔; 马丁·温肯; 海科·施瓦茨; 德特勒夫·马佩; 托马斯·维甘徳
Original assignee: GE Video Compression LLC
Current assignee: GE Video Compression LLC
Priority date: 2010-04-13
Filing date: 2010-04-13
Publication date: 2019-09-06
Anticipated expiration: 2030-04-13
Also published as: CN105872562B; CN105933716B; CN105915920A; CN105915921B; CN105915922B; CN105915923A; CN105933716A; CN105933715B; CN105915919B; CN105915921A; CN105915922A; CN105915920B; CN105915919A; CN105915918A; CN105872562A; CN105915924B; CN105915923B; CN105872563B; CN105915924A; CN105872563A

Abstract

This application involves across planar prediction.Although the demand of across planar prediction information signal notice decoder will lead to overhead, when the target for redundancy reduction can obtain better rate distortion ratio using the correlation between the coding parameter of Different Plane.Specifically, it can be made whether respectively for multiple planes using the judgement across planar prediction.In addition, or optionally, it is contemplated that a secondary flat can be determined in blocks.

Description

Method and apparatus across planar prediction

The application is divisional application, parent application application No. is 201080067394.2, the applying date is in April, 2010 It is 13 days, entitled " across planar prediction ".

Technical field

The present invention relates to the encoding scheme of the different spaces sample intelligence component for scene image planar, Mei Geping Face includes the message sample array of such as video or still image.

Background technique

In image and Video coding, image or the specific sample array set for the image are typically dissected into block, The block and specific coding parameter association.Image is usually to be made of multiple sample arrays.In addition, an image can also be associated with additionally Aid sample array, the sample array (for example) show transparent information or depth map.One image sample array (including auxiliary sample This array) one or more so-called plane groups can also be assembled, each plane group is by one or more sample arrays herein Composition.The plane group of one image can independently encode, alternatively, if the image is associated with more than one plane group, using come It is encoded from the prediction of other plane groups of same image.Each plane group is typically dissected into multiple blocks.The block (or The corresponding block of sample array) it is to be predicted and across image prediction or image interior prediction.Each block can have different sizes And it can be square or rectangle.One image segmentation can be fixed at multiple blocks by grammer, or can be (at least partly) in place Flow internal signal notice.Predefined size of the syntactic element signal notice often sent for block subdivision.This syntactic element It can be shown that whether and how a block is subdivided into smaller block and associated coding parameter, such as predicting purpose.Needle To whole samples an of block (or corresponding block of sample array), the decoding of associated coding parameter is in a manner of some Show.It in this example, is using identical Prediction Parameters ensemble prediction in whole samples of a block, the Prediction Parameters are all in this way Benchmark index (mark a reference picture) in encoded image set, kinematic parameter (show a reference picture and this are current The measured value of block motion between image), the parameter, the intra-prediction mode that show interpolation filter etc..Kinematic parameter can be by It is indicated with the motion vector of a horizontal component and a vertical component, or indicated by order motion parameter, such as including six The affine motion parameters of component.May also more than one particular prediction parameter sets (such as benchmark index and kinematic parameter) be with Single block is associated.In the case of this kind, for each set of this particular prediction parameter, generate for the block (or sample number The corresponding block of group) single medium range forecast signal, and final prediction signal is by including superposition medium range forecast signal One combination is established.Corresponding weighting parameters and possibly also include that a constant offset (adding to the weighted sum) can be directed to a figure Picture or a reference picture or a reference picture collection are combined into fixation or it includes in the Prediction Parameters set for being directed to corresponding block In.Difference between original block (or corresponding block of sample array) and its prediction signal is also referred to as used as residual signals, the difference It is usually transformed and quantifies.Often, apply two-dimensional transform to the residual signals (or the corresponding sample number for the residual error block Group).For transition coding, use the block (or corresponding block of sample array) of particular prediction parameter sets that can apply Add and is further divided before transformation.Transform blockiis can be equal to or less than the block for prediction.It is also likely to be a transform blockiis Including the more than one block for predicting.Different transform blockiis can have different size, and transform blockiis can indicate square or square Shape block.After the conversion, gained transformation coefficient is quantified, obtains so-called transformation coefficient level.Transformation coefficient level and prediction Parameter and if it exists when, subdivided information is coded by entropy.

In image and video encoding standard, an image (or a plane group) is subdivided into area as provided by grammer A possibility that block, is extremely limited.Being typically only capable to define whether (and possibly how) has the block for defining size in advance can It is subdivided into smaller block.An example is lifted, maximum resource block size H.264 is 16 × 16.16 × 16 blocks are also referred to as used as Hong Qu Block, in first step, each image segmentation is at macro zone block.For each 16 × 16 macro zone block, signal notifies whether it is encoded into 16 × 16 blocks or two 16 × 8 blocks or two 8 × 16 blocks or four 8 × 8 blocks.If 16 × 16 blocks are subdivided At four 8 × 8 blocks, then each 8 × 8 block may be encoded as 8 × 8 blocks or two 8 × 4 blocks or two 4th × 8 areas Block or four 4 × 4 blocks.Show to be divided into the small set possibility of block to have in current image and video encoding standard The advantages of be for signal notice subdivided information side information rate can keep smaller, but have the drawback that for the block pass Bit rate needed for defeated Prediction Parameters is quite big, is described in detail later.The side information rate of signal notice predictive information usually indicates really For significant a large amount of total bit rates of a block.When this side information reduce when, code efficiency increases, for example, can by using compared with Large-scale block size come realize side information reduction.The practical image or image of video sequence are by with any of special properties Shape objects composition.An example is lifted, this object or object part are with unique texture or unique motion for its feature.It is usually identical Prediction Parameters set can be applied to this object or object part.But object bounds usually and misfit large-scale prediction block (for example, According to 16 × 16 macro zone blocks H.264) possible block border.

Encoder usually determines that subdivision leads to the minimum of particular rate distortion cost measuring (in limited kind of possibility set) Change.For arbitrary shape of object, a large amount of block of cells so may cause.And since this block of cells is one transmitted with needs Prediction Parameters set is associated, therefore side information rate becomes most of total bit rate.But due in block of cells it is several still It indicates the area of same target or object a part, therefore is identical or very much like to the Prediction Parameters of multiple gained blocks.

In other words, an image segment or piece together into compared with miniature part or piece block or block together substantially and influence code efficiency and Encoder complexity.If outline, the space that an image is subdivided into multiple smaller blocks permission coding parameters are more finely set above, by This allows this coding parameter to be more preferably adapted to image/video material.On the other hand, with more fine granularity setting coding parameter to notice The required side information content of setting value of the decoder in relation to needing adds all more high load capacities.Furthermore notably encoder (further) Spatially subdivision image/video become block any freedom degree, increase severely possible coding parameter setting value amount, and thus usually So that for causing the search of optimal ratio/distortion tradeoff coding parameter setting value more difficult.

Summary of the invention

It is an object of the present invention to provide a kind of different spaces sample information components for scene image planar Coding encoding scheme, each plane includes message sample array, and the program can obtain better rate distortion ratio.

The present invention potentially conceives, although across planar prediction information signal is notified that the demand to decoder will lead to Overhead, but when the target for redundancy reduction can be obtained more using the correlation between the coding parameter of Different Plane Good rate is distorted ratio.

According to one embodiment, the message sample array of representation space sample information signal first is spatially placed in tree root Area, then according to the multiway tree subdivided information for being drawn from a data flow, by the subset in the recursively repeated segmentation tree root area, and A subset in the tree root area is at least divided into various sizes of smaller simple bonding pad.In order to allow with regard to rate distortion sense On, it finds out with the excessive tiny subdivision of reasonable coding complexity and the good compromise excessively between coarse subdivision, message sample array The maximum area's size for carrying out the tree root area that space is divided into is to include in the data flow and extract in decoding end from the data flow. Accordingly, decoder may include a withdrawal device, be configured as extracting maximum area's size and multiway tree subdivided information from data flow；One Subdivider is configured as a message sample array space of representation space sample information signal being divided into maximum area's size Tree root area, and according to the multiway tree subdivided information, at least one subset in the tree root area is passed through into recursively multi-division tree The subset of root zone and be subdivided into and smaller connect different size fields merely；And a reconstructor, it is configured with the subdivision And the message sample array from the data flow is reconstructed into more small-sized simple bonding pad.

According to one embodiment, data flow also contains up to the summit that tree root area subset undergoes recursively multi-division Formula level.By this method, the signal notice of multiway tree subdivided information becomes easier to and needs less bits of coded.

In addition, reconstructor can be configured to execute one of following measures or more to depend on the granularity of intermediate subdivision Which person: at least in inner estimation mode and across decision in prediction mode it is intended to that prediction mode used；From frequency-domain transform to spatial domain, Execute and/or set the parameter across prediction；It executes and/or setting is for the parameter for interior prediction.

In addition, withdrawal device can be configured to extract and sectorized tree block with depth-first traversal sequence from data flow The associated syntactic element in leaf area.By such method, withdrawal device can develop the syntactic element in coded neighbouring leaf area Statistics amount has than using the higher probability of breadth first traversal sequence.

According to another embodiment, using another subdivider come according to another multiway tree subdivided information by the more small-sized list At least one subset of pure bonding pad is subdivided into and smaller simple bonding pad.First order subdivision can be used to execute letter by reconstructor The prediction of sample area is ceased, and second level subdivision can be used to execute converting again from frequency domain to spatial domain by reconstructor.Define residual error Subdivision is subdivided into subordinate relative to prediction, so that the less consumption position of the coding always segmented；On the other hand, by the resulting residual error of subordinate The degree of restriction and freedom degree of subdivision only have small negative effect to code efficiency, and reason is most of with similar movement benefit The image section for repaying parameter is bigger than the part with similar spectral nature.

According to still another embodiment, another maximum area size is contained in the data flow, another maximum area size definitions tree Root area size, the tree root sub-district be the tree root sub-district an at least subset according to again more multiway tree subdivided information and be subdivided into To be first divided before smaller simple bonding pad again.So transfer the independence for allowing on the one hand maximum area's size of prediction subdivision Setting, and on the other hand residual error is allowed to segment, it can so find out preferable rate/distortion tradeoff.

Still another embodiment according to the present invention, data flow include with formed the multiway tree subdivided information the second grammer member First syntactic element subset of sub-prime collection separation, wherein the combiner in the decoding end allows according to the first syntactic element subset And interblock space it is neighbouring multiway tree subdivision small-sized simple bonding pad come obtain the sample array one among subdivision.Reconstructor It can be configured to using centre subdivision and reconstruction sample array.In this manner, encoder is easier with optimal ratio/mistake of discovery Effective subdivision is adapted to the spatial distribution of the property of message sample array by true compromise.For example, if maximum area is having a size of big, Then multiway tree subdivided information may be because that tree root area becomes larger is more complicated.But then, it if maximum area is having a size of small, is more likely to It is neighbouring tree root area is the related information content with similar quality, so that this tree root area can also be handled together.Before merging is filled up This gap between stating extremely allows to segment close to optimized granularity by this.From encoder viewpoint, merges syntactic element and permit Perhaps coded program less complex more easily or in operation, if reason is that encoder mistake uses too fine subdivision, This error can be by encoder with post-compensation, by merging syntactic element with or without only adjustment adaptation sub-fraction by then setting The syntactic element being set before merging syntactic element setting is reached.

According to still another embodiment, maximum area's size and multiway tree subdivided information be for residual error subdivision and nonanticipating is thin Point.

For handling the simple bonding pad that the multiway tree of a message sample array of representation space sample information signal segments A depth-first traversal sequence rather than breadth first traversal sequence be according to an embodiment use.By by using the depth excellent There are higher probability in first traversal order, each simple bonding pad to have the neighbouring simple bonding pad being traversed, so that when weight When building individual simple bonding pads at present, the information in relation to these adjacent to simple bonding pad can be positively utilized.

It is then thin again when the tree root area that message sample array is first separated into the class type size of zero layer grade is regularly arranged When dividing at least one subset in the tree root area becomes different size of smaller simple bonding pad, zigzag can be used to sweep for reconstructor It retouches to scan the tree root area, for each tree root area for being intended to subregion, with the depth-first traversal sequential processes leaf connected merely Area then more steps into next tree root area again with zigzag scan order.In addition, having identical according to depth-first traversal sequence The leaf area of hierarchical level connected merely can also be traversed according to zigzag scan order.In this way, maintaining to have neighbouring simple A possibility that connecting leaf area is increased.

According to one embodiment, although label associated with the node of polynary tree construction is suitable according to depth-first traversal Sequence estimates context using probability come the coding in proper order for arranging, but marking in proper order, for the identical stratum in polynary tree construction Label associated with multiway tree structure node is inside formula level, but for the different estate formula layer in polynary tree construction The associated label of multiway tree structure node inside grade is, good between the context number for allowing to be provided by this Compromise, and on the other hand, adjustment adapts to the actual symbol statistics of label.

According to one embodiment, it is also dependent on for the probability estimation context of used predetermined labels according to depth Label of the first traversing order before the predetermined labels, and area corresponding with the predetermined labels is corresponded to predetermined phase To each area in the tree root area of positional relationship.Potentially conceive similar to aforementioned aspects, is guaranteed using depth-first traversal sequence high Probability: coded label also includes label corresponding to adjacent area, area corresponding with the predetermined labels, this knows can be used To adjust context more excellently for the predetermined labels.

Can be used for setting the label for the context of a predetermined labels can be that correspond to the predetermined labels opposite Answer the label in Qu Shangqu and/or left area.In addition, associated with predetermined labels are belonged to select the label of context can be limited to The identical hierarchical level of node label.

According to one embodiment, encoded signal notice include summit formula level instruction and be not equal to summit The associated flag sequence of the node of formula level, each label show whether associated nodes are intermediate node or child node, And decode the flag sequence for deriving from the data flow in proper order according to depth-first or breadth first traversal sequence, skip summit The node of formula level and be automatically directed to identical leaf node, thus reduce encoding rate.

According to another embodiment, the encoded signal notice of polynary tree construction may include the instruction of summit formula level. In this manner, the presence of label may be limited to the hierarchical level other than summit formula level, for reason is always Exclusion have summit formula level block further subregion.

In space, multiway tree subdivision belongs to the leaf node of a multiway tree subdivision and the subdivision without partition tree root zone A part in the case where, the context for encoding subdivision label may be selected so that the context be directed to it is equal big Cell it is associated label be.

According to an embodiment, the advantageous merging in a simple bonding pad made of which segments or it is grouped and is With little data amount coding.In order to reach this purpose, for simple bonding pad, a predetermined relative location relationship is defined, is permitted Perhaps make a reservation for simple bonding pad for one and identify and make a reservation for simple bonding pad with predetermined phase with this inside multiple simple bonding pads To the simple bonding pad of positional relationship.In other words, if the number is zero, it may be not present to have to be directed to inside the data flow and be somebody's turn to do Make a reservation for simple bonding pad one merges index.In addition, if making a reservation for the list that simple bonding pad has predetermined relative location relationship with this Pure bonding pad number is 1, then the coding parameter of the simple bonding pad can be used, or can be used to predict for the predetermined simple connection The coding parameter in area is without any additional syntactic element.Otherwise, that is, if making a reservation for simple bonding pad with this has predetermined opposite position The simple bonding pad number for setting relationship is that then can inhibit the introducing of an additional syntactic element, even if identified with these greater than 1 The associated coding parameter in simple bonding pad be each other identical also multiple such.

According to an embodiment, if the coding parameter adjacent to simple bonding pad is each other, one refers to proximity identification Symbol can recognize the appropriate subset for making a reservation for simple bonding pad number of the simple bonding pad with predetermined relative location relationship with this, and This appropriate subset is used when using the coding parameter or predicting that this makes a reservation for the coding parameter of simple bonding pad.

According to other embodiments, as indicating the spatial sampling of the two-dimensional information signal by recursively Multiple Segmentation It with different size of simple bonding pad is depending on one first grammer in the data flow that local area and space, which are subdivided into multiple, Subset of elements and execute, then be depending on first subset be not connected in the data flow one second grammer member Sub-prime collection and interblock space will be subdivided into the simple connection being not connected with adjacent to simple bonding pad to obtain among the sample array Area's set, joint are multiple simple bonding pad.When centre subdivision is used in from the data stream reconstruction sample array.Such as This optimization for allowing for for should segmenting does not have critical importance more, and reason is that actually meticulous subdivision can lead to Subsequent merging is crossed to be compensated.In addition, subdivision allow to reach with combined combination it is impossible separately through recurrence Multiple Segmentation The intermediate subdivision reached, therefore subdivision and combined cascade are executed by using the syntactic element set being not connected with (concatenation) allow effectively or intermediate subdivision more preferably adjusts the actual content for adapting to the two-dimensional information signal.With it Advantage compares, and is insignificant by the overhead caused by being used to indicate the additional syntactic element subset for merging details.

Detailed description of the invention

In the following, describing the preferred embodiment of the present invention for the following drawings, in which:

Block diagram of Fig. 1 display according to the embodiment encoder of the application；

Block diagram of Fig. 2 display according to the embodiment decoder of the application；

Fig. 3 A to Fig. 3 C schematically shows a specific embodiment of quaternary tree subdivision, and wherein Fig. 3 A shows the first class type Level, Fig. 3 B show that second-order laminar level and Fig. 3 C show third class formula level；

Fig. 4 schematically shows the tree construction segmented according to an embodiment for the illustrative quaternary tree of Fig. 3 A to Fig. 3 C；

Fig. 5 A, Fig. 5 B schematically show the quaternary tree subdivision of Fig. 3 A to Fig. 3 C and with the indexs for indicating individual leaf blocks Tree construction；

Fig. 6 A, Fig. 6 B scheme four forks for schematically showing tree construction and Fig. 3 A to Fig. 3 C that Fig. 4 is indicated according to different embodiments Set the binary string or flag sequence of subdivision；

Fig. 7 shows a flow chart, and display is according to embodiment step as performed by data flow withdrawal device；

Fig. 8 shows a flow chart, illustrates the function of the data flow withdrawal device according to another embodiment；

Schematic diagram of Fig. 9 A, Fig. 9 B display according to embodiment illustrative quaternary tree subdivision, emphasizes a predetermined block Neighbor candidate block；

Figure 10 shows a flow chart of the function according to another embodiment data flow withdrawal device；

Figure 11 schematically shows the composition of the image according to an embodiment in plane and plane group and illustrates and says The coding of bright across the plane adaptation/prediction of use；

Figure 12 A and Figure 12 B are schematically illustrated according to a sub-tree structure of an embodiment and corresponding subdivision to describe Succession scheme；

Figure 12 C and 12D schematically illustrate the sub-tree structure according to an embodiment describe respectively using using and The succession scheme of prediction；

Figure 13 shows a flow chart, and display realizes step performed by succession scheme by encoder according to an embodiment；

Figure 14 A and Figure 14 B show primary subdivision and subordinate subdivision, to illustrate according to embodiment specific implementation association Across-prediction one inherit scheme a possibility that；

Figure 15 shows that a block diagram illustrates a kind of coding/decoding method that the succession scheme is associated with according to an embodiment；

Figure 16 shows a schematic diagram, illustrates the scanning sequency according to an embodiment in multiway tree subdivision subinterval, should Sub-district is to undergo interior-prediction；

Figure 17 shows the block diagrams of decoder according to the embodiment；

Figure 18 A to Figure 18 C shows a schematic diagram, illustrates the subdivision possibility different according to other embodiments；

Specific embodiment

Later in the detailed description of attached drawing, appearing in the component between several pieces of drawings is indicated with common component symbol to keep away Exempt from these components of repeated explanation.The explanation of the component in relation to presenting inside an attached drawing is also applied for wherein occurring individual groups instead Other attached drawings of part, as long as deviation therein is pointed out in the explanation that other attached drawings are presented herein.

