CN110505488A - The image of extended reference pixel sample value collection prediction pixel array encodes or coding/decoding method - Google Patents
The image of extended reference pixel sample value collection prediction pixel array encodes or coding/decoding method Download PDFInfo
- Publication number
- CN110505488A CN110505488A CN201910946605.2A CN201910946605A CN110505488A CN 110505488 A CN110505488 A CN 110505488A CN 201910946605 A CN201910946605 A CN 201910946605A CN 110505488 A CN110505488 A CN 110505488A
- Authority
- CN
- China
- Prior art keywords
- sample value
- pixel
- decoding
- block
- pixel sample
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Landscapes
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
The present invention provides a kind of method for compressing image.When carrying out coding or the decoding of 2 dimension matching ways to a current block, whenever there is the new reference pixel sample value reconstructed to generate, when thering is new reference zone to generate to have the reference zone boundary of emerging new shape, the region of the reference pixel sample value reconstructed is extended first, i.e. on boundary (including at least the reference zone boundary of the emerging new shape, also allow to include part or all of original boundary) filling is obtained according to the rule pre-established in outer a certain range pixel number, to during 2 dimension matchings encode or are decoded, allow to match reference strings block to extend to except the region of the reference pixel sample value reconstructed.
Description
The application is the divisional application of following original application:
The applying date of original application: 2015-03-18
The application number of original application: 2015101185917
The invention and created name of original application: the image coding or coding/decoding method of extended reference pixel sample value collection
Technical field
The present invention relates to a kind of compression of digital video encoding and decoding system, especially computer screen images and video
The method of encoding and decoding.
Background technique
The natural form of the digital video signal of image is the sequence of image.One frame image is usually to be made of several pixels
Rectangular area, and digital video signal is exactly the sequence of video images being made of tens frames to thousands of frame images, sometimes
Also referred to as video sequence or sequence.Digital video signal is encoded be exactly according to certain sequence to image one by one into
Row coding.At any one time, that frame encoded is known as current encoded frame.Equally, to the digital video signal of compression
Video code flow be decoded and be exactly decoded according to video code flow of the same sequence to image one by one.A period of time in office
It carves, that frame decoded is known as current decoded frame.Current encoded frame or current decoded frame are all referred to as present frame.
In the international standard such as MPEG-1/2/4 of nearly all encoding video pictures, H.264/AVC and the newest world is regarded
Frequency compression standard HEVC(High Efficiency Video Coding) in, a frame image is encoded (and it is corresponding
Decoding) when, a frame image is divided into the subgraph of several pieces of MxM pixels, referred to as " (Coding Unit is referred to as coding unit
CU) ", using CU as basic coding unit, one piece one piece of subgraph is encoded.The size of common M is 8,16,32,64.Cause
This, encodes a sequence of video images and is exactly successively encoded to each coding unit of each frame.Equally, when decoding
And each coding unit of each frame is successively decoded according to same sequence, finally reconstruct entire video image sequence
Column.
For the difference for adapting to each section picture material and property in a frame image, most effective volume is targetedly carried out
Yard, the size of each CU can be different in a frame image, has plenty of 8x8, has plenty of 64x64, etc..In order to make different size
CU can be seamless spliced, a frame image be always first divided into size it is identical with NxN pixel " maximum coding is single
First (Largest Coding Unit abbreviation LCU) ", then each LCU is further divided into multiple sizes of tree not
Certain identical CU.Therefore, LCU is also referred to as " coding tree unit (Coding Tree Unit abbreviation CTU) ".It is same as CTU big
Small CU be known as depth D be 0(D=0) CU.The CU for a quarter size that the CU quartering of one D=0 obtains is called depth
D be 1(D=1) CU.The CU of the D=1 smaller CU of size that the quartering obtains again call depth D be 2(D=2) CU.
The CU of the one D=2 smaller CU of size that the quartering obtains again be known as depth D be 3(D=3) CU.The CU of D=3 again four
The smaller CU of the size that equal part obtains be known as depth D be 4(D=4) CU.The concept for using " depth " and " serial number " is for letter
Just the shape and size of CU are indicated.For example, a frame image is first divided into depth D=0 of the identical 64x64 pixel of size
LCU(N=64).One LCU can be as shown in Figure 10 and 15 by numbering in CU(Fig. 1 of D=1 of 2 32x32 pixels
CU), in CU(Fig. 1 of D=2 of 6 16x16 pixels number be 1,2,3,4,9 and 10 CU) and 8 8x8 pixels D=3 CU
(CU that number is 5,6,7,8,11,12,13 and 14 in Fig. 1) is constituted.Such 16 CU at tree constitute a CTU.
One LCU can also be as shown in Figure 2 by numbering the CU for being 0,5 and 6 in CU(Fig. 2 of D=1 of 3 32x32 pixels) and 4
The CU that number is 1,2,3 and 4 in CU(Fig. 2 of D=2 of 16x16 pixel) it constitutes, such 7 also constitute one at the CU of tree
A CTU.Therefore, under scheduled tree division rule (such as simplest simple quartering), depth and serial number itself do not have
There is special meaning, is only the shape of the CU of tree and a kind of easy representation of equal value of size.To a frame image
It is encoded, is exactly successively to be encoded to the CU one by one in CTU one by one.At any one time, it is encoding
CU is known as present encoding CU.One frame image is decoded, and according to same sequence successively to one in CTU one by one
An a CU is decoded.At any one time, the CU decoded is known as current decoding CU.Present encoding CU or current decoding
CU is referred to as current CU.In Fig. 1 and Fig. 2 CU number sequence also exactly to CU one by one carry out coding or it is decoded suitable
Sequence.
All CU in one CTU have a depth D and a serial number.As shown in figure 3, the CU of depth D=0 only has one
It is a, serial number 0.As shown in figure 4, the CU of depth D=1 has 4, serial number 0 ~ 3.As shown in figure 5, the CU of depth D=2 has 16
It is a, serial number 0 ~ 15.As shown in fig. 6, the CU of depth D=3 has 64, serial number 0 ~ 63.As shown in fig. 7, depth D=4
CU has 256, serial number 0 ~ 255.
For more generally, Fig. 3 ~ 7 show also the tree-shaped segmentation to CTU.Fig. 3 is the segmentation to depth D=0 of CTU,
Only one segmentation, serial number 0.Fig. 4 is the segmentation to depth D=1 of CTU, shares 4 segmentations, serial number 0 ~ 3.Fig. 5
It is the segmentation to depth D=2 of CTU, shares 16 segmentations, serial number 0 ~ 15.Fig. 6 is the segmentation to depth D=3 of CTU, altogether
There are 64 segmentations, serial number 0 ~ 63.Fig. 7 is the segmentation to depth D=4 of CTU, shares 256 segmentations, serial number 0 ~
255。
With MPEG-1/2/4, H.264/AVC and HEVC etc. for representative in the prior art, in order to improve coding effect
One maximum CU of CU(usually still is allowed to be frame image itself by rate) it is further divided into smaller subregion.It is described
Subregion includes but is not limited to: predicting unit (PU), converter unit (TU), and asymmetry divides the region of (AMP), and two divide generation
Region, three divide the region generated, macro block, block, microlith, and (width or height are a pixel or pixel component to item
Region), the rectangular area of variable-size, the pixel string (section) or pixel component string (section) or pixel index string of variable-size
(section).It is exactly to be encoded to subregion one by one (and accordingly that (and corresponding decoding) is encoded to a CU
Decoding).In coding, subregion is known as encoding subregion, and in decoding, subregion is known as decoding subregion.Encode sub-district
Domain reconciliation numeral region is referred to as encoding and decoding subregion.In the prior art, subregion (especially predicting unit, the transformation
Unit, the region that asymmetry divides, macro block, block, microlith, the situation of item) usually still it is known as " block ".So coding subregion
Reconciliation numeral region is often referred to as encoding block and decoding block in many cases, is referred to as encoding and decoding block.Correspondingly, by CU
It is further divided into during smaller encoding and decoding subregion is also referred to as encoding and decoding block, the depth of division is also further increased and produced
Raw new serial number becomes no longer easy, especially in the case of non-equal part divides, indicates encoding and decoding block with depth and serial number
Shape actually have become very complicated with size, therefore, the shape of encoding and decoding block might as well be directly used using depth and serial number
Shape and size are more convenient, become apparent from, is simpler.