In addition, being explained later starts from regard to Fig. 1 to Figure 11 encoder explained and decoder embodiments.This attached drawing is presented Embodiment combines many aspects of the application, if but encoding scheme inside is implemented individually it is also excellent, in this way, just subsequent attached Figure, for embodiment by the aforementioned individual aspects of short discussion, this embodiment is the implementation described with different meaning representations with regard to Fig. 1 and Figure 11 The abstract of example.

Fig. 1 shows the encoder of embodiment according to the present invention.The encoder 1010 of Fig. 1 includes that a fallout predictor 12, one is residual Poor precoder 14, a residual error reconstructor 16, a data flow inserter 18 and a block divider 20.Encoder 10 is to use With by a space-time sample information Signal coding at a data flow 22.Temporal and spatial sampling information signal may be, for example, video, that is, a figure As sequence.Each image indicates an image sample array.When other examples of space time information signal are for example including by such as light (time-of-light) the depth image of camera shooting.Again notably a spatial sampling information signal may include each frame or when Between stab more than one array, such as in the case where color video, color video is for example including each one luma samples array of frame Together with two chroma sample arrays.May also the Temporal samplings of different components (that is, brightness and coloration) to information signal can It can be different.Similarly, it is suitable for spatial resolution.Video can also be accompanied by exceptional space sample information, such as depth or transparency Information.But it is of the invention that the processing that the focus-of-attention being hereinafter described will focus on one of this array carrys out understanding clearer first Purport then turns to the processing of more than one plane.

The encoder 10 of Fig. 1 is configured to form data flow 22, exists so that the syntactic element in data flow 22 describes granularity Image between full images and individual image samples.In order to reach this purpose, divider 20 is configured as each image 24 being subdivided into Different size of simple bonding pad 26.Hereinafter, this area will be simply referred to as block or sub-district 26.

As being described in detail later, image 24 is subdivided into various sizes of block 26 using multiway tree subdivision by divider 20. In more detail, it is segmented hereinafter with regard to the specific embodiment of Fig. 1 to Figure 11 institute outline is most of using quaternary tree.As being described in detail later, May include that the cascade of subdivider 28 is used to image 24 being subdivided into aforementioned block 26 inside divider 20, then for combiner 30 its Allow this block 26 being combined into group to obtain between the subdivision of image 24 defined without subdivision and subdivider 28 It is effective subdivision or granularity.

If the dotted line of Fig. 1 is for example, fallout predictor 12, residual error precoder 14, residual error reconstructor 16 and data flow insertion Device 18 is operated in the image subdivision defined by divider 20.For example, as being described in detail later, 12 use of fallout predictor by Divider 20 defined prediction subdivision come be directed to prediction subdivision individual sub-districts and deciding whether individual sub-districts should undergo With according to selected prediction mode for individual sub-districts corresponding Prediction Parameters setting value image interior prediction or Across image prediction.

Residual error precoder 14 transfers the residual error sub-district using image 24 again to encode the image as provided by fallout predictor 12 24 prediction residual.The syntactic element reconstructive residual error that residual error reconstructor 16 is exported from residual error precoder 14, residual error reconstructor 16 also operate in the subdivision of aforementioned residual error.Data flow inserter 18 can utilize previous segmentation, that is, prediction and residual error subdivision, come sharp Determine that insertion sequence and proximity relations between syntactic element are used for residual error precoder 14 and fallout predictor 12 with such as entropy coding The syntactic element of output is inserted into data flow 22.

As shown in Figure 1, encoder 10 includes an input terminal 32, the original information signal enters encoder 10 herein.One subtracts Musical instruments used in a Buddhist or Taoist mass 34, residual error precoder 14 and data flow inserter 18 in the input terminal 32 of data flow inserter 18 and are compiled with the sequence It connects between the output end that code data flow 22 exports.Subtracter 34 and residual error precoder 14 are a part of prediction loop, this is pre- Survey time road is surrounded by residual error reconstructor 16, adder and fallout predictor 12, these components are to be prelisted with the sequence in residual error It connects between the output end and the inverting input terminal of subtracter 34 of code device 14.The output end of fallout predictor 12 is also to be connected to adder 36 Another input terminal.In addition, fallout predictor 12 includes being connected directly to an input terminal of input terminal 32 and may include and another input End is also to be connected to the output end of adder 36 via filter 38 in optional circuit.In addition, fallout predictor 12 is in operation Period generates side information, therefore the output end of fallout predictor 12 is also coupled to data flow inserter 18.Similarly, divider 20 includes One output end its be another input terminal for being connected to data flow inserter 18.

The structure of encoder 10 has been described, the further details of operation mode are described in detail as after.

It has been observed that divider 20 determines how image is subdivided into cell 26 for each image 24.It is pre- according to being intended to be used for How the subdivision of the image 24 of survey, fallout predictor 12 predict individual cells for each cell decision for corresponding to such subdivision. Fallout predictor 12 exports inverting input terminal of the prediction to subtracter 34 of cell, and output to the another input terminal of adder 36, and It will reflect how fallout predictor 12 exports from the predictive information that the previous coding part of video obtains the mode that this is predicted to data Flow inserter 18.

In the output end of subtracter 34, so acquisition prediction residual, wherein residual error precoder 14 is according to also by dividing The residual error of 20 defined of device is segmented to handle such prediction residual.It is such as further detailed for Fig. 3 A to Fig. 3 C, Figure 10 below Describe in detail bright, the subdivision of the residual error of the image 24 as used in residual error precoder 14 can be with prediction subdivision phase used in fallout predictor 12 It closes, so that each prediction sub-district is employed as residual error sub-district or further subdivision becomes smaller residual error sub-district.It is possible that complete Complete independent prediction and residual error subdivision.

Each residual error sub-district is undergone transformation from space to frequency domain by two-dimensional transform by residual error precoder 14, then for Or peculiarly it is related to the gained quantization of transform coefficients of gained transform blockiis, therefore distorted result comes from quantizing noise.For example, Data flow inserter 18 can be used (for example) entropy coding and the syntactic element for describing foregoing transformation coefficient be nondestructively encoded into number According to stream 22.

Residual error reconstructor 16 is then again to convert again using re-quantization, and transformation coefficient is re-converted into residual signals, In the residual signals be that prediction residual is obtained by the resulting prediction of subtracter 34 in adder 36 internal combination, pass through this and obtain Reconstruction part or sub-district in the current image of output end one of adder 36.The reconstruction image sub-district can be used directly in fallout predictor 12 For interior-prediction, in other words, for by by being used to predict that some is pre- in the neighbouring prediction sub-district extrapolation rebuild from previously Survey sub-district.But by by from neighbouring spectrum prediction current sub-district frequency spectrum and directly carry out inside frequency domain in-prediction theory on Belonging to may.

For interaction prediction, image version that is previously encoded and rebuilding is can be used in fallout predictor 12, has been passed through accordingly Filter 38 filters in selective circuit.Filter 38 for example may include a solution blocking filtering device and/or an adaptive filter, Has the function of the transfer for being suitble to excellently form aforementioned quantizing noise.

Fallout predictor 12 selects Prediction Parameters, and display predicts certain and by using compared with the original sample inside image 24 The mode of a prediction sub-district.As being described in detail later, Prediction Parameters may include the instruction of prediction mode to each prediction sub-district, such as In image-predict and across image prediction.In image-prediction in the case where, Prediction Parameters also include be intended in-prediction prediction son The angle instruction that area inner edge mainly extends；And in the case where across image prediction, motion vector, moving image index and final height Power motion transform parameter；And inside image and/or in the case where across image prediction the two, for filtering reconstructed image sample Selective filter information, based on this predictable prediction sub-district at present.

As being described in detail later, the aforementioned subdivision defined by divider 20 substantially influence by residual error precoder 14, The maximum rate that fallout predictor 12 and data flow inserter 18 can be reached/distortion ratio.In the case where subdivision is too thin, by fallout predictor 12 The Prediction Parameters 40 for exporting data flow 22 to be inserted into need excessively high code rate, but can by the resulting prediction of fallout predictor 12 Preferably, and it smaller may can make it that can encode by less bits by the residual signals to be encoded of residual error precoder 14.Thin In the case where point too coarse, then contrary circumstance is applicable in.It is segmented in addition, aforementioned thinking is also suitable for residual error in a similar manner: figure As converting using the finer grain of individual transform block, as a result cause to become obtained by complexity reduction and result for operation transform The spatial resolution changed increases.In other words, less residual error sub-district allows the frequency spectrum of the content inside individual residual error sub-districts to distribute It is more consistent.However, spectral resolution lowers, the ratio of significant coefficient and not significant coefficient (that is, being quantified as zero) is deteriorated.Change speech It, transformation granularity must adjust adaptation local image content.In addition, independently of the positive effect of more fine particle size, more fine particle size Increase the side information content of needs regularly to indicate to subdivision selected by the decoder.As being described in detail later, aftermentioned implementation Example provides extremely effective adjustment to encoder 10 and adapts to subdivision to encoded information signal content is intended to, and is inserted by designation date stream Enter device 18 and subdivided information insertion data flow 22 is carried out into the subdivision that signal notifies decoding end to be used in.Details shows as follows.

But before the subdivision for defining divider 20 with further details, the decoder of embodiment according to the present invention is put up with Fig. 2 is described in detail with further details.

The decoder of Fig. 2 is indicated with reference label 100 and including a withdrawal device 102, a divider 104, a residual error weight Build filter 112 and a selective postfilter 114 in the selective circuit of device 106, an adder 108, a fallout predictor 110, one. Withdrawal device 102 receives encoded data stream in the input terminal 116 of decoder 100, and extracts subdivided information from the encoded data stream 118, Prediction Parameters 120 and residual error data 122, withdrawal device 102 respectively export these information to image segmentating device 104, prediction Device 110 and residual error reconstructor 106.The first input end that there is residual error reconstructor 106 output end to be connected to adder 108.Add Another input terminal and its output end of musical instruments used in a Buddhist or Taoist mass 108 are to be connected to a prediction loop, are filtered in selective circuit in the prediction loop Wave device 112 and fallout predictor 110 are the bypass paths with the sequential series from the output end of adder 108 to fallout predictor 110, directly Connect and be similar to the connection between adder 36 and fallout predictor 12 described in Fig. 1 above, in other words, one in image-prediction and Another one is used for across image prediction.The output end of filter 112 can be connected in the output end of adder 108 or selective circuit The output end 124 of decoder 100, reconstruction information signal is, for example, to export to transcriber herein.Selective postfilter 114 can The path of guiding output end 124 is connected to improve the visual quality in the eye impressions of the reconstruction signal of output end 124.

Outline says it, the component 16,36 of the similar Fig. 1 of the effect of residual error reconstructor 106, adder 108 and fallout predictor 110 And 12.In other words, the equally operation of emulation 1 component of earlier figures.In order to reach this purpose, residual error reconstructor 106 and fallout predictor 110 by Prediction Parameters 120, and the subdivision indicated by image segmentating device 104 according to the subdivided information 118 for deriving from withdrawal device 102 It is controlled, to carry out or determine that the same way carried out predicts the prediction sub-district with fallout predictor 12, and as residual error prelists The mode of code device 14 is remapped the received transformation coefficient of institute with same particle sizes.Image segmentating device 104 is dependent on subdivided information again 118 rebuild the subdivision as selected by divider 20 in a synchronous manner.Subdivided information can be used to control data pick-up in withdrawal device, all As with regard to context selection, neighbouring decision, probability estimation, the anatomy of data flow grammer etc..

Several deviations can be carried out to previous embodiment.Certain deviations will be described in detail later with regard to performed by subdivider 28 Subdivision and combiner 30 performed by address in merging, and other deviations are put up with subsequent Figure 12 to Figure 16 and are explained.Appoint in nothing In the presence of what obstacle, all these deviations all individually or can be in subset and be applied to the detailed description portion of Fig. 1 above and Fig. 2 Point.For example, divider 20 and 104 and determined prediction subdivision and only determine each image residual error subdivision.Instead its May filtering subdivision be determined for filter 38 and 112 in selective circuit respectively.Other prediction subdivisions or other residual codings Subdivision is independent unrelated or has dependence.In addition, determining that subdivision may not be to carry out based on frame by frame by these components.Instead And the following frame of some number can be reused or be used in the subdivision carried out to some frame, the simple new subdivision of then transfer.

In providing further details of the related image segmentation as sub-district, being explained later is that focus is concentrated first The subdivided portions being responsible for are estimated in subdivider 28 and 104a.Then the responsible progress of combiner 30 and combiner 104b institute is described Merging treatment program.Finally, across the plane adaptation/prediction of description.

The mode of subdivider 28 and 104a segmented image makes it possible that an image is divisible with different size of more A block is for image or prediction and the residual coding of video data.It has been observed that image 24 can be used as one or more images Sample value array.In the case where the color space YUV/YCbCr, such as the first array can indicate luminance channel, and another two numbers Group indicates chrominance channe.These arrays can have different dimensions.Whole arrays can be grouped into one or more plane groups, Ge Geping Face group is made of one or more continuous levels, so that each plane is included in one and only one plane group.Afterwards Text is suitable for each plane group.First array of one specific plane group can be referred to as an array of a plane group thus. Possible subsequent array is subordinate array.The block segmentation of array can be carried out based on quaternary tree method, as being described in detail later. Subordinate array block segmentation can the segmentation based on an array and lead calculating.

According to aftermentioned embodiment, subdivider 28 and 104a are configured as an array being partitioned into multiple equal sizes Square blocks are hereinafter referred to as tree block.When using quaternary tree, the side length for setting block is typically 2 multiple, such as 16, 32 or 64.But for completeness, notably using it is other types of tree and binary tree or have any number of sheets purpose tree all belong to possibility. In addition, the filial generation number of tree may depend on the level of tree and which kind of signal of the tree representation depended on.

Furthermore as explained earlier, sample array can also respectively indicate the other information other than video sequence, such as depth map Or light field.Concise to ask, the focus being explained later is the typical example for focusing on quaternary tree as multiway tree.Quaternary tree is In each internal node, just there are four the trees of filial generation.Each tree block forms a quaternary tree together in a quaternary tree Each leaf subordinate quaternary tree.Quaternary tree determines that the subdivision of the given tree block is used to predict, and subordinate quaternary tree Determine the subdivision of a given prediction tree block to residual coding.

The root node of quaternary tree is corresponding with completely tree block.For example, Fig. 3 A shows a tree block 150. It must remember, each image is divided into the column of such tree block 150 and the regular grid of row, thus seamlessly Covering samples number Group.But notably for the whole blocks subdivision shown hereinafter, the seamless subdivision without overlapping does not have critical importance.It is adjacent instead Near region block can overlap each other, as long as having no the appropriate subdivision that leaf block is neighbouring leaf block.

Together with the quad-tree structure of tree block 150, each node can be further divided into as four child nodes, at one time four In the case where fork tree, indicate that tree block 150 can split into four sub-blocks, half-breadth and half with tree block 150 are high.In Fig. 3 A In, these sub-blocks are with reference label 152a to 152d instruction.In the same manner, these sub-blocks are respectively further divided again It is cut into the half-breadth that there is original sub-block for four smaller sub-blocks and half height.It in figure 3b, is lifted for sub-block 152c Example display, sub-block 152c are subdivided into as four small-sized sub-block 154a to 154d.To so far, Fig. 3 A to Fig. 3 C is shown How tree block 150 is first separated into as four sub-block 152a to 152d, and then sub-block 152c in lower-left is divided into again Four small-sized sub-block 154a to 154d；And it is last as shown in Figure 3 C, the upper right block 154b of these small-sized sub-blocks once again by Four blocks are partitioned into, respectively 1/8th width with raw element tree block 150 and 1/8th height, these are again smaller Sub-block indicate with 156a to 156d.

Fig. 4 shows the potential tree construction as shown in Fig. 3 A to Fig. 3 C based on Quadtree Partition example.Number by tree node For so-called subdivided mark value, will be further described when quad-tree structure signal discussed hereinafter notifies.The root of quaternary tree Node is shown in the figure top (being denoted as level " 0 ").This root node is and four sons shown in Fig. 3 A in four branches of level 1 Block is corresponding.Because the third party in these sub-blocks is subdivided into its four sub-blocks again in figure 3b, Fig. 4 is in level 1 Third node also has there are four branch.Once again, corresponding with the subdivision of second (upper right) child node of Fig. 3 C, there are four sub-branches It is connected to the second node of quaternary tree stratum level 2.It is not segmented further in the node of level 3.

Each leaf of quaternary tree be with item forecast parameter (that is, it is internal or across, prediction mode, kinematic parameter etc.) can The variable size block being specified is corresponding.Hereinafter, these blocks are referred to as prediction block.Especially, these leaf blocks are Fig. 3 C Shown block.The explanation of Fig. 1 and Fig. 2 are briefly referred back to, divider 20 or subdivider 28 determine the quaternary tree as explained above Subdivision.Which is by thin again in subdivider 152a-d execution tree block 150, sub-block 152a-d, small-sized sub-block 154a-d etc. The decision divided or further divided, target are obtained as indicated too tiny prediction subdivision and too coarse prediction subdivision above Between optimal compromise.Fallout predictor 12 is transferred using specified prediction subdivision come the granularity to segment according to prediction or for for example The each prediction sub-district that block shown in Fig. 3 C indicates determines aforementioned Prediction Parameters.

Prediction block shown in Fig. 3 C can be further divided into smaller block to residual coding.For each prediction block, That is, it is directed to each leaf node of a quaternary tree, it is determined by one or more subordinate quaternary trees for residual coding corresponding Subdivision.For example, one given 32 × 32 prediction blocks will be divided into four 16 when allowing 16 × 16 maximum residul difference block size × 16 blocks individually pass through the subordinate quaternary tree decision for residual coding.In this example each 16 × 16 block be with from The root node for belonging to quaternary tree is corresponding.

Described in the case where being subdivided into prediction block just like a given tree block, subordinate quaternary tree point is can be used in each prediction block Solution is divided into multiple residual error blocks.Each leaf of one subordinate quaternary tree corresponds to a residual error block, can to the residual error block Shown individual residual coding parameters (that is, pattern conversion, transformation coefficient etc.) by residual error precoder 14, the residual errors such as this are compiled Code parameter transfers to control residual error reconstructor 16 and 106 respectively again.

In other words, subdivider 28 can be configured to for each image or for each groups of pictures determine a prediction subdivision and one from Belong to prediction subdivision, can divide the image into first tree block 150 it is regularly arranged, by quaternary tree subdivision recursively subregion this One subset of a little tree blocks is subdivided into prediction block to obtain prediction, if not carrying out subregion in tree block individually, the Target area Block can be tree block, and a subset of these prediction blocks is further segmented after probable, then be the leaf block of quaternary tree subdivision；Together Reason a, if prediction block is greater than the full-size of subordinate residual error subdivision, via item forecast block is divided into subtree first Block it is regularly arranged, then segment program according to quaternary tree, the subdivision of a subset of these subtree blocks obtained into residual error area Block, if not being split into subtree block in item forecast block, which can be prediction block, if or this small pin for the case Tree block is not split into smaller area again, then the residual error block is subtree block, or the leaf block for the subdivision of residual error quaternary tree.

Such as outline above, subordinate array can be mapped to for subdivision selected by an array.When consideration and an array This puts relatively easy when the subordinate array of identical dimensional.But it must be adopted when subordinate array dimension is with an array dimension difference Use special measure.Outline says it, and under different dimensional conditions, an array subdivision, which maps to subordinate array, can pass through space reflection It carries out, that is, via the block border space reflection for segmenting an array to subordinate array.Especially, for each subordinate number Group can have scaling factor in the horizontal direction and vertical direction, determine an array to the dimension ratio of subordinate array.Subordinate array It is divided into for predicting and the sub-block of residual coding can pass through the co-location tree block of a quaternary tree, array respectively Respective subordinate quaternary tree determines that tree block is calibrated by the relative calibration factor obtained by subordinate array.When horizontal direction and hang down Histogram to scaling factor different (such as in 4:2:2 coloration time sampling) when, the prediction block and residual error of gained subordinate array Block will no longer be rectangular.In such cases, can predetermine or adaptability selection (for entire sequence, in the sequence one A image or for each Individual forecast block or residual error block) whether non-square residual error block should split into square blocks.Example Such as, in the first case, encoder and decoder will agree to the side of being subdivided into subdivision when mapping block is not rectangular every time Shape block.In a second situation, subdivider 28 will lead to via data flow inserter 18 and data flow 22 to subdivider 104a signal Know the selection.For example, subordinate array has the half-breadth of an array but contour, residual error in the case where 4:2:2 coloration time sampling The height of block is twice of width.By the way that two square blocks can be obtained once again by this block longitudinal splitting.

It has been observed that subdivider 28 or divider 20 are based on four to subdivider 104a signal notice via data flow 22 respectively Pitch the segmentation of tree.In order to reach this purpose, 28 notification data stream inserter 18 of subdivider is related for selected by image 24 Subdivision.The structure of data flow inserter and transmission primaries quaternary tree and secondary quaternary tree, therefore, transmission picture number group are partitioned into Variable size block for inside data flow or bit stream 22 prediction block or residual error block to decoding end.

The transmission of minimum and maximum admissible block size is as side information and can change according to different images.Alternatively, Minimum and maximum allows block size that can fix in encoder and decoder.These minimum and maximum block sizes can be for prediction Block and residual error block and it is different.For the signal notice of quad-tree structure, quaternary tree must be traversed, must for each node Must show whether this specific node is a leaf node (that is, corresponding block does not segment further) for quaternary tree or this is specific Whether node is branched off into its four child node (that is, corresponding block with double of sized divisions at four sub-blocks).

Signal notice inside one image is to be carried out with raster scan order by tree block, such as from left to right and by upper Under, if Fig. 5 A is in 140 displays.Such scanning sequency can also be different, such as are carried out in a manner of chessboard from bottom right to upper left.Compared with In good embodiment, it is each set block and thus each quaternary tree be to be used to signal with depth-first fashion traversal to notify the subdivided information.

In the preferred embodiment, not only subdivided information (that is, tree construction), while prediction data etc. (that is, the leaf with the tree The associated payload of node) it is with depth-first grade transmission/processing.The reason of so carrying out is that depth-first traversal has There is the advantages of better than breadth-first.In figure 5B, quad-tree structure be denoted as with leaf node a, b ..., j present.Fig. 5 A is shown The segmentation of gained block.If block/leaf node is to obtain following order: abjchidefg with breadth-first order traversal.But it presses According to depth-first sequence, this is sequentially abc ... ij.Such as from Fig. 5 A it is found that left adjacent block and top are adjacent according to depth-first sequence Always transmission/the processing before current block of near region block.In this way, motion vector prediction and context modeling can always be used to a left side And parameter specified by adjacent block is pushed up to reach improvement coding efficiency.For breadth-first sequence, not such situation, reason It is that block j is, for example, to transmit before block e, g and i.

As a result, the signal for each tree block notifies it is the quad-tree structure recurrence progress along a quaternary tree, so that It is marked for each node-node transmission one, shows whether corresponding block splits into four sub-blocks.If this label has value " 1 " (being used for "true"), then this signal advising process recursively repeats all four child nodes, that is, sub-block is suitable with raster scanning Leaf node of the sequence (upper left, upper right, lower-left, bottom right) until reaching a quaternary tree.Notice that leaf node is characterized by subdivision The value of label is " 0 ".For node be reside in a quaternary tree lowest-order laminar level and so correspond to minimum allow it is pre- The case where surveying block size, without transmission subdivided mark.For the example of Fig. 3 A to Fig. 3 C, as shown in the 190 of Fig. 6 A, first will Transmission " 1 " shows that setting block 150 is split into as its four sub-block 152a-d.Then, recursively with raster scan order 200 Encode the subdivided information of all four sub-block 152a-d.Show it not for transmission " 0 " for first two sub-blocks 152a, b It is segmented (with reference in Fig. 6 A 202).For third sub-block 152c (lower-left), by transmission " 1 ", show that this block is through segmenting (with reference in Fig. 6 A 204).Now according to recurrence method, four sub-block 154a-d of this block will be handled.It will be directed to herein First sub-block (206) transmits " 0 " and for second (upper right) sub-block (208) transmission " 1 ".The minimum sub-block of present Fig. 3 C Four block 156a-d of size will be processed.If the minimum for having reached this example allows block size, need not transmit again Data, reason are impossible further segment.Otherwise, show that " 0000 " that these blocks do not segment further will transmit, If Fig. 6 A is in 210 instructions.Then, by the encrypted communication " 00 " of lower section two (with reference in Fig. 6 A 212) to Fig. 3 B, and finally to figure The bottom right encrypted communication " 0 " (referring to 214) of 3A.Therefore indicate that the complete binary string of quad-tree structure will be for shown in Fig. 6 A.

Such binary string of Fig. 6 A indicates that the different background shade of kenel corresponds to the stratum segmented based on quaternary tree Different levels in relationship.Shade 216 indicates level 0 (correspond to block size and be equal to raw element tree block size), 218 table of shade Show level 1 (correspond to block size and be equal to raw element tree block size half), shade 220 indicates that level 2 (corresponds to block size Equal to a quarter of raw element tree block size), shade 222 indicates that level 3 (it is big equal to raw element tree block to correspond to block size Small 1/8th).Identical hierarchical level (corresponding to the same block size indicated in example binary string in kenel and Same hue) whole subdivided marks for example can be by inserter 18 using one and the same probability model does entropy coding.

The case where paying attention to for breadth first traversal, subdivided information will be transmitted with different order, be shown in Fig. 6 B.

Similar to the subdivision of each tree block for prediction, each gained prediction block is divided into residual error block must be in position It spreads defeated.There can be maximum and minimum block big for the residual coding that transmits and may change according to image as side information again It is small.Or it can be fixed for the maximum and minimum block size of residual coding in encoder and decoder.In each of quaternary tree A leaf node, as shown in Figure 3 C, corresponding prediction block may be partitioned into the residual error block of maximum allowable size.These blocks be from Belong to the composition root node that quad-tree structure is used for residual coding.For example, if the maximum residul difference block size of image be 64 × 64 and prediction block size be 32 × 32, then entire prediction block will correspond to size 32 × 32 a subordinate (residual error) four Pitch root vertex.On the other hand, if the maximum residul difference block for image is that 16 × 16,32 × 32 prediction blocks will be by four Residual error quaternary tree root node is formed, and respectively has 16 × 16 size.Inside each prediction block, subordinate quad-tree structure Signal notice be with raster scan order (left-to-right, under) by root node carry out.Similar to primary (prediction) quaternary tree The case where structure, for each node, one label of coding shows whether this specific node divides as four child nodes.Then If this label has value " 1 ", for all four corresponding child nodes and its corresponding sub-block with raster scan order (upper left, upper right, lower-left, bottom right) recursively repeats the leaf node until reaching subordinate quaternary tree.Such as the feelings of a quaternary tree Condition notifies that reason is that these nodes are corresponding for the node in subordinate quaternary tree lowest-order laminar level without signal In it is minimum may residual error block size block and can not further divide.

For entropy coding, the residual error block subdivided mark for belonging to the residual error block of same block size can be used one and same One probability model coding.

In this way, according to the example presented above with regard to Fig. 3 A to Fig. 6 A, subdivider 28 defines the primary subdivision for prediction And the sub- subordinate subdivision with different size block once segmented for residual coding purpose.Data flow inserter 18 is logical It crosses and notifies that bit sequence is established according to Fig. 6 A, even to encode primary subdivision for each tree block with zigzag scan order signal The block size of maximum once segmented and maximum hierarchical level are encoded together.For each prediction block so defined, Associated Prediction Parameters have been included in bit stream.In addition, similar information (that is, full-size size, maximum according to Fig. 6 A Hierarchical level and bit sequence) coding can for each prediction block carry out, the size of the prediction block be equal to or be less than it is residual The full-size size of difference subdivision；And carried out for each residual tree root zone block, wherein prediction block is divided in advance At more than the full-size size defined to residual error block.For each residual error block so defined, residual error data is slotting Enter the data flow.

Withdrawal device 102 extracts individual bit sequences from the data flow in input terminal 116 and notice divider 104 is related so Resulting subdivided information.In addition, aforementioned sequence can be used to be used in prediction block and residual error for data flow inserter 18 and withdrawal device 102 Transmit additional syntactic element between block, the residual error data that is such as exported by residual error precoder 14 and by fallout predictor 12 The Prediction Parameters exported.Using the advantages of such sequence be by using adjacent block coding/decoding grammer member The suitable context for some block coding individual grammar element may be selected in element.In addition, similarly, residual error precoder 14 and Fallout predictor 12 and residual error reconstructor 106 and precoder 110 can the sequential processes item forecast blocks of outline and residual above Poor block.

Fig. 7 shows the flow chart of step, which can execute the side stated come digest before when encoding by withdrawal device 102 Formula extracts subdivided information from data flow 22.In the first step, image 24 is divided into tree root block 150 by withdrawal device 102. This step is with step 300 instruction in Fig. 7.It is big from the extraction maximum predicted block of data flow 22 that step 300 is related to withdrawal device 102 It is small.Further additionally or alternatively, step 300 can be related to withdrawal device 102 and extract maximum hierarchical level from data flow 22.

Next, in step 302, withdrawal device 102 marks or one from the data stream one.Carry out first time step Rapid 302, withdrawal device 102 knows to mark the position for belonging to the first tree root block 150 for tree root Reginal-block scanning sequence 140 suitable individually First label of sequence.Therefore should labeled as with hierarchical level 0 label, in step 302, withdrawal device 102 can be used with The associated context modeling of hierarchical level 0 determines a context.Each context have individual probability estimate for The entropy coding of its associated label.The probability estimation of context can individually adapt to context and count in individual contexts symbol Number.For example, withdrawal device 102 can in order to determine the suitable context for being used to decode the label of hierarchical level 0 in step 302 A context in a set of context is selected, is stratum associated with hierarchical level 0, depending on neighbouring tree block Formula level 0 marks, and is more dependent upon the neighbouring tree block (such as top and left neighbouring tree block) for defining processing tree block at present Information contained in the bit string of quaternary tree subdivision.

In next step, that is, step 304, withdrawal device 102 check that current decoder marks whether to prompt subregion.If belonging to such Situation, then withdrawal device 102 is by current block (at present for tree block) subregion, or in step 306 instruction a kind of this subregion to subdivision Device 104a checks whether current hierarchical level is equal to maximum hierarchical level and subtracts 1 in step 308.For example, withdrawal device 102 for example also have the maximum hierarchical level extracted in step 300 from data flow.If current hierarchical level is not equal to most Big hierarchical level subtracts 1, then comes current hierarchical level incremental 1 and return step 302 from this in step 310 withdrawal device 102 The next label of data stream.At this point, being intended to decoded label in step 302 belongs to another hierarchical level, therefore foundation One of different set of context may be selected in one embodiment, withdrawal device 102, which is to belong to current hierarchical level.It should Selection may be based on subdivision bit sequence of the decoded neighbouring tree block according to Fig. 6 A.

If the inspection of one label of decoding and step 304 discloses the subregion that this label does not prompt current block, withdrawal device 102 advance steps 312 check whether current hierarchical level is 0.If belonging to such situation, withdrawal device 102 with regard to step 314 according to Next tree root block of scanning sequency 140 is handled, if or do not leave any tree root block to be handled, stop handling Extract subdivided information.

It should be noted that the description focus of Fig. 7 is the decoding for segmenting cue mark for focusing on only prediction and segmenting, thus it is real On border, step 314 is related in relation to such as decoding of the associated other storehouses (bin) of tree block or syntactic element at present.This kind of feelings Under condition, there are another or next tree root block if it exists, then withdrawal device 102 advances to step 302 by step 314, from subdivision Information decodes next label, that is, the first label of the flag sequence in relation to new tree block.

In step 312, if hierarchical level is not equal to 0, operation advances to step 316, has checked for Close other child nodes of current node.In other words, it when withdrawal device 102 is when step 316 is checked, has been examined in step 312 Looking into current hierarchical level is the hierarchical level other than 0 hierarchical level.So transfer to indicate there are parent node again, It is to belong to tree root block 150 or small-sized block 152a-d or again one of smaller area block 152a-d etc..Current decoder marks institute The tree construction node of category has a parent node, which is shared by the other three node of the current tree construction.Tool There is the scanning sequency between these child nodes of a shared parent node to be illustrated in Fig. 3 A, there is ginseng for hierarchical level 0 Examine label 200.In this way, in step 316, withdrawal device 102 checks whether that all four child nodes are all in the processing journey of Fig. 7 It is accessed in sequence.If not belonging to such situation, that is, parent node has additional child node at present, then the processing routine of Fig. 7 advances to Step 318, this is in accessed according to next child node of zigzag scan order 200 inside current hierarchical level, because This its corresponding sub-block indicates the current block of Fig. 7 now, and then, saves in step 302 from related current block or at present The data stream one of point marks.However, in step 316, if having no additional child node, the side of Fig. 7 to current parent node Method advances to step 320, and current hierarchical level successively decreases 1 herein, wherein subsequent this method is with step 312 progress.

By executing step shown in Fig. 7, withdrawal device 102 and subdivider 104a cooperation of pulling together come in encoder-side from data flow Fetch selected subdivision.The method focus of Fig. 7 is concentrated the case where aforementioned prediction is segmented.The flow chart of constitutional diagram 7, figure How 8 display withdrawal devices 102 and subdivider 104a pull together cooperation to fetch residual error subdivision from data flow.

In specific words, Fig. 8 is shown for from each prediction block of prediction subdivision gained, passes through withdrawal device 102 and subdivider The step of 104a is carried out respectively.It has been observed that these prediction blocks are between the tree block 150 of prediction subdivision according to sawtooth scan Sequentially 140 traversals, and pass through tree using the depth-first traversal accessed at present inside each tree block 150 shown in such as Fig. 3 C Block.According to depth-first traversal sequence, the leaf block by the primary tree block of subregion is with the visit of depth-first traversal sequence Ask there is the sub-block for some hierarchical level for sharing current node with the access of zigzag scan order 200, and advancing The respective subdivision of these sub-blocks is mainly scanned first before to next sub-block of such zigzag scan order 200.

For the example of Fig. 3 C, gained scanning sequency is with the display of reference label 350 between setting the leaf node of block 150.

The processing routine of prediction block for accessing at present, Fig. 8 starts from step 400.In step 400, current area is indicated The inner parameter of the current size of block is set equal to the size of the hierarchical level 0 of residual error subdivision, that is, residual error subdivision is most Big block size.It must remember that maximum residul difference block size is smaller than the minimum block size of prediction subdivision, or can be equal to or be greater than The latter.In other words, according to one embodiment, encoder can the aforementioned any possibility of unrestricted choice.

In next step, that is, step 402, execution check whether the prediction block size in relation to access block at present is greater than It is denoted as the inner parameter of current size.It may be a leaf block or the prediction subdivision of prediction subdivision if belonging to such situation One tree block and the current access prediction block without any further subregion is greater than maximum residul difference block size, such situation Under, the processing routine of Fig. 8 advances to the step 300 of Fig. 7.In other words, the prediction block accessed at present is divided into residual error tree root First label of block, the flag sequence of the first residual tree block inside such current access prediction block is in step 302 Decoding etc..

But if there is access prediction block size to be equal to or less than the inner parameter for indicating current size at present, Fig. 8's Processing routine advances to step 404, checks prediction block size herein to determine whether it is equal to and indicate the inside of current size Parameter.If it has, then segmentation step 300 can skip, processing routine directly continues to the step 302 of Fig. 7.

If but the prediction block size of access prediction block is less than the inner parameter for indicating current size, Fig. 8 at present Processing routine advance to step 406, herein hierarchical level be incremented by 1, size is set as the size of new hierarchical level at present, Such as with 2 segmentations (downloading two axis directions in quaternary tree subdivision situation).Then, the inspection for carrying out step 404 once again, passes through step 404 and 406 be formed by circuit effect be hierarchical level regularly be intended to subregion corresponding block size it is corresponding, and It is independently unrelated with the item forecast block for being less than or equal to/being greater than maximum residul difference block size.In this way, when in step 302 When coding symbols, the context modeling carried out is the hierarchical level and block size for depending on label meaning simultaneously.Needle Probability estimation, which is extremely suitble to, to be using the advantages of different contexts respectively to the label of different estate formula level or block size On the other hand the practical probability distribution that mark value occurs has the moderate context number to be managed, thus reduce context pipe Expense is managed, and increases context adjustment and is adapted to actual symbol statistics.

It is such as already described above, there is more than one sample array, these sample arrays can be grouped as one or more plane groups. Such as into the desire coded input signal of input terminal 32 can be an image of video sequence or static images.Such image is In one or more sample array forms.In an image coding context of video sequence or static images, sample array is Refer to three color planes, such as red, green and blue, or refers to that luminance plane and colorimetric plane are for example indicated in the colour of YUV or YCbCr Kenel.In addition, the sample array for indicating the depth information of α (that is, transparency) and/or 3-D video data can also be presented.It is multiple These sample arrays can be grouped into so-called plane group together.For example, brightness (Y) can be only one sample array One plane group and coloration (such as YCbCr) can be for there are two another plane groups of sample array；Or in another example In, UV can be for there are three the depth informations of a plane group of matrix and 3-D video data can be only one sample number The Different Plane group of group.For each plane group, a quad-tree structure can be used in 22 in-line coding of data flow Expression is divided into prediction block；And it is directed to each prediction block, secondary quad-tree structure expression is divided into residual error block.In this way, According to aforementioned first example, luminance component is a plane group, and chromatic component forms another plane group herein, one four Fork tree construction is the prediction block for luminance plane, and a quad-tree structure is the residual error block for luminance plane, one It is the residual error block for being directed to colorimetric plane that quad-tree structure, which is for the prediction block of colorimetric plane and a quad-tree structure,. But in aforementioned second example, there may be a quad-tree structure for the prediction block (YUV) of brightness and coloration together, one Quad-tree structure is directed to the residual error block (YUV) of brightness and coloration together, and a quad-tree structure is directed to the depth of 3-D video data The prediction block and a quad-tree structure of spending information are directed to the residual error block of the depth information of 3-D video data.