One color images is known as 3 component (component) compositions.Most common two kinds of pixel color formats (pixel
Color format) the GBR color format that is made of green component, blue component, red component and by a brightness
(luma) the common name YUV color format such as YCbCr color format of component and two coloration (chroma) components composition.Therefore, right
When one CU is encoded, a CU can be divided into 3 component planes (G plane, B plane, R plane or Y plane, U plane, V
Plane), 3 component planes are encoded respectively；Can also 3 component binding combinations of a pixel at 3 tuples,
The CU being made of these 3 tuples is integrally encoded.The arrangement mode of former pixel and its component be known as image (and its
CU planar format (planar format)), and the arrangement mode of latter pixel and its component is known as image (and its CU)
Folded packet format (packed format).
By taking the GBR color format p [x] [y] of pixel={ g [x] [y], b [x] [y], r [x] [y] } as an example, a kind of plane
The arrangement mode of format is that first arrangement width is owning for a frame image (or a CU) that W pixels tall is H pixel
Then WxH G component arranges all WxH B components, finally arrange all WxH R components:
g[1][1],g[2][1],…,g[W-1][1],g[W][1],
g[1][2],g[2][2],…,g[W-1][2],g[W][2],
………………………………………,
………………………………………,
g[1][H],g[2][H],…,g[W-1][H],g[W][H],
b[1][1],b[2][1],…,b[W-1][1],b[W][1],
b[1][2],b[2][2],…,b[W-1][2],b[W][2],
………………………………………,
………………………………………,
b[1][H],b[2][H],…,b[W-1][H],b[W][H],
r[1][1],r[2][1],…,r[W-1][1],r[W][1],
r[1][2],r[2][2],…,r[W-1][2],r[W][2],
………………………………………,
………………………………………,
r[1][H],r[2][H],…,r[W-1][H],r[W][H]。
And a kind of arrangement mode of folded packet format is the G component for first arranging first pixel, then arranges its B component and R
Component then arranges G component, B component and the R component, etc. of second pixel, finally arrange the last one (WxH) as
G component, B component and the R component of element:
g[1][1],b[1][1],r[1][1], g[2][1],b[2][1],r[2][1], …………, g[W][1],b[W]
[1],r[W][1],
g[1][2],b[1][2],r[1][2], g[2][2],b[2][2],r[2][2], …………, g[W][2],b[W]
[2],r[W][2],
………………………………………………………………………………………,
………………………………………………………………………………………,
g[1][H],b[1][H],r[1][H], g[2][H],b[2][H],r[2][H], ………, g[W][H],b[W]
[H],r[W][H]。
The arrangement mode of this folded packet format is expressed as in which also can simplify:
p[1][1],p[2][1],……,p[W-1][1],p[W][1],
p[1][2],p[2][2],……,p[W-1][2],p[W][2],
…………………………………………,
…………………………………………,
p[1][H],p[2][H],……,p[W-1][H],p[W][H]。
Other than the arrangement mode of the arrangement mode of above planar format and folded packet format, according to the difference of three components
Sequentially, can also there are the arrangement mode of other a variety of planar formats and the arrangement mode of folded packet format.
YUV color format again can be according to whether carrying out down-sampling to chromatic component be subdivided into several different sub-formats: 1 picture
The YUV4:4:4 pixel color format that element is made of 1 Y-component, 1 U component, 1 V component；2 adjacent pixels of left and right are by 2
The YUV4:2:2 pixel color format that a Y-component, 1 U component, 1 V component form；Left and right is neighbouring to press the spatial position 2x2
The YUV4:2:0 pixel color format that 4 pixels of arrangement are made of 4 Y-components, 1 U component, 1 V component.One-component
Generally indicated with the number of 18 ~ 16 bit.YUV4:2:2 pixel color format and YUV4:2:0 pixel color format are all
The down-sampling for implementing chromatic component to YUV4:4:4 pixel color format obtains.One pixel component is also referred to as a pixel sample value
(pixel sample) or it is simply referred as a sample value (sample).One sample value can be number i.e. one of 8 bits
Sample value occupies a byte.One sample value is also possible to the number of 10 bits or the number of 12 bit or 14 bit
Several or 16 bits numbers.
When any one encoding and decoding block is encoded or decoded, reconstructed pixel can be all generated, and is divided into coding or decoding
The Perfect Reconstruction that the different degrees of partial reconfiguration pixel and coding or decoding process generated among process generates after being fully completed
Pixel.If Perfect Reconstruction pixel sample value has equal numerical value with the pixel sample value that is originally inputted before coding, experienced
Coding and decoding process is known as lossless coding and decoding.If being originally inputted pixel before Perfect Reconstruction pixel sample value and coding
Sample value has unequal numerical value, then coding and decoding process experienced is known as lossy coding and decoding.To compiling one by one
When decoding block is successively encoded or decoded, generated reconstructed pixel sample value is usually all got up as storage of history data P, is used
Make reference pixel sample value when coding or the decoding of subsequent encoding and decoding block.After being used as due to the sole purpose of reference pixel sample value
Prediction pixel sample value when the coding for decoding block of sequeling or decoding, reference pixel sample value are also referred to as prediction pixel sample value or are
With pixel sample value.Reconstructed pixel historical data is saved as the memory space of reference pixel sample value and is known as the storage of reference pixel sample value
Space, also referred to as reference pixel sample value collection or referred to as reference zone, or also referred to as prediction pixel sample value array or for
With pixel sample value array.Reference pixel sample value memory space is all limited, and can only save a part of historical data.Reference pixel
Historical data in sample value memory space also may include the reconstructed pixel sample value of current codec block.
With using remote desktop as the development of the cloud computing of new generation of classic manifestations and messaging model and platform
With it is universal, between multiple stage computers, other digital devices such as main frame and smart television, smart phone, tablet computer it
Between and various all kinds of digital devices between interconnection have become reality and increasingly as a kind of main trend.This makes server
The real time screen transmission for holding (cloud) to user terminal becomes current urgent need.Since the screen video data volume for needing to transmit is huge
Greatly, the data compression of high-effect high-quality must be carried out to computer screen image.
The characteristics of making full use of computer screen image carries out the compression of Ultra-High Efficiency to computer screen image, and
A main target of newest international video compression standards HEVC.
One distinguishing feature of computer screen image is that many similar or even complete phases are usually had in same frame image
Same pixel pattern (pixel pattern).For example, the Chinese or foreign language that often occur in computer screen image, are all
It is made of a few basic strokes, many similar or identical strokes can be found in same frame image.Computer screen
Common menu, icon etc. in image, it may have many similar or identical patterns.Existing screen picture and video compress skill
Various matching ways are used in art, including (also referred to as intra frame motion compensation or intra block are multiple for intra prediction mode, Block- matching
System) mode, microlith matching way, microlith String matching mode, finely division matching way, palette matching way, 1 dimension String matching
Mode, 2 maintenance shape String matchings (matching of referred to as 2 maintenance shapes) find the matchings of various different sizes and shapes, realize to screen
The high efficient coding of curtain image.In various matching ways, Block- matching mode has match parameter few to reach in certain situations
To very high code efficiency, and the more generally 2 maintenance shape matching ways including Block- matching can be looked in very large range
To fairly precise matching to which very high code efficiency can be reached in certain situations.
Block- matching mode is exactly (referred to as to be matched with several suitable pixel sample value blocks in reference pixel sample value memory space
Reference block or abbreviation reference block) come approximate or accurately matches and (represent) that the pixel sample value block in current CU (referred to as matches and works as
Preceding piece or abbreviation current block) and view is recorded and passed through with partition mode and/or matching position (being referred to as matching relationship parameter)
Frequency code stream come transmit matching reference block with match the relationship between current block, to be stored in decoding end using reference pixel sample value
Matching reference block and matching relationship parameter in space can obtain matching current block.Matching relationship parameter usually only takes up seldom
Bit number, far fewer than the bit number that occupies of matching current block itself, thus good data compression effects can be reached.
In Block- matching mode, as shown in figure 8, matching current block ties up shapes with matching reference block usually and having identical size and 2
Shape (width and height).
In Block- matching mode, block is further subdivided into microlith, just becomes microlith matching way.
In Block- matching mode, block is done fine division again, just becomes fine and divide matching way.
2 maintenance shape matching ways are exactly with several suitable pixel sample value (referred to as matching ginsengs in reference pixel sample value collection
Examine sample value) come approximate or accurately matches pixel sample value (referred to as matching current sample) in (represent) current CU and with matching
Position, matching length and do not match sample value (being referred to as matching relationship parameter) come record and transmitted by video code flow matching ginseng
Examine sample value with match the relationship between current sample, to refer to sample value using the matching in reference pixel sample value collection in decoding end
Matching current sample can be obtained with matching relationship parameter.Matching relationship parameter usually only takes up seldom bit number, far fewer than
The bit number that matching current sample itself occupies, thus good data compression effects can be reached.
In 2 maintenance shape matching ways, as shown in figure 8, the pixel sample value in current CU is divided into sectional pixel sample
Value string, these pixel sample value strings (the referred to as current string of matching or referred to as current string) all have corresponding in reference pixel sample value collection
With reference strings or abbreviation reference strings.Without the pixel sample of corresponding reference pixel sample value in reference pixel sample value collection in current CU
Value does not match sample value referred to as.The characteristics of 2 maintenance shapes match is to match current string to have identical 2 dimension shape with reference strings are matched
Shape.In the matching of vertical 2 maintenance shape, the height of all matching strings is exactly the height of current CU all.In 2 maintenance shape of level
In matching, the width of all matching strings is exactly the width of current CU all.
2 maintenance shape matching ways are a kind of String matching modes.
Due to coding and decoding be all that CU is carried out one by one as unit of CU, to a current CU carry out encode or
When decoding, the region Buddhist monk of the pixel sample value reconstructed is uncoded or the region of decoded pixel sample value is to be cross-linked together
, as shown in Figure 9.Matching reference strings (block) must keep and match current string (block) same shape, therefore match ginseng
Examine except the very possible region (i.e. reference pixel sample value collection) for extending to the pixel sample value reconstructed of string (block) (but matching reference
String or block can not be all over except the regions of the pixel sample value reconstructed), as in Fig. 95 matching reference strings and 3
With shown in reference block, leftmost matching reference strings have 4x2 pixel to extend out, and the matching reference strings of the second from left have 8x1 picture
Element extends out, and intermediate matching reference strings have 1x10 pixel to extend out, the matching ginseng below intermediate matching reference strings
Examining block has 2x8 pixel to extend out, and the matching reference strings on the right side two have 2x1 pixel to extend out, the matching reference of rightmost
String there is 3x2 pixel to extend out, the matching reference block of the top have 8x1 pixel extend to present image coboundary it
Outside, the matching reference block of rightmost has 1x8 pixel to extend to except the right margin of present image.But in existing 2 maintenance shape
With in mode and Block- matching mode, matching reference strings (block) is cannot extend into except the region of the pixel sample value reconstructed, institute
It must be truncated in zone boundary with matching reference strings (block), matching current string (block) also must just be truncated.In this way, not only existing
There are 2 maintenance shape matching ways and Block- matching mode that there must be complicated out-of-bounds treatment mechanism, checks matching reference strings all the time
(block) whether out-of-bounds if out-of-bounds will do specially treated cause computational complexity very big, nor can give full play to 2 dimensions
The superiority of conformal matching way and Block- matching mode.
In the present patent application, to describe for purpose of brevity, the matching of 2 maintenance shapes, microlith matching and Block- matching are referred to as 2 dimensions
Match, matching string, matching microlith and match block are referred to as matching string block, and matching reference strings, matching are with reference to microlith and matching reference block
It is referred to as matching reference strings block, matching is currently gone here and there, matches current microlith and matched current block and is referred to as matching current string block, reference
String with reference to microlith and reference block is referred to as reference strings block, and current string, current microlith and current block are referred to as currently going here and there block.
It should be noted that " matching " is the operation of coding, corresponding reconstruct and decoding operate are often " duplication ".Cause
This, various matching ways such as Block- matching mode, finely divides matching way, String matching mode, 2 maintenance shapes at microlith matching way
Matching way etc. is also referred to as that block copy mode, microlith copy mode, finely to divide copy mode, string copy mode, 2 maintenance shapes multiple
Mode processed etc..In coding and decoding, " matching " is inherently the pixel usually predicted in current codec using reference image,
Therefore, matching encoding and decoding traditionally also commonly referred to as prediction encoding and decoding.
Summary of the invention
It is of the invention be mainly technically characterized by a current block (such as current PU or current CU or current CTU or
The current band of person or present image) carry out the coding of 2 dimension matching ways or when decoding, whenever there is the new reference image reconstructed
Plain sample value is also referred to as prediction pixel sample value or referred to as matched pixel sample value generates, that is, has new reference zone also referred to as to predict picture
Plain sample value array or for matched pixel sample value array generate toThere is emerging new shapeReference zoneBoundaryAlso referred to as
Prediction pixel sample value arrayBoundaryIt or is matched pixel sample value arrayWhen boundary, first to the reference pixel sample value reconstructed
RegionIt is extended, i.e., (prediction pixel is also referred to as including at least the reference zone boundary of the emerging new shape on boundary
Sample value array boundary or be matched pixel sample value array boundary also allows to include part or all of original boundary) certain model outside
The interior pixel number filled and obtained according to the rule pre-established is enclosed, to allow during 2 dimension matchings encode or is decoded
Matching reference strings block extends to except the region of the reference pixel sample value reconstructed, does not need to check that matching reference strings block is constantly
No out-of-bounds, also available more complete matching reference strings block is to improve coding efficiency.It is exported from this technical characteristics
A subordinate technical characteristic be different phase in entire coding or decoding process (for example, to the current CU of different serial numbers
When being encoded or being decoded) allow to carry out same reference zone and its boundary repeatedly different extensions, i.e., when each time extension, expansion
The numerical value of the pixel sample value of exhibition is not necessarily all mutually identical for each time.
Figure 10 is when the depth of the CTU to serial number n is encoded or decoded for the CU of 2 serial numbers 2, in order to avoid matching
The complicated out-of-bounds treatment mechanism of reference strings block and an example for improving minimal expansion region required for coding efficiency.Current CU
Depth D=2(16x16 pixel), serial number 2.New reference zone boundary is the CU that depth is 2 serial numbers 1 in the CTU of serial number n
Right margin and lower boundary.Since matching reference strings block is at least a part of within the region of the pixel sample value reconstructed, and
The width and height that match reference strings block itself can not be more than the width and height of current CU, so the pixel sample value reconstructed
The width in minimal expansion region of right margin in region be that the width of current CU subtracts one (as unit of pixel sample value), reconstructed
The height in minimal expansion region of lower boundary in region of pixel sample value be that subtract one (be single with pixel sample value for the height of current CU
Position).Figure 10 is also illustrated to same reference zone and its boundary, can be repeatedly extended, the number of the pixel sample value extended every time
Be worth it is different establish a capital it is identical.For example, to the coboundary of the current CTU of serial number n, carrying out the CU that depth is 2 serial numbers 1 in Figure 10
Coding or when decoding, carry out first time extension, in coding or the decoding for carrying out the CU that depth is 2 serial numbers 2, carry out second
Secondary extension, extension calculates the numerical value of the pixel sample value of extension using different rules twice.For another example in Figure 10, to image
Coboundary carries out first time extension when reference pixel is the reconstructed pixel without block-eliminating effect filtering, is in reference pixel
By block-eliminating effect filtering reconstructed pixel when carry out second of extension again, the numerical value of the pixel sample value extended twice is obviously different
It establishes a capital identical.