In addition, input signal is to be divided into multiple prediction blocks using a quad-tree structure, now in being described above Describe how these prediction blocks using subordinate quad-tree structure are further subdivided into residual error block.According to another embodiment, Subdivision not terminates in subordinate quaternary tree grade.In other words, it may be used using subordinate quad-tree structure from resulting block is segmented Ternary quad-tree structure is further segmented.Such segmentation may assist again then used in the purpose using additional encoding tool The coding of residual signals.

Focus concentration is described above to be finely divided by subdivider 28 and subdivider 104a respectively.It has been observed that point The processing granularity of the module of controllable afore-mentioned code device 10 and decoder 100 is not finely divided by subdivider 28 and 104a.But According to embodiment described later, subdivider 228 and 104a are then combiner 30 and combiner 104b respectively.But notably Combiner 30 and 104b are selectivity and can exempt.

But actually and as being described in detail later, if combiner provides with will be in prediction block or residual error block encoder Dry person is combined into the chance of group or group variety, so that at least part in other modules or other modules can be by these blocks group Group is handled together.For example, fallout predictor 12 can be optimized by using the subdivision of subdivider 28 to sacrifice measured part Deviation between the Prediction Parameters of prediction block, and replaced using the Prediction Parameters shared to these whole prediction blocks, as long as Prediction block be grouped the signal notice together with the shared parameter transmission for the whole blocks for belonging to the group with regard to rate/distortion than in meaning and Speech, earth signal notices more individual than the Prediction Parameters of all these prediction blocks are with more prospect property.It is shared pre- based on these Parameter is surveyed, is still to be carried out one by one with prediction block in the processing routine itself that fallout predictor 12 and 110 fetches prediction.It is also possible to Fallout predictor 12 and 110 even Prediction program is once carried out to entire prediction block group.

As being described in detail later, it is also possible to which prediction block group is not used only for the identical or shared of one group of prediction block Prediction Parameters, instead additionally or alternatively, allow encoder 10 send for this group a Prediction Parameters together with right Belong to the prediction residual of the prediction block of this group, thus the letter for notifying the Prediction Parameters of this group for signal can be reduced Number notification overhead.In aftermentioned situation, consolidation procedure only influence data flow inserter 18 rather than influence by residual error precoder 14 and The decision that fallout predictor 12 is done.But further details as being described in detail later.However, for completeness, notably aforementioned aspects is also fitted For other subdivisions, such as aforementioned residual error subdivision or filtering subdivision.

Firstly, the merging of sample set (such as aforementioned prediction block and residual error block) is motivated with more typically property meaning, That is, it is not limited to aforementioned multiway tree subdivision.But subsequent illustrates that focus will focus on previous embodiment and segment gained by multiway tree The merging of block.

Outline says it, and the purpose for being used for transmission associated coding parameter merges grammer associated with specific sample set Element allows to apply upper reduction side information rate in image and Video coding.For example, the sample array for the signal to be encoded Be usually to divide specific sample set or sample set into, can indicate rectangle block or square blocks or sample it is any its It gathers, including arbitrary shape area, triangle or other shapes.In the aforementioned embodiment, simple bonding pad is thin from multiway tree Divide resulting prediction block and residual error block.The subdivision of sample array can be fixed by grammer；Or it has been observed that subdivision can also be at least Part is notified in bit stream internal signal.In order to which the side information rate for being used for signal notice subdivided information is maintained small, grammer is logical Allow a limited number of selection only often to lead to simple subregion, block subdivision is such as become into smaller block.Sample set Be it is associated with specific coding parameter, can be shown that predictive information or residual coding mode etc..Details in relation to this subject under discussion be as It is described above.For each sample set, individual codings ginseng such as showing predictive coding and/or residual coding can be transmitted Number.In order to reach improvement code efficiency, in terms of the merging being hereinafter described, it is so-called to be also merged into two or more sample sets Sample set group, allow to reach several advantages, as being described in detail later.For example, sample set can through merging so that this Whole sample sets of one group share identical coding parameter, can transmit together with one of the sample set in group.It is logical This mode is crossed, coding parameter need not individually be transmitted for each sample set in sample set group, be replaced instead, Coding parameter is only primary to entire sample set group transmission.As a result, the side information for being used to transmit coding parameter is reduced, and total Code efficiency can improve.As for instead road, the additional purification of one or more coding parameters can be directed to a sample set group In one or more sample sets transmission.The whole sample sets that can be applied in a group are refined, or is only applied to and is directed to Its sample set transmitted.

Also encoder is provided to form relatively high-freedom degree when bit stream 22, reason in terms of the merging further described hereinafter It is that merging method dramatically increases number a possibility that for selecting one image pattern array of subregion.Because encoder can be compared with multiselect It is selected between, such as reducing particular rate/distortion measurement, therefore code efficiency can be improved.Operation encoder has several possibility. In simple approach, encoder can determine the best subdivision of sample array first.Fig. 1 is briefly referred to, subdivider 28 will be in the first order Determine best subdivision.Then, it for each sample set, checks whether and merges with another sample set or another sample set group Reduce particular rate/distortion cost measure.In this connection, Prediction Parameters associated with a merging sample set group It can re-evaluate, such as by executing new movement search and estimation；Or shared sample set and candidate samples collection have been directed to it It closes or can be assessed for considered sample set group for the Prediction Parameters of combined sample set group after measured. In more comprehensive method, particular rate/distortion cost measure can assess additional candidate sample set group.

The merging method that is notably hereinafter described and the processing sequence for having not been changed sample set.In other words, merging conception can It is embodied in one way, so that delay is not further added by, that is, the maintenance of each sample set can be decoded in moment at the same time And merging method is not used.

For example, if being greater than additional consume in coding conjunction by the bit rate that reduction coding Prediction Parameters number is saved And information is used to refer to merge to the bit rate of decoding end, merging method (as being described in detail later) causes code efficiency to increase.Into one Step must refer to that described extend for combined grammer provides extra discretion to encoder to select image or plane group subregion At multiple blocks.In other words, whether encoder is not limited to be finely divided and then check first several persons in gained block With the identical set of Prediction Parameters or similar set.As for a simple instead road, according to rate-distortion cost measure, Encoder determines to segment first, and then can check adjacent block for each block coder or be associated after measured One of block group merge whether attenuating rate-distortion cost measure.In this way, can re-evaluate and the new block group Associated Prediction Parameters, such as by being searched by carrying out new movement；Or to current block and adjacent block or block group The determining Prediction Parameters of group can be assessed for new block group.Pooling information carries out signal notice as unit of block.Effectively Ground, merge can also be interpreted as current block Prediction Parameters inference as a result, wherein the Prediction Parameters of inference are set to Equal to the Prediction Parameters of one of adjacent block.In addition, residual error can be transmitted for the block in a block group.

In this way, the potential basic conception in merging conception lower section being described later on is to become an area by merging adjacent block Block group is come bit rate needed for lowering communicating predicted parameter or other coding parameters, and each block group is all with coding parameter herein As a unique set of Prediction Parameters or residual coding parameter is associated.Other than subdivided information (if present), pooling information Also it is notified in bit stream internal signal.The advantages of merging conception causes code efficiency to increase to reduce from the side information of coding parameter It is high.Notably merging method described herein also may extend to other dimensions other than Spatial Dimension.For example, in several differences This or block sets group are combinable at a block group as the inside of video image.Merging is equally applicable to 4- compression And light code field.

In this way, briefly refer back to the explanation of Fig. 1 to Fig. 8 above, pay attention to the consolidation procedure after subdivision be it is advantageous, And it is independently unrelated with the ad hoc fashion of subdivider 28 and 104a subdivision image.More clearly say it, H.264 the latter can also be similar to that Mode segment image, in other words, each image is subdivided into predetermined size such as 16 × 16 luma samples or in data flow Internal signal notifies the regularly arranged of the rectangular or square aggregation block of size, and each macro zone block has several volumes associated there Code parameter divides the regular sublattices of 1,2,4 or several other number of partitions into and is used as including for example defining for each macro zone block Prediction granularity and corresponding Prediction Parameters in bit stream and for defining point of residual error and corresponding real transform granularity The partitioned parameters in area.

To sum up, merging the advantages of providing short discussion above, such as reduce in image and Video coding application Side information rate position.Indicate rectangular or square block or arbitrary shape area or such as any simple connection of any other sample set The specific sample set of area or sample is commonly connected specific coding parameter sets；For each sample set, coding parameter is to contain It includes in bit stream, coding parameter for example indicates Prediction Parameters, and it provides corresponding sample set is how to be subject to using encoded sample Prediction.One image pattern array divides sample set into and can be fixed by grammer, or can be by corresponding inside the bit stream Subdivided information signal notice.It can be passed for the coding parameter of the sample set with predetermined order (that is, the given sequence of grammer) It is defeated.According to pooling function, combiner 30 can for a shared sample set or a current block (such as with one or more other samples The prediction block or residual error block that this set merges), 30 signal of combiner is notified into a sample set group.One group's sample The coding parameter of set therefore need only transmission primaries.In a particular embodiment, if sample set is and has transmitted coding at present This set as parameter or existing sample set group merge, then the coding parameter of current sample set does not transmit.Instead, mesh The coding parameter of preceding sample set is set to be equal to the sample set or the sample set that current sample set merges with it The coding parameter of group.As for instead road, the additional purification of one or more of coding parameter can be to current sample set Transmission.Purification can be applied to whole sample sets of a group or only be applied to the sample set for its transmission.

According to an embodiment, for each sample set (such as aforementioned prediction block, aforementioned residual error block or aforementioned multiway tree The leaf block of subdivision), whole previous codings/decoding sample set set is referred to as " set of cause and effect sample set ".Example Such as refer to Fig. 3 C.Whole blocks that this figure is shown are all certain subdivision as a result, such as prediction subdivision or residual error subdivision or any more Member tree subdivision etc., the coding/decoding sequence defined between these blocks are with the definition of arrow 350.Consider some between these blocks Block is current sample set or simple bonding pad at present, the set of cause and effect sample set be by along sequence 350 current Whole blocks in front of block are formed.As long as but must remember to consider to be not used polynary hereinafter in relation to the discussion of Unite principle Other subdivisions of tree subdivision also fall within possibility.

The sample set merged with current sample set be can be used in being hereinafter referred to as " the collection of candidate samples set Close ", a regular subset for " set of cause and effect sample set ".How to be formed the subset mode be decoder known to, Or it can show inside data flow or bit stream from encoder to decoder.If specific current sample set is encoded/decoding, The then set non-empty of this candidate samples set is to notify in encoder in data flow internal signal, or in decoder from the number Calculate whether the shared sample set merges with a sample set in the set of this candidate samples set according to conductance, and if so, It is to merge with any one in the sample set.Otherwise, merge and be not used to this block, reason is the collection of candidate samples set It closes regular for sky.

There are different modes to measure the subset for the set that will indicate candidate samples set in the set of cause and effect sample set. For example, the measurement of candidate samples set can be based on the sample inside current sample set, with unique geometry Definition, the upper left image sample of such as rectangle block or square blocks.It starts from such unique geometry and defines sample, determine special Determine non-zero number sample, indicates that such unique geometry defines the straight space of sample adjacent to sample.For example, Ci Zhongte Determine the upper neighbouring sample and left neighbouring sample of unique geometric definition sample that non-zero number sample includes current sample set, thus it is adjacent The non-zero number of nearly sample is at most 2, if one of upper or left neighbouring sample can not obtain or be located at outside the image Side, then non-zero number is 1；Alternatively, if missing two in the case where the sample, non-zero number 0.

The set of candidate samples set can cover at least one in the non-zero number containing aforementioned neighbouring sample through determining Those of person sample set.Such as with reference to Fig. 9 A.The sample set being presently considered is combining objects, must be block X, and its geometry Shape distinct definition sample must illustrate as upper left sample, with 400 instructions.The top of sample 400 and left neighbouring sample respectively refer to It is shown as 402 and 404.The set of cause and effect sample geometry or the set of cause and effect block are emphasized in a manner of shade by adding.Therefore these areas In block, block A and B include one of neighbouring sample 402 and 404, these blocks form candidate block set or candidate samples The set of set.

According to another embodiment, the set for merging the candidate samples set that purpose is determined can extraly or exclusively Including the sample set containing a specific non-zero number sample, which can be 1 or 2, and the two has same spatial location, still It is contained in different images, that is, previous coding/decoding image.For example, it other than the block A and B of Fig. 9 A, can be used previous The block of coded image comprising in the sample of the same position of sample 400.In this manner, paying attention to only upper neighbouring sample 404 or only left neighbouring sample 402 can be used to define the non-zero number of aforementioned neighbouring sample.Generally, candidate samples set Set can lead calculating from treated data previous inside current image or other images.Leading calculation may include direction in space Information, transformation coefficient such as associated with the specific direction of current image and image gradient；Or may include time orientation information, Such as adjacent to movement representation kenel.By these in the available data of receiver/decoder and other numbers inside data flow According to and side information (if present), can lead calculate candidate samples set set.

Notably candidate samples set is led at last by the combiner 30 in encoder-side and in the merging of decoder end Device 104b is executed side by side.Just it has been observed that the two can be unrelated independently of one another based on determining to the mode defined in advance known to the two Candidate samples set set；Or encoder can notify to imply clue in bit stream internal signal, be by combiner 104b band Come to a position to execute these in the same way of the combiner 30 for the set for determining candidate samples set in encoder-side Candidate samples set leads calculation.

As being described in detail later, combiner 30 and data flow inserter 18 are cooperated to transmit one or more to be directed to each sample set Syntactic element, shows whether the sample set merges with another sample set, which can be to have merged again A part of sample set group and the set of candidate samples set in any one be for merging.Withdrawal device 102 Then these syntactic elements and notice combiner 104b are extracted accordingly.Especially, according to the specific embodiment being hereinafter described, for one Specific sample set transmits one or two syntactic element to show pooling information.First syntactic element shows that current sample set is It is no to merge with another sample set.If the first syntactic element shows that the current sample set is merged with another sample set, only The second syntactic element for having such situation just to transmit shows which one in the set of candidate samples set for merging.If leading calculation The collection of candidate samples set is combined into sky out, then can inhibit the transmission of the first syntactic element.In other words, if leading the candidate samples of calculating The set and non-empty of set, then have such situation only and just transmit the first syntactic element.Have the candidate samples set for leading calculating only The second syntactic element is just transmitted when set is containing more than one sample set, if reason is the set in the candidate samples set In contain only a sample set, then must not do further selection.Furthermore if the set of candidate samples set includes being more than One sample set, then the transmission of the second syntactic element can be suppressed；If but whole samples in the set of candidate samples set It is then no when set is associated with one and same coding parameter.In other words, the second syntactic element has the candidate sample for leading calculating one only It is just transmitted when at least two sample sets in the set of this set are associated with different coding parameter.

Inside the bit stream, the pooling information of a sample set can Prediction Parameters associated with the sample set or its It is encoded before its specific coding parameter.Prediction Parameters or coding parameter have only notifies current sample set to merge in pooling information signal It is just transmitted when not merging with any other sample set.

For example, the pooling information of some sample set (that is, a block) can encode after appropriate Prediction Parameters subset；Or More typically property defines, and has transmitted coding parameter associated with the individual sample sets.Prediction/coding parameter subset can be by one The one or more components or benchmark index and kinematic parameter vector of a or multiple reference picture indexes or kinematic parameter vector One or more components etc. formed.The Prediction Parameters or coding parameter subset transmitted can be used to from just like being described above It has led in the interim set of the biggish candidate samples set of calculation and has led the set for calculating a candidate samples set.An example is lifted, The encoded Prediction Parameters and coding parameter that current sample set can be calculated are joined with the corresponding prediction of previous candidate sample set Difference measurement between several or coding parameter or the distance according to preset distance measured value.Then, counted difference is only counted to survey Magnitude or distance are less than or equal to those of predetermined critical or the critical value for leading calculating sample set and are included finally collecting Close (that is, set of the candidate samples set reduced).Such as with reference to Fig. 9 A.Sample set must be block X at present.Related local area One subset of the coding parameter of block must be already inserted into bit stream 22.For example, it is assumed that block X is prediction block, and in the case of this kind, coding The appropriate subset of parameter can block X thus Prediction Parameters subset, it is such as (all including image reference index and motion map information Such as motion vector) one set in a subset.If block X is residual error block, the subset of coding parameter is residual information Collection, such as mapping table of the significant transformation coefficient position of transformation coefficient or instruction inside block X.Based on this information, data This information can be used to determine the subset in block A and B in both stream inserter 18 and withdrawal device 102, and the subset is in Ben Te Determine the preliminary set that aforementioned candidates sample set is constituted in embodiment.In specific words, because block A and B belong to cause and effect sample set Set, coding parameter block X coding parameter be current encoder/decoding when by both encoder and decoder can benefit With.Therefore, it can be used to exclude any number in the preliminary set of candidate samples set A and B using the aforementioned comparison of different methods Mesh block.Then, the set for the candidate samples set that gained reduces can use as explained earlier, in other words, for determining to merge Whether indicator, which indicates to merge or extract from the data flow from transmission in the data flow, merges, and depends in diminution candidate's sample Whether sample set number and the second syntactic element inside the set of this set must be depending on wherein transmitting；Or from It is extracted in the data flow, there is the second syntactic element instruction which sample set palpus inside the diminution candidate samples set To merge companion's block.

The aforementioned critical value that afore-mentioned distance is compared relative to it can for it is fixed and be both encoder and decoder institute Know, or can based on counted distance and lead calculating, the middle number of different value or it is several it is other take it is middle tendency etc..Such situation Under, inevitably, the set for reducing candidate samples set must be the appropriate subset of the preliminary set of candidate samples set.Separately Outside, having only according to distance measure is that those of minimum range sample set is just selected from the preliminary set of the candidate samples set Out.In addition, what a proper sample set only selected from the preliminary set of the candidate samples set using afore-mentioned distance measured value It closes.In aftermentioned situation, pooling information need only show which current sample set is intended to merge with single candidate samples set ?.

In this way, candidate block set can be as hereinafter with regard to forming or leading calculating shown in Fig. 9 A.Start from the current block of Fig. 9 A The upper left sample position 400 of X leads its left neighbouring 402 position of sample of calculating in encoder-side and decoder end and its pushes up neighbouring sample This 404 position.In this way, at most only there are two elements for candidate block set, that is, contain in the cause and effect set for adding picture shade of Fig. 9 A It is block B and A there are two the block in one of sample position (the case where belonging to Fig. 9 A).In this way, candidate block set is only Two direct neighbor blocks of the upper left sample position with current block are as its element.According to another embodiment, candidate regions Set of blocks can be given by whole block coded before current block, and contain the straight of the current any sample of block of expression Connect one or more samples of spatial neighbor sample.A direct left side for the neighbouring any sample for being limited to current block of straight space is neighbouring The neighbouring sample in sample and/or directly top and/or directly right neighbouring sample and/or direct bottom are adjacent to sample.Such as it is aobvious with reference to Fig. 9 B Show another block subdivision.In such cases, candidate block includes four blocks, that is, block A, B, C and D.