In coding method of the invention, most basic peculiar technical characteristic is exactly one current block (such as current PU of every coding
Either current CU or current CTU or current band or present image), there is the new reference pixel sample value reconstructed to produce
It is raw, that is, there is new reference zone to generate to haveEmerging new shapeReference zoneWhen boundary, first according to current block
Depth and serial number are directAccording to the shapes and sizes of current block, to the reference pixel sample value collection region of 2 dimensionsIt carries out outside boundary Extension, i.e., (including at least the reference zone boundary of the new shape, also allow in the reference pixel sample value collection zone boundary of 2 dimensions
Including part or all of original boundary) it is also referred to as filling in a certain range according to pre-establishing on prediction pixel sample value array the boundary outside
The obtained pixel number of rule, the reference pixel sample value collection after forming an extension, then reference pixel sample after expansion
2 dimension matching codings are carried out in value collection and are also referred to as predictive coding, in an encoding process, allow to match the out-of-bounds of reference strings block, without inspection
Look into matching reference strings block whether out-of-bounds；It in an encoding process, can also be according to the picture of the Policy Updates extended area pre-established
Prime number value.
In coding/decoding method of the invention, most basic peculiar technical characteristic is exactly that foundation is read from video codeword data stream
Information or the depth of current block and the shapes and sizes of serial number or current block, every decoding are calculated according to decoding order
One current block (such as current PU or current CU or current CTU or current band or present image), there is new weighed
The reference pixel sample value of structure generates, that is, there is new reference zone to generate toThere is emerging new shapeReference zoneBoundary When, first according to the depth of current block and serial number or directlyAccording to the shapes and sizes of current block, to the reference pixel sample of 2 dimensions
Value collection regionCarry out boundary external expansion, i.e., (included at least described emerging new in the reference pixel sample value collection zone boundary of 2 dimensions
The reference zone boundary of shape also allows to include part or all of original boundary) it is also referred to as outside prediction pixel sample value array boundary
The pixel number that filling is obtained according to the rule pre-established in a certain range, the reference pixel sample value after forming an extension
Collection, the interior 2 dimension matching and decodings that carry out of reference pixel sample value collection then after expansion are also referred to as prediction decoding, in decoding process,
Allow match the out-of-bounds of reference strings block, without check matching reference strings block whether out-of-bounds；It, can also be according to preparatory in decoding process
The pixel number of the Policy Updates extended area of formulation.
Unquestionably, coding and decoding must use pixel number consistent with each other to be extended.
Illustrate technical characteristic of the invention above by several specific specific examples.Those skilled in the art can be by this theory
Other advantages and efficacy of the present invention can be easily understood for the bright revealed content of book.The present invention can also pass through in addition different tools
Body embodiment is embodied or practiced, and the various details in this specification can not also carried on the back based on different viewpoints and application
From carrying out various modifications or alterations under spirit of the invention.
According to an aspect of the invention, there is provided a kind of image encoding method, including at least the following steps at least it
One:
1) new reference pixel sample value is generated in coding and is also referred to as prediction pixel sample value, that is, is generated new reference zone and be also referred to as
Prediction pixel sample value array toThere is emerging new shapeReference zoneBoundaryAlso referred to as prediction pixel sample value arrayBoundary When, prediction pixel sample value array is also referred to as to reference pixel sample value collection regionIt is extended, i.e., right according to the rule pre-established
Reference pixel sample value collection zone boundary (including at least the reference zone boundary of the emerging new shape, also allows to include portion
Point or whole original boundaries) be also referred to as a certain range of region (such as aforementioned minimal expansion area outside prediction pixel sample value array boundary
Domain or region with boundary direct neighbor pixel) pixel number is assigned, the reference pixel sample value collection for forming an extension is also referred to as
Prediction pixel sample value array；
2) at least prediction pixel sample value array is also referred to as using the reference pixel sample value collection of the extension to continue encoding block
Coding, and video code flow is written into coding result.
According to another aspect of the present invention, a kind of picture decoding method is additionally provided, extremely including at least the following steps
It is one of few:
1) new reference pixel sample value is generated in decoding and is also referred to as prediction pixel sample value, that is, is generated new reference zone and be also referred to as
Prediction pixel sample value array toThere is emerging new shapeReference zoneBoundaryAlso referred to as prediction pixel sample value arrayBoundary When, prediction pixel sample value array is also referred to as to reference pixel sample value collection regionIt is extended, i.e., right according to the rule pre-established
Reference pixel sample value collection zone boundary (including at least the reference zone boundary of the emerging new shape, also allows to include portion
Point or whole original boundaries) be also referred to as a certain range of region (such as aforementioned minimal expansion area outside prediction pixel sample value array boundary
Domain or region with boundary direct neighbor pixel) pixel number is assigned, the reference pixel sample value collection for forming an extension is also referred to as
Prediction pixel sample value array；
2) video code flow is parsed, is at least also referred to as prediction pixel sample value array to solution using the reference pixel sample value collection of the extension
Code block continues to decode.
In the present invention, the encoding block or decoding block be image a coding region or a decoding region, including with
Lower at least one: maximum coding unit LCU, coding tree unit CTU, the subregion of coding unit CU, CU, sub- coding unit
SubCU, coded sub-blocks coding subblock, prediction block, prediction sub-block prediction subblock, predicting unit PU,
The subregion of PU, sub- predicting unit SubPU, macro block.
Detailed description of the invention
Fig. 1 is that a kind of CU of a CTU is divided and tree example
Fig. 2 is that another CU of a CTU is divided and tree example
Fig. 3 is the CU of depth D=0 or the serial number of segmentation in a CTU
Fig. 4 is 4 CU of depth D=1 in a CTU or the serial number of segmentation
Fig. 5 is 16 CU of depth D=2 in a CTU or the serial number of segmentation
Fig. 6 is 64 CU of depth D=3 in a CTU or the serial number of segmentation
Fig. 7 is 256 CU of depth D=4 in a CTU or the serial number of segmentation
Fig. 8 is the current string (block) of the matched matching of 2 dimensions and the example for matching reference strings (block)
Fig. 9 is the example that matched matching reference strings (block) of 2 dimensions extends to the situation except the reference pixel sample value collection reconstructed
Figure 10 is minimal expansion region required in order to avoid matching the complicated out-of-bounds treatment mechanism of reference strings (block)
Figure 11 is the coding method flow diagram of one embodiment of the present of invention
Figure 12 is the coding/decoding method flow diagram of one embodiment of the present of invention
Specific embodiment
One embodiment of coding method of the invention, flow diagram is as shown in figure 11, includes at least following steps
At least one of:
1) assessment of encoding block specificity analysis, pre-treatment and coding mode selection step；To present encoding block and adjacent to encoding block
Pixel sample value characteristic carries out analysis assessment, including carries out necessary pre-treatment to pixel sample value, prejudges 2 dimensions matching coding staff
For formula if appropriate for being used to encode present encoding block, this step is optional, it can skips this step, is directly entered down
One step；The example of the method for the analysis assessment: the method for foundation or the coding result with reference to neighbouring encoding block calculates current volume
The method of the number of the pixel of different colours in code block；The example of the pre-treatment: sample value quantization, color quantizing and based on color
Pixel cluster, the color that input original pixels are indicated with palette and index；
2) extended reference pixel sample value collection step；According to the depth of present encoding block and serial number or directly according to present encoding block
Shapes and sizes, to 2 dimension reference pixel sample value collection region be extended, i.e., 2 dimension reference pixel sample value collection regional edge
The out-of-bounds pixel number that filling is obtained according to the rule pre-established in a certain range, the reference pixel sample value collection is plus extension
Region collectively form extension after reference pixel sample value collection；To the right margin in reference pixel sample value collection region, extension is most
Small width (as unit of pixel sample value) is that the width of present encoding block subtracts one；To the lower boundary in reference pixel sample value collection region,
The minimum constructive height (as unit of pixel sample value) of extension is that the width of present encoding block subtracts one；This is in order to avoid matching reference strings
Minimal expansion region required for the complicated out-of-bounds treatment mechanism of block；Actual extended area can be greater than the minimal expansion
Region；Extension filling carries out usually before step 3), but can also carry out again among step 3)；It the region of extension can be complete
Portion or part cover the region of present encoding block；
3) 2 dimension matching coding step；It is compiled using the reference pixel sample value collection after 2 dimension matching coding modes and the extension to current
Code block carries out 2 dimension matching encoding operations；It in an encoding process, can also be according to the picture of the Policy Updates extended area pre-established
Prime number value；The input of the 2 dimension matching coding is input original pixels or its pixel Jing Guo pre-treatment；The 2 dimension matching is compiled
The output of code is matching position, optional matching length, does not match sample value optionally, matches residual error；The matching position is to use
Sample value is referred in the reference pixel sample value collection to indicate to match the matching that current sample matches with the present encoding block
Interior variable where；Optionally, (referred to as matching is current with the pixel sample value string of variable-length for 2 dimension matching coding modes
String, position can both be indicated with one 2 dimension coordinate or with a linear address) it is that unit carries out matching coding；Described
The string block of a matching in the reference pixel sample value collection, referred to as matching reference strings block are formed with reference sample value, position is also
Both it can be indicated with one 2 dimension coordinate or with a linear address, thus in 2 dimension matching coding modes, the match bit
Setting both 2 can tie up the difference of coordinates or with matching reference strings block with match current string block with 2 dimension coordinates of matching reference strings block
Linear address and the difference for matching the currently linear address of string block, are commonly referred to as displacement vector to indicate；Optionally, since matching refers to
The length (length currently gone here and there equal to matching) of string is variable, so optionally, it is also necessary to which another is known as matching length
Variable indicates its length；Optionally, the sample value that do not match is according to predetermined matching criterior in the reference pixel
Matched input original pixels sample value is not found in sample value collection；The current string block of matching has identical with corresponding matching reference strings block
2 dimension shapes；Optionally, the sample value that do not match is an input original pixels sample value, so can also be with it in present encoding
Position in block indicates；Optionally, the sample value that do not match can also use an imitative matching obtained by calculation damaged
Sample value is come approximate；The matching residual error is to input original pixels sample value and the difference matched with reference to sample value, if 2 dimension matchings are compiled
The code predetermined matching criterior of mode is absolutely accurate lossless matching, then the matching residual error is zero, i.e., this 2 dimension matching
Coding mode is just without matching residual error as output, if the 2 dimension matching predetermined matching criterior of coding mode of one kind is close
As damage matching, then the matching residual error may be not zero, another kind damage matched situation be first to input original image
Plain sample value carries out the pre-treatment of sample value quantization, color quantizing or the pixel cluster based on color, then carries out the matching of 2 dimensions again and compiles
Code, in this situation, since sample value quantization, color quantizing or pixel cluster based on color damage, even if 2 dimension matchings
Encoding is lossless in itself, and the matching residual error (i.e. input original pixels sample value and the difference matched with reference to sample value) may also
It is not zero；To the present encoding block carry out 2 dimension matching codings the result is that I(I >=0) a matching goes here and there block and optional J(J
>=0) a non-matched pixel sample value, output I is to (displacement vector, optional matching length) and optional J non-matched pixel samples
Value, wherein I and J can not be zero simultaneously；
4) remaining various common coding and restructing operation step complete remaining all coding and restructing operation of present encoding block,
Various common technologies are carried out to the original pixels of input, various parameters and variable, such as intra prediction, inter-prediction, Block- matching, tune
Colour table matching, transformation, quantization, inverse transformation, inverse quantization, corresponds to prediction residual and matching residual error at imitative matching sample value prediction interpolation
Compensation (taking the inverse operation of residual error operation), predict and ask residual error, DPCM, single order and higher difference, mapping, the distance of swimming, index,
Block-eliminating effect filtering, sample value adaptive equalization (Sample Adaptive Offset), coding and restructing operation and entropy compile
Code operation；The input of this step is above-mentioned steps 3) output and input original pixels and empty from the storage of reference pixel sample value
Between and reference pixel sample value collection reference pixel；The output of this step is reconstructed pixel (including Perfect Reconstruction pixel and different journeys
The partial reconfiguration pixel of degree) and containing 2 dimension matching coding results and other coding results video code flow；The reconstructed pixel is put into
In reference pixel sample value memory space, it is used as subsequent 2 dimension matching encoding operation, remaining various common coding and restructing operation step
Required reference pixel；The video code flow is the last output of this coding method, contains corresponding coding/decoding method and carries out
Whole syntactic element, especially matching positions (i.e. displacement vector) required for decoding and reconstruct, do not match sample value at matching length
Or the syntactic elements such as its position.
One embodiment of coding/decoding method of the invention, flow diagram is as shown in figure 12, includes at least following steps
At least one of:
1) video codeword data stream parsing and partial decoding of h step；To input containing matching position, matching length, do not match sample value
The video code flow of the compressed data of (or its position) and every other syntactic element compressed data carries out entropy decoding, and parses
The meaning for the various data that entropy decoding obtains；(residual error is asked to transport as transformation decodes, predicts and do compensation parsing and partial decoding of h
The inverse operation of calculation, DPCM decoding, single order and higher difference decoding, map decoding, distance of swimming decoding, index decode) after obtained
With position (i.e. displacement vector), matching length, do not match the matching relationships parameters such as sample value (or its position) export to it is subsequent 2 dimension
With decoding step；The depth and serial number of the current decoding block that parsing obtains are exported and walked to subsequent expansion reference pixel sample value collection
Suddenly；In particular, being obtained according to from the information parsed in video codeword data stream or according to being parsed from video codeword data stream
Information add to the characteristic of current decoding block and neighbouring decoding block carry out analysis assessment as a result, determining new reference area
Domain and its new boundary；The example of the method for the analysis assessment: foundation or doing with reference to several decoding results adjacent to decoding block
Method first does a wheel part pre decoding and evaluation part to current decoding block using 2 dimension matching and decoding modes and other decoding processes
The method of pre decoding result；
2) extended reference pixel sample value collection step；According to the depth of current decoding block and serial number or directly according to current decoding block
Shapes and sizes and step 1) obtain information and as a result, to 2 dimension reference pixel sample value collection region be extended, i.e.,
In reference pixel sample value collection zone boundary (including the new boundary and part or all of original boundary) outer a certain range of 2 dimensions
Interior to fill the pixel number obtained according to the rule pre-established, the reference pixel sample value collection adds the region of extension altogether
Reference pixel sample value collection after forming extension；To the right margin in reference pixel sample value collection region, the minimum widith of extension is (with pixel
Sample value is unit) it is that the width of current decoding block subtracts one；To the lower boundary in reference pixel sample value collection region, the minimum constructive height of extension
(as unit of pixel sample value) is that the width of current decoding block subtracts one；This is the complicated out-of-bounds in order to avoid matching reference strings block
Minimal expansion region required for treatment mechanism；Actual extended area can be greater than the minimal expansion region；Extension filling
It carries out, but can also be carried out again among step 3) usually before step 3)；The region of extension can completely or partially cover
The region of current decoding block；
3) 2 dimension matching and decoding step；Using the reference pixel sample value collection after 2 dimension matching and decoding modes and the extension to current solution
Code block carries out 2 dimension matching and decoding operations；It, can also be according to the picture of the Policy Updates extended area pre-established in decoding process
Prime number value；The input of the 2 dimension matching and decoding operation is to parse and decode obtained I from video codeword data stream in step 1)
(I >=0) is to (matching position, optional matching length) and optional J(J >=0) it is a do not match sample value (or its position),
Middle I and J can not be zero simultaneously；The matching position be used to indicate that from reference pixel sample value collection where duplication
With reference sample value and paste it into current decoding block matching current sample position；It is clear that matching current sample is
Matching refers to the duplicate of sample value, and the two is numerically equal；Optionally, 2 dimension matching and decoding modes are with variable-length
Pixel sample value string (the referred to as current string of matching, position can both be indicated with one 2 dimension coordinate or with a linear address) is
Unit carries out 2 dimension matching and decodings；The matching forms the string block of a matching in the reference pixel sample value collection with reference to sample value, claims
To match reference strings block, position is also that can both indicate with one 2 dimension coordinate or with a linear address, thus tie up 2
In matching and decoding mode, the matching position both can with matching reference strings block 2 dimension coordinates with match currently go here and there block
2 dimension coordinates differences can also with matching reference strings block linear address with match currently go here and there block linear address difference, be commonly referred to as
Displacement vector indicates, optionally, since the length (length currently gone here and there equal to matching) of matching reference strings is variable, so
Another is also needed to be known as the variable of matching length to indicate its length；The current string block of matching has with corresponding matching reference strings block
Identical 2 dimension shape；Optionally, the sample value that do not match is to parse and decode obtained pixel directly from video codeword data stream
Sample value and the position for pasting it into pixel sample value in the current decoding of current decoding block, it is described do not match sample value there is usually no
In the reference pixel sample value collection；Optionally, if parsing and decoding from video codeword data stream, obtain is not match sample
Value itself but the position for not matching sample value, then the position for not matching sample value is just output to subsequent step 4) to calculate imitative
With sample value；It is described 2 dimension matching and decoding operation output be the matching current sample (numerically with match refer to sample value phase
Deng) plus optional described sample value (or its position) is not matched；It the matching current sample and that may be present described does not match
Sample value (or its position) all just constitutes the complete 2 dimension matching and decoding output of the current decoding block altogether.