In addition, candidate block set can extraly or exclusively include that (it is position current containing one or more samples Any sample same position of block, but be contained in different images, that is, encoded/decoding image) block.

Again in addition, a subset of the aforementioned block sets of candidate block set expression, is by direction in space or time side To proximity relations determine.Candidate block subset can notify or lead calculation through fixation, signal.The calculation of leading of candidate block subset is contemplated that The decision that other blocks are done in the image or other images.Lift example, or extremely phase identical as other candidate blocks As the associated block of coding parameter can not include in candidate block set.

It is suitable for a left side for the upper left sample containing current block to the description of embodiment hereinafter and pushes up neighbouring sample there was only two A block is considered as the case where at most possible candidate.

If candidate block set non-empty, referred to as the one of merge_flag label is notified by signal, shows current area Whether block merges with any candidate block.If merge_flag is equal to 0 (for "false"), this block will not be with its candidate regions One of block merges, and usually transmits whole coding parameters.If merge_flag is equal to 1 (being used for "true"), it is applicable in aftermentioned person. If candidate block set contains one and only one block, this candidate block is for merging.Otherwise, candidate block set is proper Containing two blocks.If the Prediction Parameters of this two block be it is identical, these Prediction Parameters be used for current block.Otherwise (two block With different Prediction Parameters), signal notice is referred to as the label of merge_left_flag.If merge_left_flag is In 1 (being used for "true"), then the block of the neighbouring sample position in a left side for the upper left sample position containing current block is from the candidate It is selected in block sets.If merge_left_flag is equal to 0 (for "false" "), selected from the candidate block set (that is, top is neighbouring) block another out.The Prediction Parameters of selected block be for current block.

With regard to merging the number person in outline previous embodiment, shows with reference to Figure 10 and executed by withdrawal device 102 come defeated from entering Enter and extracts the step of pooling information is carried out in the data flow 22 at end 116.

Processing starts from 450, candidate block or sample set of the identification for current sample set or block.It must remember, area The coding parameter of block is with some one-dimensional sequence and in 22 internal transmission of data flow, and accordingly, Figure 10 refers to for access at present Sample set or the block method of fetching pooling information.

It has been observed that identification and step 450 include based on neighbouring aspect and in previous decoded blocks (that is, cause and effect block collection Close) in identification.For example, those adjacent blocks may point to candidate, candidate contain in space or on the time in current block X Neighbouring some of one or more geometry predetermined samples adjacent to sample.In addition, identification step may include two grades, that is, The first order is related to leading to a preliminary candidate block sets just like aforementioned identification based on neighbouring；And simple this waits blocks accordingly for the second level Meet for the coding parameter that, from the direction block of data stream, which has transmitted before step 450 to current area Some relationship of the appropriate subset of the coding parameter of block X.

Secondly, method advances to step 452, determine whether candidate block number is greater than zero at this.If belonging to such situation, Merge_flag is then extracted from data flow in step 454.Extraction step 454 can be related to entropy decoding.Entropy is used in step 454 Decode merge_flag context can based on belonging to such as syntactic element of candidate block set or preliminary candidate block sets, Wherein can be limited to following information to the dependence of syntactic element: whether the block for belonging to concern set undergoes merging.Above and below selected The probability estimation of text adjusted can adapt to.

But if candidate block number is determined as 0 452, Figure 10 method and advances to step 456, the volume of current block herein Code parameter be from bit stream extract, or said second identify instead road in the case where, wherein being swept in withdrawal device 102 with block After retouching the next block progress of sequence (all sequences 350 as shown in Figure 3 C) processing, remaining coding parameter.

With reference to step 454, this method advance step 458 after the extraction of step 454 checks extracted merge_flag Whether prompt the appearance of current block merging or is not present.If not merging, method advances to abovementioned steps 456.Otherwise, Method is with step 460 advance, including checks whether candidate block number is equal to 1.If belonging to such situation, some in candidate block is waited Not necessarily, therefore Figure 10 method advances to step 462 for the transmission of constituency block instruction, accordingly the merging companion setting of current block For unique candidate block, wherein the coding parameter for merging companion's block after step 464 is for adjusting coding parameter or mesh The adjustment or prediction of remaining coding parameter of preceding block.By taking adjustment as an example, the coding parameter that block is omitted at present is duplication merely From merging companion's block.In another case, that is, in the case where prediction, step 464 can be related to further take out from data flow Residual error data is taken, the residual error data of the prediction residual of the omission coding parameter in relation to current block and derived from merging companion's block The combination of the prediction of these residual error datas and these omission coding parameters.

But if candidate block number is determined as advancing to step 466 greater than 1, Figure 10 method in step 460, herein Check the concern part of coding parameter or coding parameter, that is, the part not yet transferred inside the data flow of current block Associated subdivision is consistent with each other.If belonging to such situation, these shared coding parameters are set to merge reference, or candidate Block is to be set as merging companion in step 468, or individual concern coding parameters are used in the adjustment or prediction of step 464.

Combined block is notified it should be noted that merging companion itself and can be signal.In this example, merge the warp of companion Adjustment or predicted gained coding parameter are for step 464.

But otherwise, in the case where coding parameter difference, Figure 10 method advances to step 470, herein additional syntactic element It is to be drawn from data flow, that is, the merge_left_flag.The separated set of context can be used for this label of entropy decoding.For The set of context of entropy decoding merge_left_flag may also comprise a simple context.After step 470, merge_ The candidate block of left_flag instruction is set as merging companion in step 472, and in step 464 for adjusting or predicting.In step After rapid 464, withdrawal device 102 is with the next block of block sequential processes.

It is of course possible to have other instead roads.For example, combination syntactic element can in data flow internal transmission, rather than as before Separated syntactic element merge_flag and merge_left_flag is stated, combination syntactic element signal notifies merging treatment program.This Outside, whether aforementioned merge_left_flag can have identical Prediction Parameters with two candidate blocks in data flow internal transmission It is unrelated, lower the computing overhead for executing Figure 10 processing routine by this.

Such as Fig. 9 B is already described if above, and more than two blocks can be included in candidate block set.In addition, pooling information, That is, signal notifies the information whether a block merges；If so, combined candidate block can pass through one or more grammers Elemental signals notice.One syntactic element can be shown that whether the block is (such as aforementioned with any one of aforementioned candidates block Merge_flag) merge.When being all in the set non-empty of candidate block, just transmission is marked.Where is second syntactic element signal notice One candidate block is used in merging, such as aforementioned merge_left_flag, but is indicated generally at two or more candidate regions Selection between block.The first syntactic element signal is had only to notify just transmit when one of current block candidate block to be merged Two syntactic elements.Second syntactic element more has only when candidate block set contains more than one candidate block again, and/or candidate Any one of block just transmits when having the Prediction Parameters different from any other person of candidate block.Grammer may depend on to It gives how many candidate block and/or how associated with candidate block different Prediction Parameters are.

Signal notice be intended to using which block in candidate block grammer can encoder-side and decoder end simultaneously and/ Or it sets side by side.For example, if identifying three candidate block selections in step 450, grammer is selected as only three selections It is available, such as considers to be used for entropy coding in step 470.In other words, syntactic element is to be selected so that its symbols alphabet Only there are multiple elements same as the selection of existing candidate block.All other select probability to be contemplated that be zero, entropy volume Code/decoding can adjust simultaneously in encoder and decoder.

In addition, remembering as above with regard to step 464, referred to as the Prediction Parameters of merging method result can be indicated and current block Associated Prediction Parameters full set, or can indicate a subset of these Prediction Parameters, such as needle is used for multiple hypothesis The Prediction Parameters of one hypothesis of the block of prediction.

It has been observed that context modeling can be used to carry out entropy coding for the syntactic element in relation to pooling information.Syntactic element can be by Aforementioned merge_flag and merge_left_flag composition (or similar syntactic element).In a specific example, three contexts One of model or context can be used for coding/decoding merge_flag in step 454.Used context model index Merge_flag_ctx can lead as follows calculating: if candidate block set contains two elements, the value of merge_flag_ctx is Equal to the sum of the value of merge_flag of two candidate blocks.But if candidate block set contains an element, The value of merge_flag_ctx is equal to twice of the merge_flag value of this candidate block.It is each because of neighbor candidate block Merge_flag can be 1 or 0, and for merge_flag, there are three contexts can be used.Merge_left_flag can only use list One probability model coding.

But according to alternate embodiment, different context models can be used.For example, nonbinary syntactic element can map to two Hex notation sequence (so-called storehouse).The context model of several syntactic elements or syntactic element storehouse that define pooling information can base Calculating is led in the syntactic element or candidate block number of the adjacent block transmitted or other measured values, while other grammers Element or syntactic element storehouse can be to fix context model coding.

In relation to the description that block above merges, notably candidate block set can also be described in aforementioned any embodiment Same way leads calculating and has following amendment: candidate block is limited to the block using motion compensated prediction or interpretation.Have those only Element can be the element of candidate block set.The signal of pooling information notifies and context modeling can carry out in the foregoing manner.

In terms of turning to the combination with reference to aforementioned multiway tree subdivision embodiment and the merging that will now be described, if the image is to pass through Use the square blocks that the sizes such as not are divided into based on quaternary tree sub-structure, such as merge_flag and merge_left_ Flag or other shows that combined syntactic element can be interleaved with the Prediction Parameters transmitted for each leaf node of quad-tree structure. Such as Fig. 9 A is considered once again.Fig. 9 A shows the example that an image is subdivided into variable-size prediction block based on quaternary tree.Maximum ruler Very little upper two blocks are so-called tree block, that is, it is the prediction block of maximum possible size.Other blocks in the figure It is the subdivision obtained as its corresponding tree block.Block is denoted as " X " at present.Whole shade blocks are before current block Coding/decoding, therefore it forms cause and effect block sets.Calculation description such as is being led for one of embodiment candidate block set, only Candidate block set can just be become by having direct (that is, top or left) the neighbouring sample of the upper left sample position containing current block Member.In this way, block merges block " A " or block " B " at present.If merge_flag is equal to zero (for "false"), at present area Block " X " does not merge any one of two blocks.If block " A " and " B " have identical Prediction Parameters, need not distinguish, Reason, which is to merge with any one of two blocks, will lead to identical result.Therefore, in such cases, merge_ is not transmitted left_flag.Otherwise, if block " A " and " B " have different Prediction Parameters, merge_left_flag=1 (being used for "true") will Merge block " X " and " B ", and merge_left_flag is equal to 0 (being used for "false") and will merge block " X " and " A ".It is another compared with In good embodiment, additional neighbouring (transmission) block indicates combined bidirectional.

Fig. 9 B shows another example.Current block " X " and left adjacent block " B " are tree block herein, that is, it has most Block size is allowed greatly.The size on top adjacent block " A " is to set a quarter of block size.Belong to the member of cause and effect block sets The block of element adds shade.Notice that block " X " is only closed with two blocks " A " or " B " at present according to one of preferred embodiment And without merging with any other top adjacent block.In a further preferred embodiment, additional adjacent (transmission) block table Show combined bidirectional.

It continues to describe before this aspect of different sample arrays in relation to how to handle image according to the embodiment of the present application, palpus Pay attention to that related multiway tree subdivision and on the one hand signal notice is discussed above and on the other hand merges these obvious aspects of aspect and can mention The advantages of for inquiring into independently of one another.In other words, as noted above, multiway tree subdivision has specific advantages with combined combination, but Advantage merges feature e.g. to be finely divided specific implementation by subdivider 30 and 104a also from alternative herein, and It is not based on quaternary tree or multiway tree subdivision, is the macro zone block subdivision for becoming smaller subregion with these macro zone block rule subregions instead It is corresponding.On the other hand, multiway tree subdivision is thin together with the combination of the maximal tree block size transmission inside bit stream and multiway tree Dividing the use of the corresponding coding parameter together with depth-first traversal sequence transmission block has whether simultaneously with use merges spy The independent unrelated advantage of sign.Generally, intuitively consider that when sample array Encoding syntax be to not only allow for one block of subdivision, together When also allowing to merge the mode of block obtained after two or more subdivisions and extending, code efficiency improves, it may be appreciated that merges The advantages of.As a result, obtaining one group of block, it is with identical Prediction Parameters coding.The Prediction Parameters of this group of block need only encode one It is secondary.In addition, the merging in relation to sample set, once again it is understood that the sample set considered can be rectangle block or square blocks, In the case of this kind, merge the set of sample set expression rectangle block and/or square blocks.In addition, the sample set considered The set in arbitrarily shaped image area is indicated for arbitrarily shaped image area, and merging sample set.

The focus being hereinafter described is the different samples of one image when each image has more than one sample array The processing of array, hereinafter in secondary description several aspects of outline be with the independent unrelated advantage of used subdivision type, that is, It is whether independent unrelated based on multiway tree subdivision with subdivision and with it is whether independent unrelated using merging.Start from the related image of description not Before specific embodiment with the processing of sample array, the main themes of the present embodiment are each image difference sample numbers of A brief introduction The process field of group.

Focus discussed hereinafter concentrates in image or Video coding application purpose, in the different sample arrays of an image Coding parameter between block, the mode of adaptive forecasting coding parameter is applied to for example scheme between the different sample arrays of a special image The encoder and decoder or other images or video coding environment of 1 and Fig. 2.It has been observed that sample array indicates and different colours The associated sample array of component, or with additional information (such as transparence information or depth map image) associated image.With The relevant sample array of the chrominance component of image is also referred to as used as color plane.The technology being hereinafter described also referred to as is used as and adopts across plane With/prediction, the sample array area of an image can be directed to by this used in image and video encoder and decoder based on block The processing sequence of block is random order.

Image and video encoder typical case are designed for coding colour image (static image or video sequence image).It is color Chromatic graph picture includes multiple color planes, indicates the sample array of different chrominance components.Often, it is bright to be encoded as one for color image The sample array set of plane and two colorimetric planes composition is spent, the latter shows color difference components herein.In several application fields, Common coded samples array set is made of expression primary colors red, green and blue sample array three color planes.This Outside, in order to improve colored expression kenel, color image can be made of more than three color plane.In addition, an image can with show The aid sample array of the additional information of the image is associated.For example, these aid sample arrays can be and show correlated color The sample array (suitable for showing to show purpose) of the transparency of sample, or to show that the sample array of depth map (is suitble to use It is presented multiple sights, such as 3D display).

In conventional image and video encoding standard (such as H.264), color plane encodes typically together, thus specific coding Parameter (such as macro zone block and secondary macro zone block prediction mode, benchmark index and motion vector) is divided for all colored of a block Amount.Luminance plane is contemplated that as a color plane, shows specific coding parameter in bit stream；And colorimetric plane can be considered secondary Plane, corresponding coding parameter are estimated from a luminance plane.Each luma blocks are and indicate same area in the image Two colourity blocks are associated.According to used chroma, chroma sample array is than the brightness for being directed to a block Sample array is smaller.For each macro zone block as composed by a luminance component and two chromatic components, using divide into it is smaller Type block (if macro zone block is through segmenting).For each block as composed by a luma samples block and two chroma sample blocks (can be macro zone block itself or be macro zone block a sub-block), using identical Prediction Parameters set, such as benchmark index, movement ginseng Number and once in a while interior-prediction mode.It, can (such as in 4:4:4 profile H.264) in the contoured of convention video coding standard The different color planes of one image of absolute coding.In the configuration, it can be separated for the chrominance component or sub-block of a macro zone block Select macro zone block subregion, prediction mode, benchmark index and kinematic parameter.According to conventional coding standard, whole color planes are to make It is encoded together with identical specific coding parameter (such as subdivided information and Prediction Parameters) set, or for whole color planes is respective It is completely independent coding.

If color plane is to encode together, a set of subdivision and Prediction Parameters must be used for all colored point of a block Amount.So ensure that side information remains a small amount of, but compared to absolute coding, may cause the reduction of code efficiency, reason is To different color components using different blocks decomposition and Prediction Parameters, the attenuating of rate-distortion cost may cause.For example, Different motion vector or reference frame are used for chromatic component, the residual signals energy of chromatic component can be substantially reduced, and increase it Overall coding efficiency.If color plane is absolute coding, coding parameter (such as block partition, benchmark index and kinematic parameter) can It is separately selected for each chrominance component to be directed to each chrominance component and optimize code efficiency.But chrominance component can not be used Between redundancy.The polynary transmission of specific coding parameter leads to the increase (compared with assembly coding) of side information rate really, such Overall coding efficiency may be adversely affected by improving side information rate.In addition, existing video encoding standard (such as H.264 in), supporting aid sample array to be limited to aid sample array is encoded using the coding parameter sets of itself.

In this way, to so far, in the whole embodiments, the plane of delineation can be more such as aforementioned processing, but as previously discussed The overall coding efficiency (may be related from different color planes and/or aid sample array) of the coding of a sample array may increase Whether height can (for example) determine whole sample arrays for a block all with identical coding parameter volume based on block benchmark at this time Code, or whether use different coding parameter.The basic conception across planar prediction for example allows to make this based on block hereinafter Kind adaptive decision.Such as based on rate distortion criterion, all or part of sample number for a particular block is may be selected in encoder Whether group is encoded using identical coding parameter, or whether uses different coding parameter coding for different sample arrays.Pass through needle To a specific sample array block, signal notice never with the encoded co-location block of sample array whether the specific volume of inference Code parameter, also may achieve this selection.Sample may be also referred to as used as the image configurations difference sample array in group Array group or plane group.One or more sample arrays of each plane group containing an image.Then, in a plane The sample array block of group internal shares identical selected coding parameter, such as subdivided information, prediction mode and residual coding mould Formula；And other coding parameters (such as transformation coefficient level) are separately passed for each sample array of the plane group internal It is defeated.One plane group is to be encoded to a secondary flat group, that is, do not estimate from other plane groups or predictive coding parameter. For each block of two secondary flat groups, whether a new set of the selected coding parameter of adaptability selection transmits or selected coding Whether parameter estimates or predicts from a plane set or another secondary plane set.It is for coding parameter selected by particular block No is presumption or the decision predicted is included in the bit stream.Folding between across planar prediction permission side information rate and forecast quality Inner feelings has more high-freedom degree compared to the coding of the present image as composed by multiple sample arrays.Advantage is relative to by multiple samples The coding of normal image composed by this array, code efficiency improvement.

The expansible image of use/prediction or video encoder in plane, the image or Video coding of such as previous embodiment Device, so that can be for a block or color catalog array and/or aid sample for a color catalog array or aid sample array Whether one set of array, adaptability select selected coding parameter sets from sample arrays other in the same image Coded co-location block presumption or prediction, or whether be encoded separately for the selected coding parameter sets of the block And not with reference to the common positioning block of other sample arrays inside the same image.Selected coding parameter sets whether needle The decision of one sample array block or the presumption of multiple sample array blocks or prediction can be included in the bit stream.It is related to an image The different sample arrays of connection need not have same size.

It has been observed that (sample array can indicate chrominance component and/or aid sample to sample array associated with an image Array) two or multiple so-called plane groups can be arranged in, each plane group is made of one or more sample arrays herein.Contain In the sample array need not have equal sizes of specific plane group.Notice that such plane group that is arranged in includes each sample array The case where being encoded separately.

More clearly say it, according to an embodiment, for each block of a plane group, whether adaptability selects encoded block Show how a block estimates or predict from the coded co-location block of the Different Plane group for same image, Or whether these coding parameters are directed to the block and are encoded separately.Show that the coding parameter how a block is predicted includes following coding One or more of parameter: Block predictions mode show for the block using it is any prediction (interior prediction, using single fortune Moving vector and reference picture across-prediction, using two motion vectors and reference picture across-prediction, using high-order across-it is pre- Survey, that is, non-translational motion model and single reference picture, using multiple motion models and reference picture across-prediction), it is interior- Prediction mode indicates how in generation that-prediction signal, an identifier show how many prediction signals combinations to generate for the block Final prediction signal, which (which) reference picture benchmark index show for motion compensated prediction, and kinematic parameter is (such as Motion vector or affine motion parameters) show how prediction signal uses reference picture and generate, an identifier shows with reference to figure As how to be filtered to generate motion compensated prediction signal.Pay attention to generally, a block can be only sub with the one of the coding parameter Collection is associated.For example, if Block predictions mode shows that a block is interior-prediction, the coding parameter of a block can be additional Ground includes interior-prediction mode, but does not show coding parameter, such as indicates how to generate benchmark index and fortune across-prediction signal Dynamic parameter；If or Block predictions mode shows that associated coding parameter can extraly include benchmark index and fortune across-prediction Dynamic parameter, but do not show interior-prediction mode.

One of two or more plane groups can be used as a secondary flat group in the bit stream in-line coding or instruction.Needle To whole blocks of this secondary flat group, show that the coding parameter how prediction signal generates does not refer to same figure through transmission Other plane groups of picture.Remaining plane group is encoded as two secondary flat groups.For each block of two secondary flat groups, pass One or more defeated syntactic elements, syntactic element signal notice show whether the coding parameter how block is predicted puts down from other The co-location block of face group estimates or prediction, or a new set of these coding parameters whether is transmitted for the block.One One of a or multiple syntactic elements can be referred to as across planar prediction label or across planar prediction parameter.If syntactic element signal Notice does not estimate or predicts corresponding coding parameter, then being newly integrated into the bit stream for the corresponding coding parameter of the block passes It is defeated.If syntactic element signal notifies that corresponding coding parameter is to be estimated or predicted, determine in so-called reference planes group In co-location block.It can be assembled in many ways for the specified of reference planes group of the block.In one embodiment, It is specified for each two secondary flat group with particular reference to group；Such specify can be fixation, or can be (all in high-order syntactic structure Such as parameter sets, access unit header, image header or piece header) in signal notice.

In a second embodiment, the specified of reference planes group is compiled in bit stream in-line coding, and by being directed to a block One or more syntactic element signals of code notify to show whether selected coding parameter is estimated or predicted or whether divided Begin the compilation of code.

In order to be readily understood by association across planar prediction and the aforementioned possibility of embodiment is shown in detail below show with reference to Figure 11 Meaning ground display image 500 as composed by three sample arrays 502,504 and 506.It is easier to understand to ask, Figure 11 is only shown The sub-portion of sample array 502-506.Sample array be shown as walking back and forth Buddhist its be in alignment with each other in space so that sample array 502- 506 are overlapped along direction 508 each other, and the sample of sample array 502-506 leads to whole sample arrays along 508 protrusion result of direction The sample of 502-506 is spatially to be properly positioned each other.In other words, plane 502 and 506 is in the horizontal direction and Vertical Square It is adjusted each other to expansion and adapts to its spatial resolution and be in alignment with each other.

According to an embodiment, whole sample arrays of an image belong to same a part of space scene, wherein along Vertical Square To and the resolution ratio of horizontal direction can be different between independent sample array 502-506.In addition, in order to for illustrative purposes, Sample array 502 and 504 is considered to belong to a plane group 510, and sample array 506 is considered to belong to another plane group Group 512.In addition, Figure 11 shows example case, it is herein in sample array along the spatial resolution of the trunnion axis of sample array 504 Twice of the resolution ratio of 502 horizontal direction.In addition, sample array 504 is considered as being formed once relative to sample array 502 Array, sample array 502 form the subordinate array for being relevant to an array 504.As explained earlier, in such cases, such as pass through Fig. 1 subdivider 30 determines that sample array 504, which is subdivided into multiple blocks, to be used by subordinate array 502, wherein according to Figure 11 Example, because sample array 502 vertical resolution be an array 504 vertical direction resolution ratio half, each block is Half-and-half be divided equally into two horizontal Tile blocks, when being measured as unit of the sample position in sample array 502, each block by In half-and-half therefore once again as square blocks.

Such as Figure 11 for example, to subdivision selected by sample array 506 be with the subdivision of another sample group 510 not Together.It has been observed that subdivider 30 can dividually or independently select the thin of array of pixels 506 with the subdivision of plane group 510 Point.Certainly, the resolution ratio of sample array 506 can also be different from the resolution ratio of plane 502 and 504 of plane group 510.

Now, when encoding individual sample array 502-506, encoder 10 starts such as code plane group in the foregoing manner Array 504 of group 510.Block shown in Figure 11 may be, for example, aforementioned prediction block.In addition, block can be to define granularity to be used to Define the residual error block or other blocks of certain coding parameters.It is not limited by quaternary tree or multiway tree subdivision across planar prediction, but Quaternary tree or multiway tree subdivision are illustrated in Figure 11.

After the syntactic element transmission of an array 504, encoder 10 can determine to announce that an array 504 is flat for subordinate The reference planes in face 502.Encoder 10 and withdrawal device 30 can carry out signal via bit stream 22 respectively and notify this decision, while from Sample array 504 forms the fact that an array of plane group 510 obvious easily bright wherein relevance, which transfers again can also For a part of bit stream 22.All in all for each block inside sample array 502, times of inserter 18 or encoder 10 What its module can decide whether to inhibit transfer of the coding parameter of this block inside bit stream together with inserter 18, and in place Stream internal signal, which notifies and replaces, to be used in inside an array 504 for the block in bit stream internal signal notice Co-locate the coding parameter substitution of block；Or the coding parameter of co-location block of the decision inside an array 504 is The prediction of the coding parameter of the no current block that will act as sample array 502, and only transfer in bit stream internal needle to the sample Its residual error data of the current block of array 502.In the case where negative decision, coding parameter moves inside data flow as usual Turn.It is that signal notifies in data flow 22 for each block decision.In decoder end, withdrawal device 102 is used for each block Such across planar prediction information obtains the coding parameter of the individual block of sample array 502 accordingly, in other words, if across plane Using/across the plane use/prediction of predictive information prompt, then pass through the coding ginseng of the co-location block of array 504 of inference Number, or in addition from the data flow extract the block residual error data and by this residual error data with it is common derived from array 504 Position the prediction combination of the coding parameter of block；Or as usual with independent unrelated, the mesh of extraction sample array 502 of array 504 The coding parameter of preceding block.

Also it has been observed that identical bits where reference planes are not limited to reside in the block paid close attention at present across planar prediction are flat Face.Therefore as explained earlier, plane group 510 indicates the reference planes group of a secondary flat group or two secondary flat groups 512. In such cases, bit stream may contain a syntactic element, each block instruction aforementioned one of the syntactic element for sample array 506 The coding parameter of the co-location macro zone block of any plane 502 and 504 of secondary flat group or reference planes group 510 is adopted With/predict whether to carry out, in aftermentioned situation, the coding parameter of the current block of sample array 506 is to transmit as usual.

Notably can be identical for the subdivision and/or Prediction Parameters of multiple planes in a plane group internal, Yi Jiyou Only encode that primary (all two secondary flats of a plane group are from the one of approximately the same plane group internal to a plane group in it Secondary flat estimates subdivided information and/or Prediction Parameters), the adaptive forecasting or interference of subdivided information and/or Prediction Parameters be It is carried out between multiple plane groups.

Notably reference planes group can be a secondary flat group or two secondary flat groups.

Co-location between the Different Plane block of a plane group internal is readily understood by the thin of a sample array 504 Divide is to be used by subordinate sample array 502 in space, but aforementioned block segments to make used leaf block become square blocks Except.In the case where Different Plane group span plane use/prediction, co-location can define in one way, thus permit Perhaps the more high-freedom degree between the subdivision of these plane groups.The reference planes group is given, is determined in the reference planes group The co-location block in portion.Co-locate block and reference planes group lead calculate can by the similar method being explained later into Row.Specific sample 514 of one of the sample array 506 of selected two secondary flat groups 512 inside current block 516.Mesh The upper left sample of preceding block 516 is also same, as Figure 11 is shown in 514 for illustrative purposes；Or in the sample of current block 516 This any other sample inside block close to current 516 center of block or at present, geometry is through distinct definition.It calculates The position of such selected sample 515 inside the sample array 502 of reference planes group 510 and 504.Sample array 502 and 514 position of sample inside 504 is indicated respectively in Figure 11 in 518 and 520.Which actually uses in reference planes group 510 A plane 502 and 504 can be predetermined or can be notified in bit stream internal signal.It determines in the opposite of reference planes group 510 The sample inside sample array 502 or 504 closest to position 518 and 520 is answered, the block containing this sample is selected as individual Co-location block inside sample array 502 and 504.In case of fig. 11, respectively block 522 and 524.For Other planes determine the possible alternative for co-locating block as being described in detail later.

In one embodiment, show that the coding parameter of the prediction of current block 516 is flat in the difference of identical image 500 It is estimated completely inside face group 510 using the corresponding Prediction Parameters for co-locating block 522/524 and does not transmit additional side Information.Presumption may include replicate merely corresponding coding parameter or coding parameter adjustment adapt to by current plane group 512 with Difference between reference planes group 510 is taken into consideration.An example is lifted, such adjustment adaptation may include that motion parameters correction is added (such as motion vector correction) is used to consider the phase difference between brightness and chroma sample array；Or adjustment adapts to may include modification fortune The precision (such as precision of modification motion vector) of parameter is moved to consider the different resolution of brightness and chroma sample array.In volume In outer embodiment, the coding parameter for showing that the one or more that prediction signal generates is estimated is not directly used in current area Block 516 is used as the prediction of the corresponding coding parameter for current block 516 instead, and these coding ginsengs of block 516 at present Several purifications is transmitted in bit stream 22.Example is lifted, not directly using the kinematic parameter estimated, is shown between kinematic parameter instead The kinematic parameter of deviation poor (such as motion vector is poor) is for current block 516, and the kinematic parameter of presumption is to be encoded in bit stream； In decoder end, the kinematic parameter of actual use is obtained via the kinematic parameter of combination presumption and the kinematic parameter difference of transmission.

In another embodiment, the subdivision of a block, it is such as aforementioned to predict the tree block for being subdivided into prediction block (even if also With the sample block of identical Prediction Parameters set) it is from the Different Plane group according to Fig. 6 A or 6B identical image that is, bit sequence Coded co-location block and adaptively estimate or predict.In one embodiment, in two or more plane groups One of be to be encoded to a secondary flat group.For whole blocks of this secondary flat group, transmits subdivision parameter and do not estimate Other plane groups in same image.Remaining plane group is to be encoded to two secondary flat groups.For two secondary flat groups Block, transmits one or more syntactic elements, and whether signal notifies subdivided information from the co-location blocks of other plane groups Whether presumption or prediction or subdivided information are in the bit stream.One of one or more syntactic elements can be referred to as across plane Predictive marker or across planar prediction parameter.If syntactic element signal notice does not estimate or predicts subdivided information, the block it is thin Point information is not address other plane groups of the same image in the bit stream.If syntactic element signal notifies the subdivision Information is to be estimated or predicted, then determines the co-location block in so-called reference planes group.The reference of the block is flat The configuration of face group can assemble in many ways.In one embodiment, one be assigned to particular reference to plane group it is each secondary Plane group；This is specified to be fixation, or can notify to be used as parameter sets, access unit report in high-order syntactic structure signal Head, image header or piece header.In second embodiment, reference planes it is specified be in bit stream in-line coding, and pass through one Or multiple syntactic element signals notice, these syntactic elements be for a block coding come show subdivided information whether estimated or It predicts or is encoded separately.Reference planes group can be a secondary flat group or other two secondary flats group.Given reference planes group Group determines the co-location block in the reference planes group internal.Co-locating block is corresponding to the identical of current block The reference planes group of image area, or indicate to share the reference planes group internal of the largest portion of the image area and current block Block.Smaller prediction block can be partitioned by co-locating block.

In Additional examples of composition, the subdivided information of block at present, such as subdivision based on quaternary tree according to Fig. 6 A or 6B Information is used in the subdivided information presumption of the co-location block of the Different Plane group of the same image, and does not transmit additional Side information.A particular instance is lifted, if co-locating block is partitioned into 2 or 4 prediction blocks, which also divides Area is at 2 or 4 sub-blocks to predict purpose.As for another particular instance, if co-locating block is partitioned into four sub-districts Block, and one of these sub-blocks are further partitioned into four smaller sub-blocks, then current block is also partitioned into four One of sub-block and these sub-blocks (corresponding to the further analyst of the sub-block for co-locating block) are also partitioned At four smaller sub-blocks.In another embodiment, the subdivided information of presumption is not directly used in current block, instead It is used as the prediction of the practical subdivided information for current block, corresponding purification information is in bit stream.An example is lifted, by altogether The subdivided information estimated with positioning block can be refined further.For be not partitioned into smaller area in common positioning area block The corresponding each sub-block of one sub-block of block, syntactic element can be encoded in bit stream, show whether sub-block puts down at present Face group further decomposes.The transmission of such syntactic element can be using the size of sub-block as condition.Or can in bit stream signal It notifies not to be further partitioned into smaller block in current plane group in the sub-block of the further subregion of reference planes group.

In another embodiment, a block to prediction block subdivision and show the coding parameter two how sub-block to predict Person is the presumption of coded co-location block adaptability or prediction from the Different Plane group for same image.At this In the preferred embodiment of invention, one of two or more plane groups are codings as a secondary flat group.For such Whole blocks of one secondary flat group, subdivided information and Prediction Parameters are not pass with reference to other plane groups of same image It is defeated.Remaining plane group is encoded to two secondary flat groups.For the block of two secondary flat groups, one or more grammer members are transmitted Element, signal notify whether subdivided information and Prediction Parameters estimate or predict from the co-location block of other plane groups；Or Whether subdivided information and Prediction Parameters are in bit stream.One of one or more syntactic elements can be referred to as across planar prediction Label or across planar prediction parameter.If syntactic element signal notice subdivided information and Prediction Parameters be without estimating or predicting, The Prediction Parameters of sub-block obtained by the subdivided information and result of the block are not address its of identical image in the bit stream Its plane group.If syntactic element signal notice is to be estimated or predicted for the subdivided information and Prediction Parameters of the sub-block, Then determine the co-location block in the so-called reference planes group.It can be with for the specified of reference planes group of the block Various ways assemble.In one embodiment, one each two secondary flat group is assigned to particular reference to plane group；Such appointment It can be to fix or can lead in high-order syntactic structure (such as parameter sets, access unit header, image header or piece header) signal Know.In a second embodiment, the appointment of reference planes group be in bit stream in-line coding, and pass through be directed to a block coding one A or multiple syntactic element signal notices are to show whether subdivided information and Prediction Parameters are estimated or predicted or is encoded separately.Ginseng Examining plane group can be a secondary flat group or other two secondary flats group.Given reference planes group determines at this with reference to flat The co-location block of face group internal.The co-location block can be identical with current block in reference planes group The corresponding block of image area, or indicate to share largest portion image area in the reference planes group internal and the current block The block of block.Smaller prediction block can be divided by co-locating block.In the preferred embodiment, for the current block Subdivided information and the Prediction Parameters of gained sub-block be used in the Different Plane group of identical image and co-locate area The Prediction Parameters of the subdivided information of block and corresponding sub-block, and additional side information is not transmitted.As particular instance, if jointly Positioning block is divided into 2 or 4 prediction blocks, then current block is also divided into 2 or 4 sub-blocks to be used to predict purpose, and Prediction Parameters for the sub-block of current block are to lead calculation as aforementioned.Another particular instance is lifted, if co-locating block quilt Four sub-blocks are divided into, and one of these sub-blocks are further partitioned into four smaller sub-blocks, then current area Block is also divided into four sub-blocks, and one of these sub-blocks (sub-district being further divided with common positioning area block The corresponding person of block) be also divided into four smaller sub-blocks, but the Prediction Parameters of whole sub-blocks of not further subregion be as Presumption is described above.In another embodiment, subdivided information is based entirely on the co-location in reference planes group The subdivided information of block estimates, but the Prediction Parameters of sub-block presumption only serve as the pre- of the actual prediction parameter of sub-block It surveys.Deviation between actual prediction parameter and presumption Prediction Parameters is stream encryption in place.In another embodiment, the subdivision letter of presumption Breath is used as the prediction of practical subdivided information for current block, and difference is transmitted in bit stream (described above), but Prediction Parameters Completely through estimating.In another embodiment, both the subdivided information of presumption and the Prediction Parameters of presumption are used as to predict, and Difference and its presumed value between practical subdivided information and Prediction Parameters are in bit stream.

In another embodiment, for a block of a plane group, adaptability selection residual coding mode (is such as converted Type) whether compiled for identical image from the presumption of the encoded co-location block of Different Plane group or prediction or residual error Whether pattern, which is directed to the block, is encoded separately.This embodiment be analogous to the aforementioned adaptability presumption for Prediction Parameters/ The embodiment of prediction.

In another embodiment, a block (such as a prediction block) be subdivided into transform blockiis (that is, application two dimension become The sample block changed) it is to be estimated from the coded co-location block adaptability of the Different Plane group for same image Or prediction.The present embodiment is the similar aforementioned adaptability presumption/prediction embodiment for being subdivided into prediction block.

In another embodiment, a block be subdivided into transform blockiis and gained transform blockiis residual coding mode (such as Alternative types) it is to estimate or predict from the coded co-location block of the Different Plane group for same image.This Embodiment is similar to be subdivided into adaptability presumption/prediction of prediction block and the Prediction Parameters for gained prediction block above Embodiment.

In another embodiment, a block is subdivided into the subdivision letter of prediction block, associated Prediction Parameters, prediction block Breath and be coded common from the Different Plane group for same image for the residual coding mode of the transform blockiis Position the presumption of block adaptability or prediction.The combination of the present embodiment expression previous embodiment.Described in only may also estimating or predict A part in coding parameter.

In this way, afore-mentioned code efficiency can be improved in across plane use/prediction.But it is encoded as obtained by across plane use/prediction Efficiency gain can also be obtained based on other blocks subdivision used in multiway tree subdivision, and with whether implement block and merge nothing It closes.