The present invention is suitable for folded packet format-pattern or the coding and decoding of CU.Present invention is equally applicable to component plane lattice
The coding and decoding of formula image or CU.The present invention is suitable for lossless 2 dimension matching coding and decoding.Present invention is equally applicable to have
2 dimension matching coding and decoding of damage.
Presented above illustrates the basic conception that only the invention is illustrated in a schematic way, only shows and the present invention in diagram
Direct related component rather than component count, shape and size when according to actual implementation are drawn, when actual implementation each component
Kenel, quantity and ratio can arbitrarily change for one kind, and its assembly layout kenel may also be increasingly complex.
It is more implementation details and variant of the invention below.
The embodiment 1 of extended reference pixel sample value collection
Prediction pixel sample value array boundary exterior pixel is also referred to as to the reference pixel sample value collection zone boundary and is extended filling institute
The pixel number of imparting is the duplication of the pixel number of pixel adjacent in boundary in a predefined manner；For example, the weight of horizontal direction
The repetition in the direction of multiple or vertical direction repetition or other angles.
The embodiment 2 of extended reference pixel sample value collection
The pixel number filled outside the reference pixel sample value collection zone boundary is the specific pattern that some pixels are formed in boundary
The natural extension of case；For example, the natural of one or several straight line extends in boundary, the extrapolation of one or several curve is prolonged in boundary
It stretches, the natural extension in the region of several different colours in boundary with obvious line of demarcation.
The embodiment 3 of extended reference pixel sample value collection
The pixel number filled outside the reference pixel sample value collection zone boundary is that matching adjacent with boundary in boundary is current
Natural extension of the matching reference strings block corresponding to string block in the reference pixel sample value collection region；For example, in the dimension of level 2
The situation matched, interior three matchings current string such as three matchings in Fig. 8 in CU m+1 adjacent with boundary of a vertical boundary are worked as
Shown in preceding string, their corresponding matching reference strings can extend to the right naturally in the reference pixel sample value collection region,
These pixels extended naturally to the right are replicated and the pixel number as filling.
The embodiment 4 of extended reference pixel sample value collection
The pixel number filled outside the reference pixel sample value collection zone boundary is the pixel number of solid color；For example, one
The pixel number of kind of the black perhaps frequency of occurrences highest one in a kind of pixel number of grey or the former frame of present frame
The highest a kind of pixel of color of the frequency of occurrences on the pixel number of kind color or the reference pixel sample value collection zone boundary
Numerical value.
The embodiment 5 of extended reference pixel sample value collection
The reference pixel sample value collection zone boundary is divided into several segments, is above determined using the different rules pre-established at every section
The pixel number filled outside the fixed segment boundary.
The embodiment 6 of extended reference pixel sample value collection
By the reference pixel sample value collection region for 2 dimension that reconstructed pixel sample value forms, can be from uncoded or decoded
The region of pixel sample value, which gradually changes, to come；At the beginning of being encoded or being decoded to present frame, the preparation of whole frame is deposited
The region for putting reconstructed pixel sample value is essentially all the region of uncoded or decoded pixel sample value；In coding or decoding process
In, it is every to generate a part of reconstructed pixel sample value, with regard to some uncoded or decoded pixel sample value regions transform at reference
The a part in pixel sample value collection region；Therefore, the extended reference pixel sample value collection step every might not start to work as one
Preceding piece all carry out when being encoded or being decoded it is primary, but can with it is every start that a current CTU is encoded or decoded when into
Row is primary；When starting that current CTU is encoded or decoded, just (usually right part inside and around current CTU
And lower part) a certain range all filled with the pixel number that obtains according to the rule pre-established, in this way, to current CTU
In block one by one when being encoded or being decoded, it is not necessary to each piece to carry out the extended reference pixel sample value collection step
Suddenly.
The embodiment 7 of extended reference pixel sample value collection
The extended reference pixel sample value collection step, every might not start to carry out a current block or current CTU coding or
All carried out when decoding it is primary, but can with it is every start that a present frame is encoded or decoded when carry out it is primary, entirely working as
(right part and lower part) a certain range is all filled with the pixel number obtained according to the rule pre-established inside and around previous frame
Get up.
The embodiment 8 of extended reference pixel sample value collection
To the different CTU in present frame, the different rules pre-established can be used to obtain filling out for extended area
The pixel number filled；To the different blocks in present frame or in current CTU, the different rules pre-established can be used
Obtain the pixel number of the filling for extended area.
The embodiment 9 of extended reference pixel sample value collection
The new reference zone boundary is at least the reference zone boundary generated in one or a combination set of following state:
1) encoding block or a decoding block are carried out during encoding or being decoded；
2) after an encoding block or a decoding block being completed coding or are decoded；
3) band is carried out during encoding or being decoded；
4) after a band being completed coding or is decoded；
5) one group of encoding block or one group of decoding block are carried out during encoding or being decoded；
6) after one group of encoding block or one group of decoding block being completed coding or are decoded；
7) one group of CTU is carried out during encoding or being decoded；
8) after one group of CTU being completed coding or is decoded；
9) a line or multirow CTU are carried out during encoding or being decoded；
10) after a line or multirow CTU being completed coding or are decoded；
11) one or more columns per page CTU is carried out during encoding or being decoded；
12) after one or more columns per page CTU being completed coding or is decoded；
13) a CTU array is carried out during encoding or being decoded；
14) after a CTU array being completed coding or is decoded；
15) a frame image is carried out during encoding or being decoded；
16) after a frame image being completed coding or is decoded.
The embodiment 10 of extended reference pixel sample value collection
The new reference zone boundary is at least on the reference zone boundary that one of following state generates:
1) reference pixel is different degrees of partial reconfiguration pixel；
2) reference pixel is Perfect Reconstruction pixel；
3) reference pixel is after predicting but without the reconstructed pixel of other reconstruct steps；
4) reference pixel is after replicating but without the reconstructed pixel of other reconstruct steps；
5) reference pixel is without block-eliminating effect filtering and/or sample value adaptive equalization (Sample Adaptive Offset
That is SAO) step reconstructed pixel；
6) reference pixel is to have passed through block-eliminating effect filtering and/or sample value adaptive equalization (Sample Adaptive Offset
That is SAO) step reconstructed pixel.
The embodiment 11 of extended reference pixel sample value collection
In coding or decoding, to the same region and its boundary in a frame image, at least secondary extension is carried out.It is at least secondary
The example of extension:
Example 1) first time extension is carried out in coding or the decoding of depth D=1, the is carried out in coding or the decoding of depth D=2
Secondary extension；
Example 2) first time extension is carried out when reference pixel is the reconstructed pixel without block-eliminating effect filtering and/or SAO, joining
It examines and carries out second when pixel is the reconstructed pixel that have passed through block-eliminating effect filtering or/and SAO and extend.
The embodiment 12 of extended reference pixel sample value collection
To a kind of specific embodiment of the extension of reference zone and its boundary, at least come using one or a combination set of following method
It realizes:
1) using the array of an expression extended area, the numerical value assignment of the pixel sample value of extended area is given the array；
2) expanded using one from the coordinate and its numerical value of pixel sample value and the coordinate of extension rear region of extension forefoot area to determine
The rule for opening up the numerical value of pixel sample value on the coordinate of rear region carries out reference zone using the rule " indirectly i.e. not straight
Connect using the array for indicating extended area " extension；
3) it is converted into the mapping of the coordinate of extension forefoot area using a coordinate extension rear region, reuses extension forefoot area
Coordinate on pixel sample value numerical value come determine extension rear region coordinate on pixel sample value numerical value, to reference zone carry out
" indirectly i.e. not directly using the array for indicating extended area " extension.
Method 3) be actually method 2) a kind of special case.
Method 2) and method 3) an example:
The value range of the coordinate (x, y) of reference zone (extension forefoot area) is 0≤x < W and 0≤y < H, pixel on coordinate
The numerical value of sample value is P(x, y)；For the region (extension rear region) outside reference zone boundary, the value range of coordinate (x, y)
It is x<0 or x>=W or y<0 or y>=H；It is operated using following limit Clip
The coordinate (x, y) of extension rear region is converted (mapping) into the coordinate (x of extension forefoot area0, y0), in which:
x0=Clip(0, W-1, x)
y0=Clip(0, H-1, y)
Later, the numerical value of pixel sample value is all unified before extension and on the coordinate of extension rear region is obtained with following calculation rule:
P(Clip(0, W-1, x), Clip(0, H-1, y)).
Claims (10)
1. a kind of image encoding method, it is characterised in that at least include the following steps:
1) it extends prediction pixel sample value array step: generating new prediction picture when generating new prediction pixel sample value in coding
Plain sample value array toThere is emerging new shapePrediction pixel sample value arrayWhen boundary, right according to the rule pre-established
A certain range of region assigns pixel number, the prediction outside the prediction pixel sample value array boundary of the emerging new shape
Pixel sample value array collectively forms the prediction pixel sample value array after extension plus the region of extension；
2) continue coding step: at least encoding the mode for being also referred to as predictive coding and the prediction picture after the extension using 2 dimension matchings
Plain sample value array continues to encode to present encoding block；In an encoding process, also allow to update extension according to predetermined way
Prediction pixel sample value array afterwards；And video code flow is written into coding result.
2. a kind of picture decoding method, it is characterised in that at least include the following steps:
1) video codeword data stream parsing and entropy decoding step: entropy decoding is carried out to the video code flow of input, and parses entropy decoding
The meaning of obtained various data；According to the information that is parsed from video codeword data stream or according to from video codeword data stream
The middle obtained information that parses is along at least to current decoding block progress partial-decoding operation as a result, determine new prediction pixel
Sample value array is to generate the prediction pixel sample value array boundary of emerging new shape；
2) it extends prediction pixel sample value array step: generating new prediction picture when generating new prediction pixel sample value in decoding
Plain sample value array toThere is emerging new shapePrediction pixel sample value arrayWhen boundary, carried out according to the rule pre-established
Boundary external expansion assigns pixel to region a certain range of outside the prediction pixel sample value array boundary of the emerging new shape
Numerical value, the prediction pixel sample value array collectively form the prediction pixel sample value array after extension plus the region of extension；
3) continue decoding step；At least the mode and the prediction picture after the extension that matching and decodings are also referred to as predictive coding are tieed up using 2
Plain sample value array continues to decode to current decoding block；In decoding process, also allow to update extension according to predetermined way
Prediction pixel sample value array afterwards.
3. coding/decoding method according to claim 2, it is characterised in that the rule pre-established is according at least to current
The depth and serial number of block or directly according to the shapes and sizes of current block, carry out boundary to prediction pixel sample value array and extend out
Exhibition.
4. coding/decoding method according to claim 2, it is characterised in that the prediction pixel sample value of the emerging new shape
Array boundary is at least the prediction pixel sample value array boundary generated in one or a combination set of following state:
1) during being decoded to a decoding block；
2) after completing decoding to a decoding block；
3) during being decoded to a band；
4) after completing decoding to a band；
5) during being decoded to one group of decoding block；
6) after completing decoding to one group of decoding block；
7) during being decoded to one group of CTU；
8) after completing decoding to one group of CTU；
9) during being decoded to a line or multirow CTU；
10) after completing decoding to a line or multirow CTU；
11) during being decoded to one or more columns per page CTU；
12) after completing decoding to one or more columns per page CTU；
13) during being decoded to a CTU array；
14) after completing decoding to a CTU array；
15) during being decoded to a frame image；
16) after completing decoding to a frame image.
5. coding/decoding method according to claim 2, it is characterised in that the prediction pixel sample value of the emerging new shape
Array boundary is at least the prediction pixel sample value array boundary generated in one or a combination set of following state:
1) reference pixel is different degrees of partial reconfiguration pixel；
2) reference pixel is Perfect Reconstruction pixel；
3) reference pixel is after predicting but without the reconstructed pixel of other reconstruct steps；
4) reference pixel is after replicating but without the reconstructed pixel of other reconstruct steps；
5) reference pixel is without block-eliminating effect filtering and/or sample value adaptive equalization (Sample Adaptive Offset
That is SAO) step reconstructed pixel；
6) reference pixel is to have passed through block-eliminating effect filtering and/or sample value adaptive equalization (Sample Adaptive Offset
That is SAO) step reconstructed pixel.
6. coding/decoding method according to claim 2, it is characterised in that:
To a kind of specific embodiment of the extension of reference zone and its boundary, at least come using one or a combination set of following method
It realizes:
1) using the array of an expression extended area, the numerical value assignment of the pixel sample value of extended area is given the array；
2) expanded using one from the coordinate and its numerical value of pixel sample value and the coordinate of extension rear region of extension forefoot area to determine
The rule for opening up the numerical value of pixel sample value on the coordinate of rear region, is extended reference zone using the rule；
3) it is converted into the mapping of the coordinate of extension forefoot area using a coordinate extension rear region, reuses extension forefoot area
Coordinate on pixel sample value numerical value come determine extension rear region coordinate on pixel sample value numerical value, to reference zone carry out
Extension.
7. coding/decoding method according to claim 6, it is characterised in that:
Reference zone is that the value range of the coordinate (x, y) of extension forefoot area is 0≤x < W and 0≤y < H, pixel on coordinate
The numerical value of sample value is P(x, y)；For the region outside reference zone boundary, the value range of coordinate (x, y) is x<0 or x>=W
Or y<0 or y>=H；It is operated using following limit Clip
Coordinate (x, the y) conversion of extension rear region is mapped to the coordinate (x of extension forefoot area0, y0), in which:
x0=Clip(0, W-1, x)
y0=Clip(0, H-1, y)
Later, the numerical value of pixel sample value is all unified before extension and on the coordinate of extension rear region is obtained with following calculation rule:
P(Clip(0, W-1, x), Clip(0, H-1, y)).