Just the previous embodiment of across plane adaptation/prediction can be applied to image and video encoder and decoder, by one Color plane and (if present) the aid sample array associated with the image of image be divided into block and by these blocks with Coding parameter is related.For each block, a coding parameter sets can be included in bit stream.For example, these coding parameters can be description How to be predicted and decoded parameter in one block of decoder end.As particular instance, coding parameter can indicate macro zone block or block Prediction mode, subdivided information, interior-prediction mode, the benchmark index for motion compensated prediction, kinematic parameter be such as displaced to Amount, residual coding mode, transformation coefficient etc..There can be different size from the associated different sample arrays of an image.

Next, description above referring to figs. 1 to described in Fig. 8 inside the splitting scheme based on tree for coding parameter reinforce One scheme of signal notice.As for other schemes, that is, merge and across plane use/prediction, reinforce signal notification scheme (after Commonly referred to as to inherit in text) effect and advantage be with previous embodiment independent description, but aftermentioned scheme can be with previous embodiment Any one of or alone or in combination formula combine.

Generally, it is used to encode the improvement encoding scheme of side information (referred to as in the inside of a splitting scheme based on tree To inherit, it is described as follows) allow to obtain following advantages relative to the processing of conventional coding parameter.

In conventional image and Video coding, image or the specific sample array set for image are typically dissected into multiple Block, these blocks are associated with specific coding parameter.Image is usually to be made of multiple sample arrays.In addition, image It can be associated with additional aid sample array, such as can be shown that transparent information or depth map.One image sample array (including auxiliary Sample array) one or more so-called plane groups can be grouped as, each plane group is by one or more samples herein Array composition.The plane group of one image can absolute coding, if or the image be it is associated with more than one plane group, one The plane group of image can predict from other plane groups of same image.Each plane group is typically dissected into multiple blocks. The block (or corresponding block of sample array) be in across image prediction or image-prediction and predicts.Block can have Different size and can be rectangular or rectangle.One image segmentation, which becomes multiple blocks, to be fixed by grammer, or can (at least partly) It is notified in bit stream internal signal.The syntactic element signal notice often transmitted has the subdivision of predefined size block.These grammers Element can be shown that whether a block segments and how to be subdivided into smaller block, and be associated with coding parameter for for example predicting mesh 's.For whole samples (or block of corresponding sample array) of a block, the decoding of associated coding parameter is with predetermined Mode shows.In this example, whole samples of a block are predicted using the identity set of Prediction Parameters, such as benchmark index (reference picture of the identification in coded images set), kinematic parameter (show between a reference picture and current image The measurement of one block motion), the parameter, the interior-prediction mode that show interpolation filter etc..Kinematic parameter can have horizontal component And the motion vector of vertical component indicates, or with higher order kinematic parameter (affine motion parameters composed by such as six ingredients) It indicates.It is related to single block for may have more than particular prediction parameter sets (such as benchmark index and kinematic parameter) Connection.In the case of this kind, for each set of these particular prediction parameters, generate that (or sample array is corresponding for the block Block) single medium range forecast signal, final prediction signal be by include overlapping medium range forecast signal combination establish.Relatively It answers weighting parameters and also constant offset (adding to the weighted sum) image or a reference picture or a reference picture can may be directed to Gather and fixes；Or it can include in the Prediction Parameters set for corresponding block.Original block (or corresponding sample array Block) and its prediction signal between difference also referred to as be used as residual signals usually it is transformed and quantization.Often two-dimensional transform is applied to Residual signals (or corresponding sample array of residual error block).For transition coding, particular prediction parameter sets have been used Block (or corresponding sample array block) can be in the segmentation that takes a step forward of the application transformation.Transform blockiis can be equal to or be less than and uses In the block of prediction.May also a transform blockiis include in block for prediction more than one.Different transform blockiis can have There is different size, transform blockiis can indicate square blocks or rectangle block.After the conversion, gained transformation coefficient is quantified, and obtains Obtain so-called transformation coefficient level.Transformation coefficient level and Prediction Parameters and (if present) subdivided information are to be entropy encoded.

According to several images and video encoding standard, provided by grammer an image (or a plane group) is subdivided into it is more A possibility that a block, is extremely limited.Usually only indicate whether that (and possibly how) can segment with the block of predefined size At smaller block.It for example, is 16 × 16 according to largest block H.264.16 × 16 block is also referred to as used as Hong Qu Block becomes macro zone block in each picture portion of first step.For each 16 × 16 macro zone block, whether signal notice is encoded into 16 × 16 blocks, or it is encoded into two 16 × 8 blocks or two 8 × 16 blocks or four 8 × 8 blocks.If 16 × 16 blocks Being subdivided into four 8 × 8 blocks, then these 8 × 8 blocks respectively can be encoded into 8 × 8 blocks or two 8 × 4 blocks, or Two 4 × 8 blocks or four 4 × 4 blocks.Show in current image and video encoding standard subdivision become multiple blocks can The minimal set of energy property has the advantages that be maintained small with the side information rate for signaling to subdivided information, but has transmission needle The shortcomings that bit rate required by Prediction Parameters to block becomes larger, as being described in detail later.With the side for signaling to Prediction Parameters Side information rate usually indicates the total bit rate of the significant quantity of a block really.And when this side information is reduced, such as it can be used larger When type block size is reached, code efficiency can be improved.The practical image or image of one video sequence are appointing by tool special properties Shape objects of anticipating composition.An example is lifted, these objects or object part are with unique texture or unique motion for its feature.Usually Identical Prediction Parameters set can be applied to such object or object part.But object bounds are not overlapped large-scale prediction block usually Possible block border (such as in 16 × 16 macro zone blocks H.264).Encoder usually determines to lead to smallest particular rate-distortion The subdivision of cost measure (in limited possibility set).For arbitrary shape of object, great quantity of small block so can lead to. And because these small-sized blocks are associated with the Prediction Parameters set that must be transmitted, therefore side information rate becomes the significant of total bit rate Part.But because several small-sized blocks still indicate the area of same object or object part, the Prediction Parameters of multiple gained blocks are It is identical or very much like.Instinctively, when grammer is to expand in one way, one block of subdivision is not only allowed for, while also thin Code efficiency can be improved when sharing coding parameter between the multiple blocks of gained after point.In the subdivision based on tree, by by be based on Class type relationship prescribed coding parameter of tree or part thereof gives one or more parent nodes, may achieve for a given block collection The coding parameter of conjunction is shared.As a result, shared parameter or part thereof can be used to reduce for gained block signature notice after subdivision Side information needed for coding parameter actual selection.Reducing can be reached by deleting the signal notice of the parameter of subsequent block, Or it can be reached by the prediction model of parameter and/or the shared parameter of context model for being directed to subsequent block.

The basic conception of aftermentioned succession scheme is by coming by the social strata relation shared information based on tree along the grade blocks Bit rate needed for reducing transmission encoded information.Shared information is to notify (in addition to subdivided information) in bit stream internal signal.It inherits The advantages of scheme is to cause code efficiency to increase by lowering side information rate result for coding parameter.

In order to reduce side information rate, according to aftermentioned embodiment, for the individual encoding parameters of specific sample set, that is, Simple bonding pad can indicate the rectangle block or square blocks or arbitrary shape area or any other sample set of multiway tree subdivision Conjunction is to notify in an efficient way in data flow internal signal.Aftermentioned succession scheme allows coding parameter need not be in sample set Each sample set clearly include in bit stream.Coding parameter can indicate Prediction Parameters, show that corresponding sample set is to make With encoded sample predictions.Multinomial possible and example has been also applied for herein really as explained earlier.It is such as already described above, and just As being described in detail later, related following several schemes, the sample array of an image is based on tree-shaped being divided into multiple sample sets and can leading to It crosses grammer to fix, or can be notified by the corresponding subdivided information signal inside bit stream.It has been observed that being directed to the volume of sample set The sequential delivery that code parameter can define in advance, this is sequentially given by grammer.

According to the scheme of succession, the withdrawal device 102 of decoder or decoder is configured as leading calculating related in a specific way The information of the coding parameter of not simple bonding pad or sample set.In specific words, coding parameter or part of it are (such as pre- Survey the parameter of purpose) it is to be shared between each block along the given splitting scheme based on tree, along the shared group of the tree Group is determined by encoder or inserter 18 respectively.In a specific embodiment, the whole of the one of cut tree given internal node The shared of the coding parameter of child node is using the shared label instruction of specific binary values.As for instead road, for each node The purification of coding parameter can be transmitted, so that the accumulation purification of parameter can be applied to along the class type relationship of the block based on tree Whole sample sets of the block of one given leaf node.In another embodiment, it is passed along the block social strata relation based on tree A part of the defeated coding parameter for internal node can be used for a given leaf node for the block coding parameter or its The context adaptability entropy coding of a part and decoding.

Figure 12 A and Figure 12 B display use the basic conception of the succession of the special case based on Quadtree Partition.But such as above for several times Instruction, other multiway tree subdivision schemes can also be used.The tree is displayed at Figure 12 A, and the tree phase with Figure 12 A Corresponding space segmentation is displayed at Figure 12 B.Segmentation shown in it is similar with regard to shown in Fig. 3 A to 3C.Outline says it, succession side The node that case will allow side information to be assigned to the different n omicronn-leaf layers inside the tree construction.It is assigned to according to side information at this The node of the different layers of tree, such as internal node of the tree of Figure 12 A or its root node, the tree class type of the block shown in Figure 12 B It may achieve different degrees of shared side information in relationship.For example, if whole leaf nodes in layer 4 are determined, Figure 12 A the case where It is lower all with identical parent node, virtually share side information, it means that in Figure 12 B with 156a to 156d instruction most Small-sized block shares this side information, and no longer needs to believe for all these small-sized block 156a to 156d complete transmission side Breath, that is, transmit four times, but be so maintained the option of encoder.However, it is also possible to determine the 1 (layer of class type level of Figure 12 A 2) the whole district, that is, tree 150 upper right corner of block a quarter part include sub-block 154a, 154b and 154d and it is aforementioned again Smaller sub-block 156a to 156d is used as wherein to share the area of coding parameter.In this way, increasing shared side information Area.It is next to increase whole sub-blocks that level be aggregation layer 1, that is, sub-block 152a, 152c and 152d and aforementioned smaller Block.In other words, in such cases, entirely tree block has the side information for being assigned to the block, the whole of this tree block 150 Sub-block shares side information.

It inherits in explanation later, following annotation is for describing embodiment:

A. the reconstruction sample of current leaf node: r

B. the reconstruction sample of adjacent leaf: r '

C. the fallout predictor of current leaf node: p

D. the residual error of current leaf node: Res

E. the reconstructive residual error of current leaf node: RecRes

F. calibration and inverse transformation: SIT

G. label: f is shared

As the first example of succession ,-prediction signal notice can be described in internal node.It more accurately says it, describes How signal notice in the internal node division based on tree block-prediction mode is to predict purpose.By from root section Point to leaf node come traverse tree, internal node (including root node) can transmitting portions side information, which will be corresponding by its Child node utilize.More clearly say it, shared label f is to send for internal node and have following meaning:

If f has numerical value 1 ("true"), whole child nodes of the given internal node share in identical-prediction mode. Other than the shared label f with numerical value 1, also signal notifies interior-predictive mode parameters to come by whole child nodes to internal node It uses.As a result, all then child node does not carry any prediction mode information and any shared label.In order to rebuild whole phases Leaf node is closed, decoder is from corresponding internal node using interior-prediction mode.

If f has numerical value 0 ("false"), the child node of corresponding internal node does not share in identical-prediction mode, The each child node for belonging to internal node carries one and separately shares label.

Figure 12 C show it is aforementioned in internal node-prediction signal notice.It transmits in the internal node of layer 1 by interior-prediction The given shared label of pattern information and side information, and child node does not carry any side information.

Example is inherited as second, can be described across-prediction purification.It more clearly says it, how describes in the block based on tree Divide internal schema, side information of the signal notice across-prediction mode is for for example passing through the given movement ginseng of motion vector Several purification purposes.By the way that from root node, by traversing tree to leaf node, internal node (including root node) can transmitting portions Side information, the information will be refined by its corresponding child node.More clearly say it, shared label f is sent out for internal node It send and there is following meaning:

If f has numerical value 1 ("true"), whole child nodes of the given internal node share same motion vector ginseng It examines.Other than the shared label f with numerical value 1, also signal notifies motion vector and benchmark index to internal node.As a result, all Subsequent child node does not carry additional shared label, can carry the purification of this motion vector references inherited instead.For whole The reconstruction of related leaf node, decoder give leaf node motion vector are added to refine to belonging to its corresponding internal parent node at this The motion vector references value of the succession of numerical value 1 with shared label f.So indicate the motion vector essence in a given leaf node It is made as being intended to be applied to the actual motion vector internal parent node corresponding thereto of leaf node since then for motion compensated prediction Difference between motion vector references value.

If f has numerical value 0 ("false"), the child node of corresponding internal node is not shared necessarily identical across-prediction Mode, and the essence for using the kinematic parameter for deriving from corresponding internal node and carrying out kinematic parameter is not through in the child node System, each child node for belonging to internal node carry one and separately share label.

Figure 12 D shows the purification of aforementioned movement parameter.The internal node of layer 1 is the shared label of transmitting and side information.Belong to The child node of leaf node only carries kinematic parameter purification, such as the inside child node of layer 2 does not carry side information.

Referring now to Figure 13.Figure 13 flow for displaying figure illustrates decoder (decoder of such as Fig. 2) and is used for from data (it is subdivided into different size of one message sample array of stream reconstruction representation space example information signal by multiway tree Leaf area) operation mode.It has been observed that each leaf area has a series of relative hierarchical levels selected from multiway tree subdivision In a hierarchical level.For example, whole block shown in Figure 12 B is all leaf area.Leaf area 156c be, for example, and hierarchical level 4 (or level 3) is associated.Each leaf area has coding parameter associated there.The example of these coding parameters have been described above as Before.For each leaf area, coding parameter is indicated with an individual grammar element set.Each syntactic element is selected from grammer member Individual syntax element type in plain type set.Each syntax element type be, for example, prediction mode, motion vector component, The instruction etc. of interior-prediction mode.According to Figure 13, decoder carries out the following steps.

In step 550, inherited information is to be drawn from data flow.In the case of figure 2, withdrawal device 102 is responsible for step 550.Inherited information instruction inherits whether be used for current message sample array.Be hereinafter described will display inherited information have it is several can Can, such as shared signal notice for marking f and polynary tree construction to be divided into a second part and two second part.

Message sample array has constituted a subdivision of an image, such as tree block, such as the tree block of Figure 12 B 150.Whether such inherited information instruction uses succession for certain tree block 150.Such inherited information for example can be for all It predicts the tree block of subdivision and is inserted into data flow.

In addition, inherited information instruction is composed by leaf area set and corresponds to multiway tree if instruction is using inheriting At least one of the message sample array of one hierarchical level of the hierarchical level sequence of subdivision inherits area and is less than this Gather associated each hierarchical level in leaf area.In other words, inherited information instruction is for current sample number group (such as tree block 150) whether succession is used.If it is, indicating that volume is shared at least one leaf area inherited inside area or sub-district of this tree block 150 Code parameter.In this way, inheriting area may not be leaf area.In the example of Figure 12 B, inherit area (for example) can for by sub-block 156a extremely 156b is formed by area.Can be bigger in addition, inheriting area, also additionally cover sub-block 154a, b and d, and even in addition, inheriting area It can be tree block 150 itself, whole leaf blocks share coding parameter associated with the succession area.

But it should be noted that more than one succession area can be defined inside a sample array or tree block 150.For example, it is assumed that Lower-left sub-block 152c is also divided into smaller block.In such cases, sub-block 152c can form a succession area.

In step 552, inherited information is checked, if use succession.If it is, the processing routine of Figure 13 advances to step 554, herein in relation to each across succession area, the succession subset of at least one syntactic element including predetermined syntax element type is It is extracted from data flow.In later step 556, then it is opposite to inherit subset is copied into inside the syntactic element set one for this It answers syntactic element to inherit subset, or is used as the prediction of the succession subset, which indicates that at least one inherits the leaf in area Area's collection is combined into associated coding parameter.In other words, for each succession area indicated inside the inherited information, data flow Succession subset including syntactic element.Again in other words, succession is related at least some syntax element type that can be used for inheriting Or syntactic element classification.For example, prediction mode or succession can be undergone across-prediction mode or interior-prediction mode syntactic element. Such as inherit area inside the data flow contained by succession subset may include across-prediction mode syntactic element.Inherit subset It also include additional syntactic element, syntax element type is to depend on aforementioned fixed grammer member associated with the succession scheme The value of plain type.For example, in the case where being to inherit the fixed component of subset across-prediction mode, motion compensation is defined Syntactic element (such as motion vector component) can be included or can not included in the succession subset by grammer.For example, false If the upper right a quarter (that is, sub-block 152b) of tree block 150 is to inherit area, then individually can indicate to be used for across-prediction mode This succession area, or be used for together with motion vector and motion vector index across-prediction mode.

Being contained in and inheriting whole syntactic elements of subset is leaf block (that is, the leaf block being copied into inside the succession area The corresponding coding parameter of 154a, b, d and 156a to 156d), or it is used as its prediction.Using prediction, for Individual leaf encrypted communication residual errors.

The transmission for being aforementioned shared label f for the possibility that leaf block 150 transmits inherited information.In step 550, The extraction of inherited information includes aftermentioned in this example.More clearly say it, decoder can be configured to using from lower class type layer Grade to higher-order laminar level hierarchical level sequence, at least one hierarchical level segment with the multiway tree times What, which is inherited, gathers corresponding n omicronn-leaf area, extracts and check the shared label f for deriving from the data flow, about whether individual inherit mark Whether note or shared label show to inherit.For example, the succession set of hierarchical level can pass through the class type layer 1 of Figure 12 A To the formation of layer 3.In this way, for and nonleaf node and be position be any layer 1 to layer 3 sub-tree structure any node, can have There is a shared label associated there inside the data flow.Decoder be with from layer 1 to the sequence of layer 3 (such as with depth Preferential or breadth first traversal sequence) extract these shared labels.Once one of shared label is equal to 1, then decoder is known Dawn is contained in the corresponding leaf block for inheriting area and shares the succession subset, followed by the extraction in step 554.For current node Child node, it is no longer necessary to inherit the inspection of label.In other words, the succession label of these child nodes does not pass inside data flow Defeated, reason is that the succession subset that these obvious node areas already belong to wherein syntactic element is the shared succession area.

Shared label f can notify the position of quaternary tree subdivision to intersect with aforementioned signal.E.g., including subdivided mark and shared mark The intersection bit sequence of both notes can are as follows:

10001101(0000)000,

It is identical subdivided information shown in Fig. 6 A, and the shared label to intersperse with two, which is drawn lines strong by lower section It transfers to indicate to share coding parameter in the sub-block of tree 150 lower-left a quarter of block in Fig. 3 C.

Definition instruction inherit area inherited information another way be using two subdivisions defined each other with slave mode, Explanation as distinguished reference prediction subdivision and residual error subdivision above.Outline says it, and the leaf block once segmented can form the succession The succession subset in area, the fixed wherein syntactic element of the succession regional boundary is shared area；And inside these succession areas of subordinate segment definition Block, the succession subset for the block syntactic element is replicated or is used as predicting.

For example, it is contemplated that extension of the residual tree as prediction tree.It is smaller further to consider that prediction block can be further divided into Type block is used for residual coding purpose.For the corresponding each prediction block of leaf node of quaternary tree related to prediction, for residual The corresponding subdivision of difference coding is determined by one or more subordinate quaternary trees.

In such cases, substitution uses any Prediction Parameters in internal node, and inventor considers that residual tree is with following sides Formula interpretation, residual tree also indicate that the purification of prediction tree is indicated using constant prediction mode (by the corresponding leaf for predicting association tree Node signal notice) but there is the reference sample for passing through purification.Aftermentioned example illustrates such situation.

For example, Figure 14 A and 14B show that the Quadtree Partition for interior-prediction, neighbouring reference sample are for primary subdivision A particular leaf node emphasize that and Figure 14 B shows the residual error quaternary tree subdivision of identical prediction leaf node with the ginseng that has refined Examine sample.Whole sub-block shown in Figure 14 B is directed to the shared phase being contained in inside data flow of individual leaf blocks emphasized in Figure 14 A With across-Prediction Parameters.In this way, Figure 14 A shows the Quadtree Partition example conventionally used for interior-prediction, it is specific shown here as one The reference sample of leaf node.But in the preferred embodiment of inventor, via using reconstructed leaf node in residual tree Neighbouring sample (such as 4 (b) gray shade lines instruction), for leaf node each in residual tree calculate one separate in-it is pre- Survey signal.Then, this prediction signal is added to unusual manner one given residual error leaf of acquisition by the way that residual coding signal will be quantified The reconstruction signal of node.Then this reconstruction signal is used as the reference signal of subsequent Prediction program.It is to be noted that the decoding of prediction Sequence is identical as residual error decoding sequence.

As shown in figure 15, in decoding program, for each residual error leaf node, via use reference sample r ' according to practical Interior-prediction mode (by predicting related quaternary tree leaf node instruction), calculates prediction signal p.

After SIT processing routine,

RecRes=SIT (Res)

It calculates the signal r rebuild and storage is used for next prediction calculation procedure:

R=RecRes+p

Decoding program for prediction is identical as residual error decoding sequence shown in Figure 16.

Each residual error leaf node is to decode as mentioned in the previous paragraph.Reconstruction signal r is stored in buffer, as shown in figure 16.It should In buffer, reference sample r ' will be taken in prediction next time and decoding program.

The combined type for the various aspects that digest is stated before with regard to Fig. 1 to Figure 16 separates after subset describes specific embodiment, The Additional examples of composition of the application will be described, focus be concentrate on it is aforementioned in some terms, but embodiment indicate aforementioned several realities Apply the universalness of example.

In specific words, many aspects of previous embodiment main combination the application of the framework in relation to Fig. 1 and Fig. 2, can also Excellently using in other application purposes or other coding fields.It such as often addresses above, such as multiway tree subdivision can non-economic cooperation And it and/or is used without across plane use/prediction and/or without succession.For example, the transmission of maximum block size, depth The use of first traversing order, according to the hierarchical level of individual subdivided marks context adapt to and inside bit stream it is maximum Side information bit rate is saved in the transmission of hierarchical level, and all these aspects are all excellent but independent of one another.When consideration is across plane It is also such when Utilization plan.The advantages of across plane utilization, is independent for the exact way that is subdivided into simple bonding pad with an image Unrelated and its advantage is independent unrelated with the use of Merge Scenarios and/or succession.It is equally applicable to be related to merging and inherit excellent Point.

Therefore, the embodiment outlined below summarises the embodiment above-mentioned in terms of across plane use/prediction.By The summary to above-described embodiment is represented in the following examples, many above-mentioned details can be considered can be combined in described below Embodiment.

Figure 17 shows represent the data flow of different spaces sample intelligence component in the plane of scene image for decoding The module of decoder, each plane include message sample array.Decoder can correspond to decoder shown in Fig. 2.Specifically, Module 700 is responsible for through payload as processing such as residual error data or spectral decomposition data, Lai Jinhang message sample The reconstruction of each array 502 to 506, wherein the residual data or spectral decomposition data are relevant to each message sample array 502 to 506 be subdivided into a manner of as defined in the coding parameter of such as Prediction Parameters relevant to simple join domain it is simple Join domain.For example, the module can be showed by the framed block of institute including frame block 102 in the decoder situation of Fig. 2.So And the decoder of Figure 17 needs not be hybrid decoder.Across and/or interior prediction may be not used.This is equally applicable to conversion and compiles Code, that is, residual error can encode in spatial domain rather than pass through spectral decomposition two-dimensional transformations.

Another module 702 is responsible for will be with the list of the first array (such as array 506 of message sample array) from data flow The relevant coding parameter export of pure join domain comes.Therefore, module 702 defines the task for 700 task execution of module.? In the case where Fig. 2, withdrawal device 102 is considered being responsible for the task of module 702.It should be noted that array 506 can be the second number in itself Group can obtain relative coding parameter by way of across plane use/prediction.

Next module 704 is used to be used for the simple connection of the second array 504 of message sample array from data flow Across the plane interchange information export in region comes.In the case where Fig. 2, withdrawal device 102 is considered being responsible for the task of module 702.

Next module 706 is used for across the plane interchange information based on the simple join domain for the second array, is the Which the subset appropriate of the simple bonding pad of each of two arrays or simple bonding pad determines in next module 708 and 710 Activation.In the case where Fig. 2, withdrawal device 102 cooperates to carry out the task of module 706 with subdivider 104.Withdrawal device 102 into Row is practical while extract, and subdivider control traverses the sequence of simple bonding pad, that is, any part across plane interchange information about Any part of simple bonding pad.In embodiment in more detail above, each simple bonding pad is individually defined mutual across plane Information is changed, about whether carrying out across plane use/prediction.However, this is not problem.If the appropriate son in simple bonding pad It is determined in the unit of collection, that is advantageous.For example, across plane interchange information can define, one or more is bigger to be connected merely Area is met, each of the bonding pad includes one or more adjacent simple bonding pads, for every in these bigger areas One, execute once across plane use/prediction.

Module 708 is used for: in the case where Fig. 2, passing through the responsible export co-location relationship that cooperates with subdivider 104 Withdrawal device will be used for the coding parameter of the appropriate subset of each simple bonding pad or simple bonding pad of the second array 540 at least Partly export to come from the coding parameter of the corresponding simple bonding pad in part of the first array 506 of the task that is performed；And It is decoded in a manner of as defined in such derived coding parameter suitable with each simple bonding pad of the second array or simple bonding pad When the relevant load data of subset, which transfers to be carried out by other modules in Fig. 2, that is, 106 to 114.

Optionally for module 708, module 710 is used for: in the corresponding simple bonding pad in part for ignoring the first array 506 Coding parameter while, leading-out needle is to each simple bonding pad of the second array 504 or simple bonding pad from data flow The coding parameter of appropriate subset, the task are considered being responsible for by the withdrawal device 102 in Fig. 2；And with the phase derived from data flow Mode as defined in the coding parameter of pass decodes related to the appropriate subset of each simple bonding pad of the second array or simple bonding pad Payload data, relatively, the task is by other modules in Fig. 2, i.e., 106 to 114, be responsible for connecting merely always It manages in area and is carried out under the control of adjacent and co-location relationship subdivider 104.

As above about described in Fig. 1 to Figure 16, the array of message sample do not need to represent video image or still image or Its color component.Sample components can also represent the depth map of such as certain scenes or other two sampling physics numbers of transparent print According to.

As discussed above, the payload data in relation to each of multiple simple bonding pads area may include such as Very big conversion coefficient in the conversion block of the spatial domain of conversion coefficient or the residual error data residual block corresponding with identification in Transformation Domain The very big figure of position.On the whole, for example, load data can be in spatial domain or spectrum domain directly or as its certain The residual error of type prediction spatially describe its relevant simple bonding pad data.In turn, coding parameter is not limited It is made as Prediction Parameters.Coding parameter can be indicated for converting payload data conversion or can define for working as reconstruction information Filter used in the simple bonding pad of individual is rebuild when sample array.

It is segmented it has been observed that the simple bonding pad that the message sample array is subdivided into can be planted derived from a multiway tree, and can be Rectangular or rectangular shape.In addition, being particularly described is only specific embodiment for segmenting the embodiment of a sample array, can also make With other subdivisions.Several possibilities are displayed at Figure 18 A to Figure 18 C.Such as Figure 18 A shows that a sample array 606 is subdivided into that The regular two-dimensional arrangement of this non-overlap tree block 608 adjoined, part tree block are subdivided into according to polynary tree construction With different size of sub-block 610.It has been observed that although quaternary tree subdivision is illustrated in Figure 18 A, in any other number Each parent node segmentation of child node also falls within possibility.Figure 18 B shows an embodiment, accordingly, by directly multiway tree being applied to segment To both full-pixel array 606, a sample array 606 is subdivided into different size of sub-block.In other words, both full-pixel array 606 are regarded as tree block processes.Figure 18 C shows another embodiment.According to this embodiment, sample array is configured to that The regular two-dimensional of the macro zone block of this square adjoined or rectangle configures, and each of these macro zone blocks 612 block is independent Ground is related to partition information, and according to the partition information, macro zone block 612 is left not to be partitioned or be partitioned to be referred to by partition information Show the regular two-dimensional configuration of the block of size.So it is found that whole subdivisions of Figure 18 A to 18C cause the sample array 606 to be subdivided into Simple bonding pad, according to the embodiment of Figure 18 A to 18C, each simple bonding pad is shown with non-overlap.But several instead roads are also Belonging to may.For example, each block can overlap each other.But overlapping can be limited to some underlapped any adjacent block of each block Degree, or each block sample is made to be at most one in the adjacent block being arranged side by side along a predetermined direction and current block Overlapping block.In other words, the latter indicates that left and right adjacent block can be overlapped current block, thus the current block is completely covered, But it does not overlap each other, is similarly suitable for vertical and diagonal adjacent block.Still optionally further such as Figure 17, module 706 In the granularity that is carried out of judgement and across plane use/prediction thus can be plane.Therefore, implemented according to further Example, there are more than two plane, a principal plane and two possible secondary planes, for each possible secondary plane, module 606 determine respectively, and across the plane interchange information in data flow indicates respectively, across plane use/predict whether to be applied to Each plane.If so, can further be manipulated in a manner of above-mentioned simple bonding pad, but, wherein it is mutual across plane Change information only by across plane interchange information indicate those of graphic memory and it is processed.

Although with regard to the several aspects of device contextual declaration, it is apparent that these aspects also illustrate that saying for opposite induction method Bright, a block or a device correspond to a method and step or a method and step feature herein.Similarly, in method and step context Described in aspect also illustrate that corresponding block or project or corresponding intrument feature description.Part or all of method and step can It is executed by (or use) hardware device, such as microprocessor, programmable calculator or electronic circuit.In several embodiments, Certain one in most important method and step or certain multiple can be executed by such device.

Coding/compressed signal of the invention can be stored on digital storage mediums, or can be situated between in the transmission of such as internet Matter (such as wireless transmission medium or wired transmissions medium) transmission.

According to several specific implementations requirement, the embodiment of the present invention can be implemented in hardware or software.Specific implementation can It is executed using digital storage mediums, such as floppy disk, DVD, Blu-ray disc, CD, ROM, PROM, EPROM, EEPROM or flash memory, thereon Storage can electronic type read control signal, be pull together with programmable computer system cooperate (or can pull together to cooperate) thus hold Row individual method.Therefore, digital storage mediums can be computer-readable.

Several embodiments according to the present invention include with can electronic type read control signal data medium, can with can Computer system pull together cooperation thus execute one of method described herein.

In general, it is a kind of computer program product with program code, the journey that the embodiment of the present invention is implementable Sequence code can be operated for executing one of these methods when computer program product is run on computers.Program code Such as it can be stored on machine-readable carrier.

Other embodiments include being stored on machine-readable carrier to be used to execute one of method described herein Computer program.

In other words, therefore, the embodiment of the method for the present invention is a kind of computer program, which has program generation Code is used to execute one of method described herein when the computer program is run on computers.

Therefore another embodiment of the present invention is a kind of data medium (or digital storage mediums or computer-readable Jie Matter) it include the computer program that record is used to execute one of method described herein thereon.

Therefore the another embodiment of the method for the present invention is the computer for indicating to be used to execute one of method described herein The data flow or signal sequence of program.The data flow or signal sequence are for example configured as connecting via data communication, such as through By the Internet transmission.

Another embodiment for example including a kind of processing means, such as computer or can program logic device, be configured as Or adjustment is to execute one of method described herein.

Another embodiment includes the meter for being equipped with the computer program to execute one of method described herein thereon Calculation machine.

In several embodiments, programmable logic device (such as field programmable gate array) can be used to execute described herein Some or all of method function.In several embodiments, field programmable gate array can pull together cooperation with microprocessor to hold One of row method described herein.Outline says it, and this method is preferably executed by any hardware device.

Previous embodiment is only used for the principle illustrated the present invention.It is understood that the modification of the details of configuration described herein and Variation is that skilled artisan is obviously apparent from.Therefore intention is only limited by the range of appended claims rather than by for illustrating Illustrate that the specific detail of embodiments herein is limited.

Claims

1. it is a kind of for decoding the method for representing the data flow of different spaces sample intelligence component of the image of scene in plane, often A plane includes message sample array, which comprises

Each letter is rebuild according to mode as defined in coding parameter by handling payload data relevant to simple bonding pad Cease sample array, wherein each message sample array is subdivided into the simple bonding pad；

From exporting coding relevant to the simple bonding pad of the first array in the message sample array in the data flow Parameter；

Across plane exchange from leading-out needle in the data flow to the simple bonding pad of the second array in the message sample array Information；

According to across plane interchange information described in the simple bonding pad for second array, for second array Each of simple bonding pad or the simple bonding pad appropriate subset, determine:

Leading-out needle is corresponded in the coding parameter of simple bonding pad to each of second array from the part of first array The coding parameter of the appropriate subset of simple bonding pad or the simple bonding pad, and with the so derived coding parameter regulation Mode decode it is related to the appropriate subset of each simple bonding pad of second array or the simple bonding pad The payload data；Or

Volume of the leading-out needle to the simple bonding pad of second array or the appropriate subset of simple bonding pad from the data flow Code parameter, and decoded and second array in a manner of as defined in the coding parameter relevant derived from the data flow The relevant payload data of appropriate subset of simple bonding pad or simple bonding pad,

Wherein, for each simple bonding pad of second array or simple bonding pad appropriate subset payload data Including the residual error data in spatial domain or Transformation Domain, each simple bonding pad or simple bonding pad for second array The coding parameter of appropriate subset represents residual coding mode,

Wherein, described image is the image of video.

2. according to the method described in claim 1, wherein, described image comes from video, and the data flow is encoded as companion With the video for having depth information.

3. according to the method described in claim 1, wherein, the video includes the luma samples array as first array Together with two chroma sample arrays including second array.

4. according to the method described in claim 1, wherein, the plane of described image includes that two colorations being coded separately are put down Face and a luminance plane.

5. it is a kind of for generating the method for representing the data flow of different spaces sample intelligence component of the image of scene in plane, often A plane includes message sample array, which comprises

For each message sample array, determine:

Payload data relevant to the simple bonding pad that each message sample array is subdivided into,

Coding parameter relevant to the simple bonding pad that each message sample array is subdivided into, the coding parameter regulation is for weight Build each message sample array and for the mode that the payload data is handled,

The coding parameter relevant to the simple bonding pad of the first array in the message sample array is inserted into institute State data flow；

Determine across the plane interchange information of the simple bonding pad for the second array in message sample array,

Across the plane interchange information is inserted into the data flow,

Wherein, the determination is executed so that for the suitable of the simple bonding pad of each of second array or the simple bonding pad Work as subset, the related each simple bonding pad for being directed to second array of across the plane interchange information instruction or described connects merely Connect the appropriate subset in area coding parameter whether:

It is exported from the coding parameter that the part of first array corresponds to simple bonding pad, for such derived volume It is appropriate with each simple bonding pad of second array or the simple bonding pad that mode as defined in code parameter is decoded The relevant payload data of subset, or

It is exported from the data flow, for the solution in a manner of as defined in the coding parameter relevant derived from the data flow Code payload data relevant to the appropriate subset of simple bonding pad or simple bonding pad of second array,

Wherein, described image is the image of video.

6. according to the method described in claim 5, wherein, described image comes from video, and the method includes described in coding Data flow has the video with depth information.

7. according to the method described in claim 5, wherein, the video includes the luma samples array as first array Together with two chroma sample arrays including second array.

8. according to the method described in claim 5, wherein, the plane of described image includes that two colorations being coded separately are put down Face and a luminance plane.

9. it is a kind of for decoding the decoder for representing the data flow of different spaces sample intelligence component of the image of scene in plane, Each plane includes message sample array, and the decoder is configured as:

Wherein, described image is the image of video.

10. a kind of for generating the coding for representing the data flow of different spaces sample intelligence component of the image of scene in plane Device, each plane include message sample array, and the encoder is configured as:

For each message sample array, determine:

The coding parameter relevant to the simple bonding pad of the first array in the message sample array is inserted into institute State data flow；And

Across the plane interchange information is inserted into the data flow,

Wherein, the encoder be configured as executing the determination so that for the simple bonding pad of each of second array or The appropriate subset of the simple bonding pad, across the plane interchange information instruction is in relation to for each simple of second array Whether the coding parameter of the appropriate subset of bonding pad or the simple bonding pad:

Wherein, described image is the image of video.

11. the method that one kind flows for storing data, comprising:

The storing data stream on digital storage media, the data flow represent the different spaces sampling letter of the image of plane field scape Component is ceased, each plane includes message sample array, and the data flow includes:

For each message sample array, payload number relevant to the simple bonding pad that each message sample array is subdivided into According to；

Coding parameter relevant to the simple bonding pad of the first array in the message sample array, the coding parameter regulation pair In the mode that the payload data of first array is handled,

For across the plane interchange information of the simple bonding pad of the second array in the message sample array,

For the appropriate subset of the simple bonding pad of each of second array or the simple bonding pad, across the plane exchange The coding of appropriate subset of the information instruction in relation to each simple bonding pad or the simple bonding pad for second array Whether parameter:

It is exported from the coding parameter that the part of first array corresponds to simple bonding pad, for such derived volume It is appropriate with each simple bonding pad of second array or the simple bonding pad that mode as defined in code parameter is decoded The relevant payload data of subset；Or

Wherein, described image is the image of video.

12. a kind of method for transmitting video, comprising:

Transmitting data stream over a transmission medium, the data flow represent the different spaces sample intelligence point of the image of plane field scape Amount, each plane includes message sample array, and the data flow includes:

Wherein, described image is the image of video.

13. method described in any one of 1 to 12 according to claim 1, wherein described image comes from video, and the number The video with depth information is encoded as according to stream.

14. method described in any one of 1 to 12 according to claim 1, wherein the video includes being used as first array Luma samples array together with two chroma sample arrays including second array.

15. method described in any one of 1 to 12 according to claim 1, wherein the plane of described image includes independently being compiled Two colorimetric planes and a luminance plane of code.