8. coding/decoding method according to claim 2, it is characterised in that:
Being extended to direct neighbor pixel outside the prediction pixel sample value array boundary and filling assigned pixel number is side
The duplication of the pixel number of direct neighbor pixel in boundary.
9. coding/decoding method according to claim 2, it is characterised in that:
To different decoding blocks, allow to obtain the pixel of the filling for extended area using the different rules pre-established
Numerical value.
10. the coding/decoding method according to any claim in claim 2 to 9, it is characterised in that:
The decoding block or current block are a decoding regions of image, are comprised at least one of the following: maximum coding unit
LCU, coding tree unit CTU, the subregion of coding unit CU, CU, sub- coding unit SubCU, coded sub-blocks coding
Subblock, prediction block, prediction sub-block prediction subblock, the subregion of predicting unit PU, PU, sub- predicting unit
SubPU, macro block.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910946605.2A CN110505488B (en) | 2014-03-18 | 2015-03-18 | Image coding or decoding method for expanding prediction pixel array |
Applications Claiming Priority (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410099776 | 2014-03-18 | ||
| CN201510118591.7A CN104935945B (en) | 2014-03-18 | 2015-03-18 | The image of extended reference pixel sample value collection encodes or coding/decoding method |
| CN201910946605.2A CN110505488B (en) | 2014-03-18 | 2015-03-18 | Image coding or decoding method for expanding prediction pixel array |
Related Parent Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510118591.7A Division CN104935945B (en) | 2014-03-18 | 2015-03-18 | The image of extended reference pixel sample value collection encodes or coding/decoding method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110505488A true CN110505488A (en) | 2019-11-26 |
| CN110505488B CN110505488B (en) | 2022-01-07 |
Family
ID=54122859
Family Applications (2)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510118591.7A Active CN104935945B (en) | 2014-03-18 | 2015-03-18 | The image of extended reference pixel sample value collection encodes or coding/decoding method |
| CN201910946605.2A Active CN110505488B (en) | 2014-03-18 | 2015-03-18 | Image coding or decoding method for expanding prediction pixel array |
Family Applications Before (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510118591.7A Active CN104935945B (en) | 2014-03-18 | 2015-03-18 | The image of extended reference pixel sample value collection encodes or coding/decoding method |
Country Status (1)
| Country | Link |
|---|---|
| CN (2) | CN104935945B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112055219A (en) * | 2020-08-05 | 2020-12-08 | 浙江大华技术股份有限公司 | String matching prediction method and device and computer readable storage medium |
| CN115119046A (en) * | 2022-06-02 | 2022-09-27 | 绍兴市北大信息技术科创中心 | Image coding and decoding method, device and system with reference to pixel set |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP3800891A1 (en) | 2016-10-04 | 2021-04-07 | Ki Baek Kim | Image data encoding/decoding method and apparatus |
| US12022199B2 (en) | 2016-10-06 | 2024-06-25 | B1 Institute Of Image Technology, Inc. | Image data encoding/decoding method and apparatus |
| IT201700024221A1 (en) * | 2017-03-03 | 2018-09-03 | Sisvel Tech S R L | METHODS AND APPARATUSES FOR ENCODING AND DECODING SUPERPIXEL BORDERS |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1589027A (en) * | 2004-07-29 | 2005-03-02 | 联合信源数字音视频技术（北京）有限公司 | Image boundarg pixel extending system and its realizing method |
| US20140355898A1 (en) * | 2012-01-13 | 2014-12-04 | Thomson Licensing | Method and device for encoding a block of an image and corresponding reconstructing method and device |
| CN104244007A (en) * | 2013-06-13 | 2014-12-24 | 上海天荷电子信息有限公司 | Image compression method and device based on arbitrary shape matching |
| CN104378644A (en) * | 2013-08-16 | 2015-02-25 | 上海天荷电子信息有限公司 | Fixed-width variable-length pixel sample value string matching strengthened image compression method and device |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20060123939A (en) * | 2005-05-30 | 2006-12-05 | 삼성전자주식회사 | Method and apparatus for encoding and decoding video |
| CN101291436B (en) * | 2008-06-18 | 2011-02-16 | 北京中星微电子有限公司 | Video coding/decoding method and device thereof |
-
2015
- 2015-03-18 CN CN201510118591.7A patent/CN104935945B/en active Active
- 2015-03-18 CN CN201910946605.2A patent/CN110505488B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1589027A (en) * | 2004-07-29 | 2005-03-02 | 联合信源数字音视频技术（北京）有限公司 | Image boundarg pixel extending system and its realizing method |
| US20140355898A1 (en) * | 2012-01-13 | 2014-12-04 | Thomson Licensing | Method and device for encoding a block of an image and corresponding reconstructing method and device |
| CN104244007A (en) * | 2013-06-13 | 2014-12-24 | 上海天荷电子信息有限公司 | Image compression method and device based on arbitrary shape matching |
| CN104378644A (en) * | 2013-08-16 | 2015-02-25 | 上海天荷电子信息有限公司 | Fixed-width variable-length pixel sample value string matching strengthened image compression method and device |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112055219A (en) * | 2020-08-05 | 2020-12-08 | 浙江大华技术股份有限公司 | String matching prediction method and device and computer readable storage medium |
| CN115119046A (en) * | 2022-06-02 | 2022-09-27 | 绍兴市北大信息技术科创中心 | Image coding and decoding method, device and system with reference to pixel set |
| CN115119046B (en) * | 2022-06-02 | 2024-04-16 | 绍兴市北大信息技术科创中心 | Image coding and decoding method, device and system for reference pixel set |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104935945B (en) | 2019-11-08 |
| CN110505488B (en) | 2022-01-07 |
| CN104935945A (en) | 2015-09-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106464881B (en) | The method and apparatus of block adaptive color space conversion decoding | |
| CN106797465B (en) | A kind of method and device of decoding video data | |
| CN105120271B (en) | Video coding-decoding method and device | |
| CN106797479B (en) | For the intra block duplication prediction limitation of parallel processing | |
| CN105874795B (en) | When wavefront parallel processing is activated to the rule of intra-picture prediction mode | |
| CN104041035B (en) | Lossless coding and coherent signal method for expressing for composite video | |
| CN103650496B (en) | Intra prediction pixel-based for being encoded in HEVC | |
| CN111800640B (en) | Method and device for encoding and decoding image by alternately changing direction and back-and-forth scanning string matching | |
| CN109716774A (en) | The frame mode of variable number for video coding | |
| CN108028940A (en) | The limitation of escape pixel signal values in pallet mode video coding | |
| CN105493507B (en) | Residual prediction for intra block duplication | |
| CN104704843B (en) | The instruction of video property | |
| CN109196869A (en) | Intra prediction mode based on neighbouring signalling | |
| CN108028932A (en) | Palette prediction initialization and merging for video coding | |
| CN108293112A (en) | Elastic registration tree construction in video coding | |
| CN104853211A (en) | Image compression method and apparatus employing various forms of reference pixel storage spaces | |
| CN107251557A (en) | The coding/decoding of chrominance resolution details | |
| CN107211150A (en) | Dynamic updates quality to higher chroma samples rate | |
| CN107787583A (en) | Reference picture list structure in block in replication mode | |
| CN106105207A (en) | The palette predictor signaling with haul distance code for video coding | |
| CN104935945B (en) | The image of extended reference pixel sample value collection encodes or coding/decoding method | |
| CN107690803A (en) | The adaptive constant illumination method decoded for HDR and wide color gamut video | |
| CN106170092A (en) | Fast encoding method for lossless coding | |
| CN105491376A (en) | Image encoding and decoding methods and devices | |
| CN106464922A (en) | Conformance and inoperability improvements in multi-layer video coding |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |