WO2015032350A1 - 一种使用块匹配的图像压缩方法和装置 - Google Patents

一种使用块匹配的图像压缩方法和装置 Download PDF

Info

Publication number
WO2015032350A1
WO2015032350A1 PCT/CN2014/086054 CN2014086054W WO2015032350A1 WO 2015032350 A1 WO2015032350 A1 WO 2015032350A1 CN 2014086054 W CN2014086054 W CN 2014086054W WO 2015032350 A1 WO2015032350 A1 WO 2015032350A1
Authority
WO
WIPO (PCT)
Prior art keywords
block
reconstructed
reference block
value
decoding
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.)
Ceased
Application number
PCT/CN2014/086054
Other languages
English (en)
French (fr)
Inventor
林涛
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tongji University
Original Assignee
Tongji University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Tongji University filed Critical Tongji University
Priority to US14/917,026 priority Critical patent/US20170155899A1/en
Publication of WO2015032350A1 publication Critical patent/WO2015032350A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
    • H04N19/103—Selection of coding mode or of prediction mode
    • H04N19/105—Selection of the reference unit for prediction within a chosen coding or prediction mode, e.g. adaptive choice of position and number of pixels used for prediction
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/134—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or criterion affecting or controlling the adaptive coding
    • H04N19/136—Incoming video signal characteristics or properties
    • H04N19/137—Motion inside a coding unit, e.g. average field, frame or block difference
    • H04N19/139—Analysis of motion vectors, e.g. their magnitude, direction, variance or reliability
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
    • H04N19/17—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object
    • H04N19/174—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object the region being a slice, e.g. a line of blocks or a group of blocks
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
    • H04N19/17—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object
    • H04N19/176—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object the region being a block, e.g. a macroblock
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
    • H04N19/182—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being a pixel
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
    • H04N19/503—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving temporal prediction
    • H04N19/51—Motion estimation or motion compensation
    • H04N19/537—Motion estimation other than block-based
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/42—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals characterised by implementation details or hardware specially adapted for video compression or decompression, e.g. dedicated software implementation
    • H04N19/43—Hardware specially adapted for motion estimation or compensation
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
    • H04N19/503—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving temporal prediction
    • H04N19/51—Motion estimation or motion compensation
    • H04N19/563—Motion estimation with padding, i.e. with filling of non-object values in an arbitrarily shaped picture block or region for estimation purposes

Definitions

  • the present invention relates to a digital video compression encoding and decoding system, and more particularly to a method and apparatus for encoding and decoding computer screen images and video.
  • a notable feature of computer screen images is that there are often many similar or even identical pixel patterns within the same frame of image.
  • Chinese or foreign text that often appears in computer screen images is composed of a few basic strokes, and many similar or identical strokes can be found in the same frame image.
  • Menus, icons, etc. which are common in computer screen images, also have many similar or identical patterns.
  • the intra prediction method used in the existing image and video compression technology only refers to adjacent pixel samples, and cannot improve the compression efficiency by using the similarity or the similarity in one frame image.
  • Intra-motion compensation in the prior art block matching coding is performed with blocks of fixed size (such as 4x4, 8x8, 16x16, 32x32, 64x64 pixels), but fixed size
  • the matching block must be completely in the set of reconstructed reference pixel samples, and thus cannot overlap with the matched block that has not been reconstructed in the encoding, especially when the matching block is large, the corresponding pixel samples of the matching block and the matched block.
  • the distance between them is very long, that is, the matching of distances cannot be performed, which greatly affects the efficiency of block matching coding. Therefore, it is necessary to break through the prior art, in particular to solve the problem that the matching block and the matched block cannot overlap in the existing matching coding technology, so as to greatly improve the compression effect.
  • the natural form of a digital video signal of a screen image is a sequence of images.
  • Encoding a digital video signal encodes a frame by frame image. At any one time, the image of the frame being encoded is referred to as the current encoded image.
  • decoding a compressed code stream (a bit stream, also referred to as a bit stream) of a digital video signal is decoding a compressed code stream of one frame by one frame of image. At any one moment, the image of the frame being decoded is called the current decoding. image.
  • the current encoded image or the currently decoded image is collectively referred to as the current image.
  • coding Unit when encoding one frame of image, one frame of image is divided into sub-images of several blocks of MxM pixels, which is called "Coding Unit (CU)", with CU as the basic coding unit.
  • the sub-images are encoded one by one.
  • the size of the commonly used M is 4, 8, 16, 32, 64. Therefore, encoding a video image sequence is to sequentially encode each coding unit, that is, the CU of each frame image.
  • the CU being coded is referred to as the current coded CU.
  • the CUs are sequentially decoded for each coding unit, and finally the entire video image sequence is reconstructed.
  • the CU being decoded is referred to as the currently decoded CU.
  • the current coding CU or the current decoding CU are collectively referred to as the current CU.
  • each CU in one frame of image can be different, some are 8x8, some are 64x64, and so on.
  • LCU Large Coding Unit
  • one frame of image is always first divided into "Largest Coding Unit (LCU)" having the same size and having NxN pixels, which is also called a CU with a depth of 0. Then, a CU with a depth of 0 can be divided into 4 sizes with exactly the same A CU with a depth of 1 pixel.
  • LCU Large Coding Unit
  • a CU with a depth of 1 can be further divided into 4 sizes with exactly the same A CU with a depth of 2 pixels. In this way, continue to divide, and finally reach a preset maximum depth D, that is, the corresponding CU size reaches the minimum value. until.
  • a CU having a maximum depth D is referred to as a "Smallest Coding Unit (SCU)".
  • SCU Smallest Coding Unit
  • An LCU can be divided into four CUs of depth 1 that are 32x32 pixels in size.
  • a CU of depth 1 can be divided into 4 CUs with a depth of 2 and a size of 16x16 pixels.
  • a CU of depth 2 can be divided into four CUs of depth 3 of 8 ⁇ 8 pixels, that is, SCUs with the deepest depth.
  • the CU when encoding and decoding a CU, the CU can be split into four square sub-blocks, and the four sub-blocks are separately subjected to predictive coding and decoding.
  • an order In order to perform predictive coding and decoding in an order, an order must be specified for all of the smallest sub-blocks in an LCU. In the case where the size of the LCU is 64x64 pixels and the size of the smallest sub-block is 4x4 pixels, one LCU has a total of 256 minimum sub-blocks, and the encoding and decoding specified by HEVC and the corresponding reconstruction order are as shown in FIG.
  • the basic law for discharging the order shown in Fig. 1 is as follows.
  • First (highest) hierarchical ordering divide a 64x64 pixel block (LCU) into four 32x32 pixel blocks, and the order of the four blocks is upper left, upper right, lower left, and lower right. That is, first sort all the smallest sub-blocks in the upper left block (numbered from 0 to 63), and then sort all the smallest sub-blocks in the upper right block (serial number is 64-127), and then in the lower left block. All the smallest sub-blocks are sorted (numbered from 128 to 191), and finally all the smallest sub-blocks in the lower right block are sorted (numbered from 192 to 255).
  • Second level sorting divide a 32x32 pixel block into four 16x16 pixel blocks, and the order of the four blocks is also the upper left, upper right, lower left, and lower right. That is, first sort all the smallest sub-blocks in the upper left block (order The number is 0 to 15 or 64 to 79 or 128 to 143 or 192 to 207. Then, all the smallest subblocks in the upper right block are sorted (numbered 16 to 31 or 80 to 95 or 144 to 159 or 208 to 223).
  • sort all the smallest sub-blocks in the lower left block (number is 32 ⁇ 47 or 96-111 or 160-175 or 224-239), and finally sort all the smallest sub-blocks in the lower right block ( The serial number is 48-63 or 112-127 or 176-191 or 240-255).
  • Third level sorting divide a 16x16 pixel block into four 8x8 pixel blocks, and the order of the four blocks is also the upper left, upper right, lower left, and lower right.
  • Fourth (lowest) hierarchical ordering divide a block of 8x8 pixels into four blocks of 4x4 pixels (minimum sub-block), and the order of the four smallest sub-blocks is also upper left, upper right, lower left, lower right .
  • the four smallest sub-blocks have the serial numbers 12, 13, 14 respectively. 15, 15, as shown in Figure 1.
  • the smallest sub-blocks in the upper left, upper right, lower left, and lower right blocks of each level can be sorted, and finally the sequence numbers of all 256 minimum sub-blocks shown in FIG. 1 are obtained.
  • 2x2 4 minimum sub-blocks up and down and left and right constitute one 8x8 pixel block.
  • Up and down, left and right adjacent 2x2 4 8x8 pixel blocks form a 16x16 pixel block.
  • Up and down, left and right adjacent 2x2 4 16x16 pixel blocks form a 32x32 pixel block.
  • Up and down, left and right adjacent 2x2 4 32x32 pixel blocks form a 64x64 pixel block (LCU).
  • a color pixel usually consists of three components.
  • the two most commonly used pixel color formats are the GBR color format consisting of a green component, a blue component, and a red component, and a YUV color consisting of a luma component and two chroma components.
  • the format commonly known as the YUV color format, actually includes multiple color formats, such as the YCbCr color format. Therefore, when encoding a CU, a CU can be divided into three component planes (G plane, B plane, R plane or Y plane, U plane, V plane), and the three component planes are respectively coded; The three component bundles of one pixel are combined into one 3-tuple, and the CUs composed of these 3-tuples are encoded as a whole.
  • the arrangement of the former pixel and its components is called the planar format of the image (and its CU), and the arrangement of the latter pixel and its components is called the stacked format of the image (and its CU). Format).
  • the GBR color format and the YUV color format of the pixel are both 3-component representation formats of the pixel.
  • the palette index table In addition to the 3-component representation format of pixels, another common prior art representation of pixels is the palette index table. Current format. In the palette index representation format, the value of a pixel can also be represented by the index of the palette. The palette space stores the value or approximate value of the three components of the pixel that needs to be represented. The address of the palette is called the index of the pixel stored in this address. An index can represent one component of a pixel, and an index can also represent three components of a pixel. The palette can be one or more. In the case of multiple palettes, a complete index is actually composed of the palette number and the index of the numbered palette. The index representation format of a pixel is to represent this pixel with an index.
  • the index representation format of a pixel is also referred to as an indexed color or a pseudo color representation format of a pixel in the prior art, or is often referred to directly as an indexed pixel or a pseudo pixel (pseudo pixel). ) or pixel index or index. Indexes are sometimes referred to as indices.
  • the representation of a pixel in its index representation format is also referred to as indexing or indexing.
  • CMYK presentation formats Other commonly used prior art pixel representation formats include CMYK presentation formats and grayscale representation formats.
  • the YUV color format can be subdivided into several seed formats according to whether the chroma component is downsampled: a YUV 4:4:4 pixel color format consisting of 1 Y component, 1 U component, and 1 V component.
  • the left and right adjacent pixels are composed of two Y components, one U component, and one V component in a YUV 4:2:2 pixel color format; four pixels arranged in a left and right adjacent position by 2x2 spatial position are composed of four pixels.
  • YUV4: 2:0 pixel color format consisting of Y component, 1 U component, and 1 V component.
  • a component is generally represented by a number of 8 to 16 bits.
  • the YUV4:2:2 pixel color format and the YUV4:2:0 pixel color format are all downsampled for the YUV4:4:4 pixel color format.
  • a pixel component is also referred to as a pixel sample or simply as a sample.
  • the most basic element when encoding or decoding can be one pixel, one pixel component, or one pixel index (ie, index pixel).
  • a pixel or a pixel component or an index pixel, which is the most basic element of encoding or decoding, is collectively referred to as a pixel sample, sometimes referred to as a pixel value, or simply as a sample.
  • intra block matching also known as intra motion compensation or intra block copying
  • block such as 8x8 pixel samples
  • coarser microblocks such as 4x2 pixel samples or 8x2 pixel samples or 2x4 pixel samples or 2x8 pixel samples
  • lines ie, microblocks with a height of 1 or a width of 1, such as 4x1 pixel samples or 8x1
  • Pixel samples or 1x4 pixel samples or 1x8 pixel samples or arrange pixel samples in a block into a string that is much longer than the width (such as a string with a width of 1 pixel and a length of 64 pixels) Or a string having a width of 2 pixel samples and a length of 32 pixel samples) followed by a microblock matching (also called an intra microblock) with a microblock or a variable length substring within the string as the smallest matching unit.
  • Copying or line matching (also known as bar matching or intra-line copying or intra-frame copying)
  • a coding block or a decoding block refers to coding or decoding in a frame image.
  • the coding block and the decoding block are collectively referred to as a block.
  • Blocks include, but are not limited to, commonly referred to as blocks, microblocks, lines (strips), and strings;
  • Block matching includes, but is not limited to, so-called block matching, block copying, microblock matching, microblock copying, line matching, strip matching, line copying, strip copying, string matching, string copying;
  • Matching blocks include, but are not limited to, so-called matching blocks, matching microblocks, matching lines, matching bars, matching strings;
  • the matched blocks include, but are not limited to, commonly referred to as matched blocks, matched microblocks, matched lines, matched bars, matched strings;
  • a block is an area made up of several pixel values.
  • a block may be composed of "pixels”, or may be composed of “components of pixels”, or may be composed of "index pixels”, or may be composed of a mixture of the three, or may be mixed by any two of the three. composition.
  • Block matching coding is to search within a predetermined search range among the reconstructed reference pixel sample sets and between the coded blocks (ie, the matched blocks) when encoding one coded block.
  • a matching block with the smallest matching error (referred to as an optimal matching block) is then written into the video compressed code stream by the relative position between the matched block and the optimal matching block (referred to as a motion vector, ie, MV).
  • Block matching decoding is to determine the position of the matching block in the reconstructed reference pixel sample set according to the MV parsed from the video compressed code stream when decoding the compressed code stream segment of a decoding block, and then The matching block is copied and pasted to the position of the decoded block (ie, the matched block), that is, the value of the decoded block is directly or indirectly set equal to the value of the matching block.
  • the matching block in order to completely calculate the matching error and copy and paste the entire matching block to the position of the matched block being encoded, decoded, and reconstructed, the matching block must be completely reconstructed in the reference pixel.
  • the matching block is the original complete reconstructed matching block.
  • the matching block and the matched block cannot have overlapping portions, that is, one block cannot partially match itself (referred to as partial self-matching).
  • determining the reconstructed reference pixel sample set includes:
  • All LCUs that have been encoded or decoded and reconstructed according to a predetermined encoding or decoding order usually including at least the LCU (left LCU) located on the left side of the current LCU, and the LCU (upper LCU) located on the upper side of the current LCU, located at the current LCU on the upper left side of the LCU (upper left LCU).
  • the current CU is a 16x16 pixel CU consisting of 16 smallest sub-blocks with sequence numbers 192-207.
  • its reconstructed reference pixel sample set (the portion indicated by the shaded hatching in Figure 2) includes all of the smallest sub-blocks whose sequence number is less than 192.
  • Figure 2 also illustrates the locations of the matched blocks, i.e., the current CU and the matching block.
  • the matching blocks as a whole are all in the set of reconstructed reference pixel samples.
  • the matched block does not intersect with the matching block, ie there is no overlap.
  • FIG. 3 is a second example of a current CU and its reconstructed reference pixel sample set.
  • the current CU is an 8x8 pixel CU consisting of 4 smallest sub-blocks numbered 244-247.
  • its reconstructed reference pixel sample set (the portion shaded by hatching in Figure 3) includes all of the smallest sub-blocks whose sequence number is less than 244.
  • Figure 3 also illustrates the locations of the matched blocks, i.e., the current CU and the matching block.
  • the matching blocks as a whole are all in the set of reconstructed reference pixel samples.
  • the matched block does not intersect with the matching block, ie there is no overlap.
  • FIG. 4 is a third example of a current CU and its reconstructed reference pixel sample set.
  • the current CU is a 16x16 pixel CU consisting of 16 smallest sub-blocks with sequence numbers 80-95.
  • its reconstructed reference pixel sample set (the portion shaded by hatching in Figure 4) includes all of the smallest sub-blocks whose sequence number is less than 80.
  • Figure 4 also illustrates the locations of the matched blocks, i.e., the current CU and the matching block.
  • the matching blocks as a whole are all in the set of reconstructed reference pixel samples.
  • the matched block does not intersect with the matching block, ie there is no overlap.
  • Figure 5 is a fourth example of a current CU and its reconstructed reference pixel sample set.
  • the current CU is an 8x8 pixel CU consisting of 4 smallest sub-blocks numbered 36-39.
  • its reconstructed reference pixel sample set (the portion shaded by hatching in Figure 5) includes all of the smallest sub-blocks whose sequence number is less than 36.
  • Figure 5 also illustrates the locations of the matched blocks, i.e., the current CU and the matching block.
  • the matching blocks as a whole are all in the set of reconstructed reference pixel samples.
  • the matched block does not intersect with the matching block, ie there is no overlap.
  • Figure 6 is a fifth example of a current CU and its reconstructed reference pixel sample set.
  • the current CU is an 8x8 pixel CU consisting of 4 smallest sub-blocks numbered 168-171.
  • its reconstructed reference pixel sample set (the portion shaded by hatching in Figure 6) includes all of the smallest sub-blocks whose sequence number is less than 168.
  • Figure 6 also illustrates the locations of the matched blocks, i.e., the current CU and the matching block.
  • the matching blocks as a whole are all in the set of reconstructed reference pixel samples.
  • the matched block does not intersect with the matching block, ie there is no overlap.
  • Any pixel sample in a frame of image is typically, but not limited to, a coordinate with respect to a predetermined reference point in the image (ie, the origin, usually but not limited to the point of the top left pixel sample of the image) ( X, Y) is expressed, and the value of the pixel sample P whose coordinates are (X, Y) is represented by P(X, Y).
  • the direction of increase of X is usually (but not limited to) to the right, and the direction of increase of Y is usually (but not limited to) downward.
  • (Xc, Yc) be the coordinate of the top left pixel sample of the matched block of width Nx height Ny
  • (Xr, Yr) be a matching block (must have the same width and height as the matched block)
  • the coordinates of the top left pixel sample According to the common knowledge of plane analytic geometry, the matching block as a whole is in the reconstructed reference pixel sample set, that is, all the pixel samples of the matching block are necessary for the reconstructed pixel sample.
  • the matching block and the matched block cannot have any Intersecting (ie overlapping) parts. This necessary condition is expressed by the relationship between coordinates, width and height:
  • the operation of copying and pasting the matching block to the position of the matched block that is, directly or indirectly assigning the values of all pixel samples of the matching block to
  • the operation of the matched block can be done with (but not limited to) any one or a combination of the following assignment statements:
  • the matching block as a whole is within the reconstructed reference pixel sample set, the matching block and the matched block must not intersect, and the above eight kinds of assignment statements are completely equivalent.
  • the assignment statement of the original complete reconstructed matching bar is (but is not limited to) any one or combination of the following assignment statements:
  • the assignment statement of the original complete reconstructed matching bar is (but is not limited to) any one or combination of the following assignment statements:
  • the matching block is a matching string (ie, a matching block whose width is a small non-independent parameter W and the length is an independently variable parameter L)
  • the coordinates (X, Y) of the pixel sample become a one-dimensional
  • the value of the address K and the pixel sample value P of the address K is represented by P(K).
  • Kc be the address of the first pixel sample of the matched string of length L
  • Kr is the address of the first pixel sample of the matching string (which must have exactly the same length as the matched string).
  • Original finish The assignment statement of the entire reconstructed matching string is (but is not limited to) any one or combination of the following assignment statements:
  • Synonyms of matching blocks include, but are not limited to, reference blocks, prediction blocks.
  • synonyms of the matched block include, but are not limited to, the original block, the current block, and the current coded block.
  • synonyms of the matched block include, but are not limited to, a reconstructed block, a reconstructed block, a current block, and a current decoded block.
  • Synonyms of matching bars include, but are not limited to, reference bars, prediction bars.
  • synonyms of the matched bar include but are not limited to the original bar, the current bar, and the current encoding bar.
  • synonyms of the matched bar include, but are not limited to, a reconstruction bar, a reconstruction bar, a current bar, and a current decoding bar.
  • Synonyms of matching strings include, but are not limited to, reference strings, prediction strings.
  • synonyms of the matched string include, but are not limited to, the original string, the current string, and the current encoded string.
  • Synonyms of the matched string include, but are not limited to, a reconstructed string, a reconstructed string, a current string, and a current decoded string during decoding and reconstruction.
  • synonyms for matching coding include, but are not limited to, copy coding and generalized predictive coding
  • synonyms for matching decoding include, but are not limited to, copy decoding and generalized predictive decoding
  • predictive coding is an abbreviation for generalized predictive coding
  • predictive decoding is an abbreviation for generalized predictive decoding. Therefore, block matching coding, microblock matching coding, line matching coding, bar matching coding, string matching coding, and arbitrary shape matching coding synonyms, respectively, include but are not limited to block copy coding and block prediction coding, microblock copy coding, and microblocks, respectively.
  • Predictive coding, line copy coding and line prediction coding, strip copy coding and bar prediction coding, string copy coding and string prediction coding, arbitrary shape copy coding and arbitrary shape predictive coding, likewise, block matching decoding, microblock matching decoding, line matching Synonyms of decoding, strip matching decoding, string matching decoding, and arbitrary shape matching decoding include, but are not limited to, block copy decoding and block prediction decoding, microblock copy decoding and microblock prediction decoding, line copy decoding and line prediction decoding, and strip copy, respectively.
  • copy coding includes, but is not limited to, block copy coding, intra block copy coding, Micro-block copy coding, intra-frame micro-block copy coding, line copy coding, intra-line line copy coding, strip copy coding, intra-frame copy coding, string copy coding, intra-frame copy coding, arbitrary shape copy coding, intra-frame arbitrary Shape copy coding, including but not limited to block copy decoding, intra block copy decoding, microblock copy decoding, intra block copy decoding, line copy decoding, intra line copy decoding, strip copy decoding, intra strip copy Decoding, string copy decoding, intra-frame copy decoding, arbitrary shape copy decoding, intra-frame arbitrary shape copy decoding.
  • predictive coding includes, but is not limited to, block predictive coding, intra block predictive coding, microblock predictive coding, intra microblock predictive coding, line predictive coding, intraframe line predictive coding, and strip prediction. Coding, intra-frame predictive coding, intra-predictive coding, intra-frame predictive coding, arbitrary shape predictive coding, intra-frame arbitrary shape predictive coding, predictive decoding including but not limited to block predictive decoding, intra block predictive decoding, microblock predictive decoding Intra-frame microblock prediction decoding, line prediction decoding, intra-line prediction decoding, strip prediction decoding, intra-frame prediction decoding, string prediction decoding, intra-frame prediction decoding, arbitrary shape prediction decoding, intra-frame arbitrary shape prediction decoding.
  • the matching block as a whole is limited to the set of reconstructed reference pixel samples, the matched block and the matching block may not intersect, and the matching block of a close distance that may exist in a large amount in the image cannot be effectively found.
  • the coding efficiency of such images and patterns is very low.
  • the present invention provides that a matching block only needs to be partially among the reconstructed reference pixel sample sets, in particular, the matched block and the matching block can intersect.
  • a method and apparatus for image encoding and decoding ie, having overlapping portions, also referred to as partial self-matching.
  • a matching block that is only partially, but not completely, among the reconstructed reference pixel sample sets is referred to as a partially reconstructed matching block.
  • a matching block that is completely within the reconstructed reference pixel sample set is referred to as the original complete reconstructed matching block.
  • the matching blocks corresponding to the current block do not necessarily need to be all located in the reconstructed reference pixel sample set, as long as at least one pixel sample is located in the reconstructed reference pixel sample set. Thus, the overlapped portion of the matched block and the matching block is allowed.
  • Two 16x16 pixel CUs (matched blocks) and three 8x8 pixel CUs (matched blocks) are illustrated in FIG.
  • Two 16x16 pixel CUs (matched blocks) and five 8x8 pixel CUs (matched blocks) are illustrated in FIG.
  • Their corresponding matching blocks are all partially reconstructed matching blocks, that is, only a part (the part indicated by the shaded hatching in the figure) is among the reconstructed reference pixel sample sets.
  • the portion of a matching block that is among the reconstructed reference pixel sample sets is simply referred to as the reconstructed portion of the matching block, and the remainder is simply referred to as the unreconstructed portion of the matching block.
  • the reconstructed part of a partially reconstructed matching block (the part represented by the shaded hatching) has four cases:
  • the reconstructed portion of a partially reconstructed matching block is the upper portion of the matching block
  • the reconstructed portion of a partially reconstructed matching block is the left portion of the matching block
  • the reconstructed portion of a partially reconstructed matching block is the upper left portion of the matching block
  • the reconstructed portion of a partially reconstructed matching block is the portion of the matching block from which the lower right portion is removed, that is, the unreconstructed portion is the lower right portion.
  • the reconstructed portion has a ⁇ shape and is composed of two left and right rectangles, which are called a left rectangle and a right rectangle, respectively.
  • the block is a line (that is, a block whose height or width is 1 sample)
  • the above four cases become two cases, that is, case 1) and case 2), and case 3) and case 4) cannot exist. .
  • the block is a string (ie, the pixel samples in the block are arranged into a string whose length is much larger than the width), the above four cases become one case, that is, the reconstructed part of a partially reconstructed matching string is The front of the matching string.
  • the matching block and the matched block shown in FIGS. 7 and 8 may be a matching block and a matched block in a stacked format, or may be a matching block and a matched block of one component (sample) of a planar format. Therefore, the method and apparatus of the present invention can be applied to encoding, decoding, and reconstructing pixels of LCUs and CUs in a stacked format, and can also be applied to encoding pixel values of a plane of LCUs and CUs in a planar format. , decoding, and refactoring.
  • the most basic characteristic feature is the candidate matching block used as a reference (ie, located within a predetermined search range) when performing optimal block matching of block matching coding on the current coding block.
  • All matching blocks that may be the optimal matching block do not necessarily need to be all located in the set of reconstructed reference pixel samples, ie may contain unreconstructed parts.
  • some pixel samples of the reconstructed portion of the matching block must first be used.
  • the matching block is copied and pasted in the block matching decoding of the current decoding block (ie, the value of the decoding block is directly or indirectly set equal to the matching block.
  • the value of the matching block does not necessarily need to be all located in the set of reconstructed reference pixel samples, that is, may contain unreconstructed portions.
  • some pixel samples of the reconstructed part of the matching block or some adjacent pixel samples of the reconstructed part or other pixel samples must first be used to fill the completion. The unreconstructed portion of the matching block is then copied and pasted to the location of the current CU (ie, the matched block).
  • the matching block may contain unreconstructed portions when performing block matching encoding and decoding.
  • an unreconstructed portion that is, a partially reconstructed matching block
  • some pixel samples of the reconstructed portion of the matching block or some pixel samples adjacent to the reconstructed portion must be used first or other Pixel samples are filled to fill the unreconstructed portion of the matching block, ie, some pixel samples of the reconstructed portion of the matching block or certain pixel samples or other pixel samples of the reconstructed portion
  • the value of the value is directly or indirectly assigned to the unreconstructed portion of the matching block, followed by other subsequent operations of encoding and decoding.
  • the completion (ie, directly or indirectly assigned) the unreconstructed portion of the matching block may be filled in some suitable manner.
  • the block is a line (that is, a block whose height or width is 1 sample)
  • the matching block becomes a matching bar
  • the four cases of the reconstructed portion also become two cases.
  • the block is a string (that is, the pixel samples in the block are arranged into a string whose length is much larger than the width)
  • the matching block becomes a matching string, and the four cases of the reconstructed portion also become one case.
  • the matched block or the reconstructed portion of the matching strip having a width of 1 is the upper portion of the matching block or the matching strip of width 1 (case 1 of FIG. 9)
  • a number of rows of pixel sample fill completions (ie, directly or indirectly assigned to) the lower unreconstructed portion can be taken from the upper reconstructed portion.
  • the rows taken from the top can be arbitrary but pre-agreed (to ensure consistency and correctness of the encoder and decoder), such as the topmost rows in the reconstructed section, or the most reconstructed sections.
  • a plurality of rows of pixel samples may be taken from the upper reconstructed portion multiple times, and the padding completion (ie, directly or indirectly assigned) the lower unreconstructed portion may be repeated.
  • the plurality of rows taken from the upper reconstructed portion may be all rows of the upper reconstructed portion or a portion of the upper reconstructed portion. The rules of how to take must be agreed in advance to ensure the consistency and correctness of the encoder and decoder.
  • the fill completion method is to copy part or all of the pixel samples from the reconstructed portion of the matching block or the matching strip of width 1, and paste the topping block from top to bottom (ie, in the vertical direction) or
  • the unreconstructed portion of the matching strip of width 1, that is, the value of some or all of the pixel samples of the reconstructed portion of the matching block or the matching strip of width 1 is directly or indirectly from top to bottom (ie, in the vertical direction). Assigned to the unblocked portion of the matching block or matching strip of width 1.
  • the matched block or the reconstructed portion of the matching strip having the height of 1 is the left portion of the matching block or the matching strip of height 1 (case 2 of FIG. 9)
  • a plurality of columns of pixel sample fill completions ie, directly or indirectly assigned to the right unreconstructed portion may be taken from the reconstructed portion of the left portion.
  • the columns taken from the left can be any number of columns that are pre-agreed (to ensure consistency and correctness of the encoder and decoder), such as the leftmost columns in the reconstructed portion, or in the reconstructed portion Several columns on the far right.
  • a plurality of columns of pixel samples may be fetched from the left reconstructed portion multiple times, and the padding completion (ie, directly or indirectly assigned) the right unreconstructed portion may be repeated.
  • the plurality of columns taken out from the reconstructed portion of the left portion may be all columns of the reconstructed portion of the left portion, or may be Is a part of the column in the left reconstructed part. The rules of how to take must be agreed in advance to ensure the consistency and correctness of the encoder and decoder.
  • the fill completion method is to copy part or all of the pixel samples from the reconstructed part of the matching block or the matching strip of height 1, and paste the filled matching block from left to right (ie, horizontally) or
  • the unreconstructed portion of the matching strip of height 1 is the value of some or all of the pixel samples of the reconstructed portion of the matching block or the matching strip of height 1 from left to right (ie, horizontally) or Indirectly assigned to the unblocked portion of the matching block or matching strip of height 1.
  • the process of filling completion can be performed in two steps.
  • one or more rows of reconstructed pixel sample fill completions (ie, directly or indirectly assigned to) the lower left unreconstructed portion are taken one or more times from the upper left reconstructed portion.
  • a number of columns of pixel sample fill completions (ie, directly or indirectly assigned to) the right unreconstructed portion are taken one or more times from the left reconstructed and padded completion (ie, assigned) portions. It is also possible to use another equivalent method.
  • the first step is to take a number of columns of reconstructed pixel sample fill completions (ie, directly or indirectly assigned) from the upper left reconstructed portion one or more times to the upper right unreconstructed portion.
  • the second step one or more rows of pixel sample fill completions (ie, directly or indirectly assigned to the lower unreconstructed portion) are taken from the upper reconstructed and filled completion (ie, assigned) portions.
  • the fill completion method is to copy part of the pixel samples from the reconstructed part of the matching block, first filling from top to bottom (ie, in the vertical direction) and then from left to right (ie, horizontally).
  • Matching the unreconstructed portion of the block or equivalently, first from left to right (ie in the horizontal direction) and then from top to bottom (ie in the vertical direction) to fill the unreconstructed portion of the matching block, ie the matching block
  • the value of some or all of the pixel samples of the reconstructed portion is directly or indirectly assigned to the unreconstructed portion of the matching block from top to bottom (ie, in the vertical direction) and then from left to right (ie, in the horizontal direction).
  • the unreconstructed portion of the matching block is directly or indirectly assigned from left to right (ie, in the horizontal direction) and then from top to bottom (ie, in the vertical direction).
  • the reconstructed portion of the matching block is the portion of the matching block where the lower right portion is removed, that is, the unreconstructed portion is the lower right portion (case 4 of Fig. 10), and the reconstructed portion has a ⁇ shape, and two rectangles are left and right.
  • the composition is called the left rectangle and the right rectangle, respectively.
  • the general way of filling completion ie, assignment
  • the extracted ⁇ -shaped pixel samples can be degenerated into a rectangle located within the left rectangle for filling completion (ie, directly or indirectly assigned) to the lower right unreconstructed portion.
  • Another special case of the general approach is that the extracted ⁇ -shaped pixel samples can be degenerated into a rectangle located within the right rectangle for filling completion (ie, directly or indirectly assigned) to the lower right unreconstructed portion.
  • the fill completion method is to copy part of the pixel samples from the reconstructed part of the matching block, from top left to bottom right (ie in the 45° direction) or from left to right (ie in the horizontal direction) or from Up-to-down (ie, in the vertical direction) pasting the unreconstructed portion of the matching block, that is, the value of some or all of the pixel samples of the reconstructed portion of the matching block from top left to bottom right (ie, in the 45° direction) Or directly or indirectly assigned to the unreconstructed portion of the matching block from left to right (ie in the horizontal direction) or from top to bottom (ie in the vertical direction).
  • the reconstructed portion of a partially reconstructed matching string is usually the front of the matching string. If the reconstructed portion is larger than the unreconstructed portion, several pixel sample fill completions can be taken from the previously reconstructed portion ( That is, directly or indirectly assigned to the rear unreconstructed part. Several pixel samples taken from the front can be arbitrary but pre-agreed (to ensure encoder and decoding) Several pixel samples of consistency and correctness, such as the first few of the reconstructed parts, or the last of the reconstructed parts.
  • the padding completion method is to copy part or all of the pixel samples from the reconstructed part of the matching string, and paste the unreconstructed part of the matching string from front to back, that is, the reconstructed part of the matching string.
  • the value of some or all of the pixel samples is directly or indirectly assigned to the unreconstructed portion of the matching string from front to back.
  • a large class of partially reconstructed matching blocks is a matching block with matching blocks that are matched in position.
  • This particular class of specially reconstructed matching blocks is referred to as a matching block that intersects the matched block, and is also referred to as a matching block that partially matches itself, referred to as a partially self-matching matching block or a partially self-matching block.
  • a partially self-matching matching block or a partially self-matching block In the example of the five pairs of matching blocks and matched blocks shown in FIG. 7, four pairs are partial self-matching examples, and one pair is not partially self-matching. Five of the six pairs of matching blocks and matched blocks shown in FIG. 8 are partially self-matching examples, and one pair is not partially self-matching.
  • (Xc, Yc) be the coordinate of the top left pixel sample of the matched block with the width Nx height Ny
  • (Xr, Yr) be a matching block (there must be the same width and height as the matched block)
  • the matching block that intersects the matched block is equivalent to a matching block that satisfies the following relationship:
  • Xr ⁇ Xc and Yr ⁇ Yc will never be established at the same time.
  • the coordinates (Xc, Yc) and (Xr, Yr) have the following relationships with the above four cases, and can be used as sufficient conditions for judging which case the reconstructed portion of the matching block belongs to:
  • FIG. 1 A schematic flow chart of the encoding method of the present invention is shown in FIG.
  • the encoding method of the present invention includes, but is not limited to, part or all of the following steps:
  • step 3 1) inputting the position and size of a currently matched block while inputting the position and size of a plurality of candidate matching blocks within a predetermined search range for the matched block and the matching blocks of the plurality of candidates For each matching block, the following steps are performed: determining whether the matching block is in the reconstructed reference pixel sample temporary storage area from the position and size of the matched block and the position and size of the matching block, if , performing the next step in sequence, otherwise, skip to step 3);
  • Step 5 1) taking the original complete reconstructed matching block from the reconstructed reference pixel sample temporary storage area, and selecting the original complete reconstructed matching block as the input matching block of the moving vector search in the subsequent step 5); Step 5);
  • FIG. 1 A schematic flowchart of the decoding method of the present invention is shown in FIG.
  • the decoding method of the present invention includes, but is not limited to, part or all of the following steps:
  • step 3 1) determining, according to the motion vector obtained from the input video compression code stream (ie, the optimal motion vector searched by the encoder), the matching block corresponding to the matched block at the current decoding position (ie, the optimal search by the encoder) Whether the matching block is in the reconstructed reference pixel sample temporary storage area, if yes, the next step is performed sequentially, otherwise, skip to step 3);
  • FIG. 1 A schematic diagram of the encoding device of the present invention is shown in FIG.
  • the encoding device includes but is not limited to some or all of the following modules:
  • the module input includes the position and size of a currently matched block, and the module input further includes a candidate within a predetermined search range. Matching the position and size of the block, the module determines, according to the position and size of the matched block and the position and size of the matching block, whether the matching block is entirely in the reconstructed reference pixel sample temporary storage module;
  • Reconstructing the reference pixel sample temporary storage module all previously reconstructed pixel samples temporarily stored to the position of the matched block in the current encoding, used as reference pixel samples of the matched block in the current encoding (ie, candidates) The pixel value of the matching block);
  • a motion vector search module the function of the module is for an input current matched block, in the reconstructed reference pixel sample set, ie, the reconstructed reference pixel sample temporary storage module, according to a predetermined evaluation criterion, Within a predetermined search range, the search results in an optimal matching block and a corresponding optimal motion vector;
  • the optimal matching block may be the original complete reconstructed matching block from the reconstructed reference pixel sample temporary storage module. , may also be a matching block from the padding completion that fills the unreconstructed partial module with the reconstructed portion;
  • Remaining coding operation module performing a pair of matched blocks, the current coding CU where the matched block is located, and the current coding The remaining encoding operations of the code LCU.
  • FIG. 1 A schematic diagram of the decoding apparatus of the present invention is shown in FIG.
  • the decoding device includes but is not limited to part or all of the following modules:
  • the module input includes but is not limited to the motion vector obtained from the input video compressed code stream, and the module decodes from the motion vector and the current decoding The position and size of the matched block in the medium are determined whether the matching block corresponding to the motion vector is in the reconstructed reference pixel sample temporary storage module;
  • Reconstructing the reference pixel sample temporary storage module all previously reconstructed pixel samples temporarily stored to the position of the matched block in the current decoding, used as reference pixel samples of the matched block in the current decoding (ie, The pixel sample value of the matching block);
  • the function of the module is from the reconstructed reference pixel sample temporary storage module Copying the original complete reconstructed matching block at the position specified by the motion vector, or copying the padding completion matching block generated by the module 3), and pasting the original complete reconstructed matching block or the padding matching block
  • the value of the matched block in the current decoding that is, the value of the original complete reconstructed matching block or the padding complemented matching block is directly or indirectly assigned to the matched block in the current decoding
  • the remaining decoding operation module performing, for the matched block in the current decoding, the remaining decoding operations of the current decoding CU and the currently decoded LCU where the matched block is located in the current decoding.
  • the decoding method of the present invention may equally include, but is not limited to, some or all of the steps having the same final technical effects as follows:
  • step 3 1) determining, according to the motion vector obtained from the input video compression code stream (ie, the optimal motion vector searched by the encoder), the matching block corresponding to the matched block at the current decoding position (ie, the optimal search by the encoder) Whether the matching block is in the reconstructed reference pixel sample temporary storage area, if yes, the next step is performed sequentially, otherwise, skip to step 3);
  • step 5 copying the original complete reconstructed matching block from the reconstructed reference pixel sample temporary storage area, and pasting the original complete reconstructed matching block to the position of the matched block, that is, the original
  • the value of the complete reconstructed matching block is directly or indirectly assigned to the matched block; skip to step 5);
  • the decoding device may equivalently include, but is not limited to, some or all of the following modules having the same final technical effect:
  • the module input includes but is not limited to the motion vector obtained from the input video compressed code stream, and the module decodes from the motion vector and the current decoding The position and size of the matched block in the medium are determined whether the matching block corresponding to the motion vector is in the reconstructed reference pixel sample temporary storage module;
  • the function of the module is to reconstruct the reference pixel from the reference At a position specified by the motion vector in the value buffer module, the reconstructed pixel sample in the partially reconstructed matching block is copied, and the reconstructed pixel sample is first pasted to the corresponding of the matched block.
  • Positioning then pasting all or part of the reconstructed pixel samples and/or adjacent portion of the reconstructed pixel samples of the partially reconstructed matching block directly into the matched block that has not been pasted into Positioning, filling the entire matched block, that is, first directly or indirectly assigning the value of the reconstructed pixel sample to the corresponding portion of the matched block, and then all or part of the reconstructed pixel sample And/or the value of the adjacent partially reconstructed pixel sample of the partially reconstructed matching block is directly or indirectly assigned to the portion of the matched block that has not been assigned, completing all assignments of the entire matched block;
  • the remaining decoding operation module performing, for the matched block in the current decoding, the remaining decoding operations of the current decoding CU and the currently decoded LCU where the matched block is located in the current decoding.
  • the partially reconstructed matching block includes, but is not limited to, a matching block that satisfies
  • a block, the corresponding matched block is referred to as a partially self-matching matched block. This is the most common part of a refactored matching block.
  • a method of assigning the value of the pixel sample value of the matching block to the matched block is to assign all the values of the reconstructed portion of the matching block first or directly to the matched block. a portion of the corresponding position, if there is a portion of the matched block that is not assigned, then all the value repeats of the reconstructed portion of the matching block are directly or indirectly assigned to the unassigned portion of the matched block until All the matched blocks are assigned.
  • Such a method of assigning all of the values of the reconstructed portion of the matching block to the matched block directly or indirectly is referred to as a partial self-matching full assignment method.
  • Partial self-matching has a special case where the unreconstructed parts of a partially reconstructed matching block are all completely in the corresponding matched block. This special case is called a ⁇ -shaped partial self-matching.
  • the ⁇ -shaped partial self-matching is equivalent to: Xc - Nx ⁇ Xr ⁇ Xc and Yc - Ny ⁇ Yr ⁇ Yc.
  • the ⁇ -shaped partial self-matching full assignment method can be completed with (but not limited to) any one or a combination of the following assignment statements:
  • the ⁇ -shaped part of the partially reconstructed matching block is in the form of the assignment statement 1, 2, 3, 4 of the fully-matched fully-assigned method and the assignment statement 1, 2, 5, 7 of the original complete reconstructed matching block.
  • the essential difference between the two is that the scope of application of the two is completely different: in the case of the original complete reconstructed matching block, the relation
  • a special type of partially reconstructed matching block is: a matching block that satisfies Xc-Nx ⁇ Xr ⁇ Xc and Yc-Ny ⁇ Yr ⁇ Yc, that is, a matching block that intersects the matched block and the reconstructed portion has a strict ⁇ shape. .
  • Another special type of partially reconstructed matching block is: a matching block that satisfies Xc-Nx ⁇ Xr ⁇ Xc and Yc-Ny ⁇ Yr ⁇ Yc but two of the equal signs cannot be simultaneously established, that is, intersects with the matched block and has
  • the reconstructed portion is a matching block of a degradable ⁇ shape.
  • a partially reconstructed matching block which is a matching block satisfying Xc-Nx ⁇ Xr ⁇ Xc-Nx/2 and Yc-Ny ⁇ Yr ⁇ Yc-Ny/2, that is, a reconstructed portion intersecting the matched block Greater than three-quarters of the matching blocks of the matching block.
  • some matching blocks are used.
  • a rectangle consisting of reconstructed pixel samples adjacent to the left side of the unreconstructed portion is filled with completion, and some matching blocks are used with the upper side adjacent to the unreconstructed portion.
  • a rectangle consisting of pixel samples is constructed to fill the completion, and both may allow some or even all of the reconstructed pixel samples of the rectangle to be outside of the partially reconstructed matching block.
  • the manner of constructing or padding completion ie, directly or indirectly assigning
  • an unreconstructed portion of a partially reconstructed matching block includes, but is not limited to, one of the following ways or Its combination:
  • All or part of the value of the unreconstructed portion is directly or indirectly set equal to the value of some or all of the reconstructed reference pixel samples within the reconstructed portion of the reference block
  • All or part of the value of the unreconstructed portion is directly or indirectly set to be equal to a value equal to a part or all of the reconstructed reference pixel samples in the reconstructed portion of the reference block in a predetermined manner.
  • All or part of the value of the unreconstructed portion is directly or indirectly set to a value equal to the reconstructed reference pixel sample outside but adjacent to the reconstructed portion of the reference block
  • All or part of the values of the unreconstructed portion are directly or indirectly set to be equal to the value of the reconstructed reference pixel samples outside the reconstructed portion of the reference block but adjacent to each other in a predetermined manner.
  • All or part of the value of the unreconstructed portion is directly or indirectly set equal to a predetermined value.
  • a method of assigning a portion of a decoding block corresponding to an unreconstructed portion of a reference block includes, but is not limited to, one or a combination of the following:
  • the predetermined value is directly or indirectly assigned to a portion of the decoded block that corresponds to the unreconstructed portion of the reference block.
  • padding completion i.e., directly or indirectly assigning
  • all and each of the frames of the frame has reconstructed the unreconstructed portion of the matching block, including but not Limited to implementation with extended reconstructed reference pixel sample sets.
  • the manner in which the set of reconstructed reference pixel samples is extended includes, but is not limited to, one or a combination of the following:
  • the value of the expanded portion of the pixel sample is directly or indirectly set to a value equal to a portion of the reconstructed reference pixel sample within the reconstructed reference pixel sample set;
  • the value of the expanded portion of the pixel sample is directly or indirectly set equal to a value of a portion of the reconstructed reference pixel sample within the reconstructed reference pixel sample set that is adjacent to the extended portion;
  • the value of the pixel sample of the extended portion is directly or indirectly set equal to a value obtained by extrapolating the value of a part of the reconstructed reference pixel sample in the reconstructed reference pixel sample set in a predetermined manner.
  • the value of the expanded portion of the pixel sample is directly or indirectly set equal to the value of a portion of the reconstructed reference pixel sample adjacent to the extended portion within the reconstructed reference pixel sample set in a predetermined manner.
  • the value of the pixel value of the extended portion is directly or indirectly set equal to a predetermined value
  • Extensions can be dynamic, constantly updated, or static.
  • the extension is large enough to contain a possible unreconstructed portion of any one of the possible partially reconstructed matching blocks.
  • the value of the unreconstructed portion of any partially reconstructed matching block is automatically set to a value equal to the pixel value of the extended portion.
  • the reconstructed part of the matching block is firstly used. Directly or indirectly assigning to a corresponding portion of the matched block, and then directly or indirectly assigning a value of the pixel sample of the matching block within the extended portion to a portion of the matched block that has not been assigned The two operations may actually be merged: assigning the matching block (including the reconstructed portion and the pixel sample portion within the extended portion) directly or indirectly to the matched block
  • the matching block is a matching string
  • the assignment statement of the fully-assigned self-matching method is
  • Figure 16 is a schematic illustration of a self-matching full assignment method for the left and right portions of the present invention.
  • Both the matching bar and the matched bar have 8 pixel samples.
  • the reconstructed portion of the partially reconstructed matching strip has 5 pixel samples, and the matched strip is a periodic repetition of the 5 pixel samples.
  • the reconstructed portion of the partially reconstructed matching strip has 3 pixel samples, and the matched strip is a periodic repetition of the 3 pixel samples.
  • the reconstructed portion of the partially reconstructed matching strip has 1 pixel sample, and the matched strip is a periodic repetition of the 1 pixel sample.
  • Figure 1 is an example of the smallest sub-block sorting number in the latest international video compression standard HEVC.
  • FIG. 2 is an example of a current CU and its reconstructed reference pixel sample set and the matching block and matched block positions in the prior art.
  • FIG. 3 is a second example of a current CU and its reconstructed reference pixel sample set and the matching block and matched block positions in the prior art.
  • FIG. 4 is a third example of a current CU and its reconstructed reference pixel sample set and the matching block and matched block positions in the prior art.
  • FIG. 5 is a fourth example of a current CU and its reconstructed reference pixel sample set and the matching block and matched block positions in the prior art.
  • FIG. 6 is a fifth example of a current CU and its reconstructed reference pixel sample set and the matching block and matched block positions in the prior art.
  • FIG. 7 is a view showing four examples in which only a part of the matching block does not fall within the set of reconstructed reference pixel samples in the present invention.
  • Figure 8 is a view showing six examples in which only a part of the matching block does not fall within the set of reconstructed reference pixel samples in the present invention.
  • Figure 9 is a schematic diagram of the present invention for filling a non-reconstructed portion with a reconstructed portion of a matching block (the reconstructed portion is the upper and left portions of the matching block)
  • Figure 10 is a diagram showing the filling of the reconstructed unreconstructed portion with the reconstructed portion of the matching block in the present invention (the reconstructed portion is the upper left portion of the matching block and the portion where the lower right portion is removed)
  • FIG. 11 is a schematic flow chart of an encoding method of the present invention
  • Figure 13 is a block diagram showing the composition of the encoding apparatus of the present invention.
  • Figure 14 is a block diagram showing the composition of a decoding apparatus of the present invention
  • Figure 15 is a schematic illustration of the use of reconstructed pixel samples (possibly outside the matching block) to fill in the reconstructed unreconstructed portion of the present invention.
  • 16 is a schematic diagram of a self-matching full assignment method for the left and right portions in the present invention.

Landscapes

  • Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Compression Or Coding Systems Of Tv Signals (AREA)

Abstract

本发明提供一种图像压缩方法和装置。对编码块进行编码时，在已重构参考像素样值集之中，按照预定的评估准则，搜索得到一个或多个最优匹配块（参考块或预测块）。本发明的图像压缩方法和装置的特征是匹配块（参考块或预测块）与被匹配块（编码块或解码块或重构块）可有重叠部分，匹配块不一定全部都在已重构参考像素样值集之中，即仅匹配块的一部分在已重构参考像素样值集之中。因此，在编码器搜索匹配块以及编码器和解码器用匹配块来复原被匹配块时，必须补全所述匹配块中不在已重构参考像素样值集之中的那部分的像素样值。

Description

一种使用块匹配的图像压缩方法和装置 技术领域
本发明涉及一种数字视频压缩编码及解码系统，特别是计算机屏幕图像和视频的编码及解码的方法和装置。
背景技术
随着以远程桌面为典型表现形式的新一代云计算与信息处理模式及平台的发展和普及，多台计算机之间、计算机主机与智能电视、智能手机、平板电脑等其他数字设备之间及各种各类数字设备之间的互联已经成为现实并日趋成为一种主流趋势。这使得服务器端(云)到用户端的实时屏幕传输成为当前的迫切需求。由于需要传输的屏幕视频数据量很大，以平板电脑2048x1536像素分辨率且60帧/秒刷新率的24位真彩色屏幕图像为例，需要传输的数据达每秒2048x1536x60x24＝4320兆比特，如此多的数据要想在现实的网络条件下实现实时传输是不可能的，因此对于计算机屏幕图像的有效的数据压缩必不可少。
充分利用计算机屏幕图像的特点，对计算机屏幕图像进行超高效率的压缩，也是正在制定中的最新国际视频压缩标准HEVC(High Efficiency Video Coding)和其他若干国际标准、国内标准、行业标准的一个主要目标。
计算机屏幕图像的一个显著特点是同一帧图像内通常会有很多相似甚至完全相同的像素图样(pixel pattern)。例如，计算机屏幕图像中常出现的中文或外文文字，都是由少数几种基本笔划所构成，同一帧图像内可以找到很多相似或相同的笔划。计算机屏幕图像中常见的菜单、图标等，也具有很多相似或相同的图样。现有的图像和视频压缩技术中采用的帧内预测(intra prediction)方式，仅参考相邻的像素样值，无法利用一帧图像中的相似性或相同性来提高压缩效率。现有技术中的帧内运动补偿(intra motion compensation)方式，用几种固定大小(如4x4，8x8，16x16，32x32，64x64像素)的块来进行块匹配(block matching)编码，但固定大小的匹配块必须完全在已重构参考像素样值集之中，因而与正在编码中尚未重构的被匹配块不能重叠，特别当匹配块比较大时，匹配块与被匹配块的对应像素样值之间距离就很远，也就是无法进行距离相近的匹配，大大影响块匹配编码的效率。因此，必须突破现有技术，特别是解决现有匹配编码技术中匹配块与被匹配块不能重叠的难题，以大幅度提高压缩效果。
屏幕图像的数字视频信号的自然形式是图像的序列。一帧图像通常是由若干像素组成的矩形区域，如果一个数字视频信号每秒有50帧图像，那么一段30分钟的数字视频信号就是一个由30x60x50＝90000帧图像组成的视频图像序列，有时也简称为视频序列或序列。对数字视频信号进行编码就是对一帧一帧图像进行编码。在任一时刻，正在编码中的那一帧图像称为当前编码图像。同样，对数字视频信号的压缩码流(码流也称为比特流)进行解码就是对一帧一帧图像的压缩码流进行解码。在任一时刻，正在解码中的那一帧图像称为当前解码 图像。当前编码图像或当前解码图像都统称为当前图像。
最新国际视频压缩标准HEVC中，对一帧图像进行编码时，把一帧图像划分成若干块MxM像素的子图像，称为“编码单元(Coding Unit简称CU)”，以CU为基本编码单位，对子图像一块一块进行编码。常用的M的大小是4，8，16，32，64。因此，对一个视频图像序列进行编码就是对各帧图像的各个编码单元即CU依次进行编码。在任一时刻，正在编码中的CU称为当前编码CU。同样，解码时也是对各个编码单元即CU依次进行解码，最终重构出整个视频图像序列。在任一时刻，正在解码中的CU称为当前解码CU。当前编码CU或当前解码CU都统称为当前CU。
为适应一帧图像内各部分图像内容与性质的不同，有针对性地进行最有效的编码，一帧图像内各CU的大小可以是不同的，有的是8x8，有的是64x64，等等。为了使不同大小的CU能够无缝拼接起来，一帧图像总是先划分成大小完全相同具有NxN像素的“最大编码单元(Largest Coding Unit简称LCU)”，也称为深度为0的CU。然后，一个深度为0的CU可划分成4个大小完全相同具有
Figure PCTCN2014086054-appb-000001
像素的深度为1的CU。一个深度为1的CU也可再进一步划分成4个大小完全相同具有
Figure PCTCN2014086054-appb-000002
像素的深度为2的CU。如此，继续划分下去，最后可以达到一个预先设定的最大深度D，也就是相应的CU的大小达到最小值
Figure PCTCN2014086054-appb-000003
为止。具有最大深度D的CU称为“最小编码单元(Smallest Coding Unit简称SCU)”。在最常用的N＝64，D＝3的情形，一帧图像划分成大小为64x64像素的LCU即深度为0的CU。一个LCU可划分成4个大小为32x32像素的深度为1的CU。一个深度为1的CU可划分成4个大小为16x16像素的深度为2的CU。一个深度为2的CU可划分成4个大小为8x8像素的深度为3的CU即深度最大的SCU。在HEVC中，对一个CU进行编码和解码时，还可以把这个CU分裂(split)成4个正方形的子块(sub-block)，对这4个子块分别进行预测编码和解码。为了有序进行预测编码和解码，必须对一个LCU中的所有最小子块规定一个次序(order)。在LCU的大小为64x64像素而最小子块的大小为4x4像素的情形，一个LCU共有256个最小子块，HEVC规定的编码和解码以及相应的重构次序如图1所示。排出图1所示次序的基本规律如下所述。
1)第一(最高)层次排序：把一个64x64像素的块(LCU)划分成4个32x32像素的块，对这4个块的排序是左上，右上，左下，右下。也就是先对左上块中的所有最小子块进行排序编号(序号为0～63)，再对右上块中的所有最小子块进行排序编号(序号为64～127)，然后对左下块中的所有最小子块进行排序编号(序号为128～191)，最后对右下块中的所有最小子块进行排序编号(序号为192～255)。
2)第二层次排序：把一个32x32像素的块划分成4个16x16像素的块，对这4个块的排序也同样是左上，右上，左下，右下。也就是先对左上块中的所有最小子块进行排序编号(序 号为0～15或64～79或128～143或192～207)，再对右上块中的所有最小子块进行排序编号(序号为16～31或80～95或144～159或208～223)，然后对左下块中的所有最小子块进行排序编号(序号为32～47或96～111或160～175或224～239)，最后对右下块中的所有最小子块进行排序编号(序号为48～63或112～127或176～191或240～255)。
3)第三层次排序：把一个16x16像素的块划分成4个8x8像素的块，对这4个块的排序也同样是左上，右上，左下，右下。
4)第四(最低)层次排序：把一个8x8像素的块划分成4个4x4像素的块(最小子块)，对这4个最小子块的排序也同样是左上，右上，左下，右下。
因此，
一个64x64像素块的左上角32x32像素块的左上角16x16像素块的左上角8x8像素块划分成4＝2x2个4x4像素的最小子块，这4个最小子块的序号分别为0，1，2，3，如图1所示；
一个64x64像素块的左上角32x32像素块的左上角16x16像素块的右上角8x8像素块划分成4＝2x2个4x4像素的最小子块，这4个最小子块的序号分别为4，5，6，7，如图1所示；
一个64x64像素块的左上角32x32像素块的左上角16x16像素块的左下角8x8像素块划分成4＝2x2个4x4像素的最小子块，这4个最小子块的序号分别为8，9，10，11，如图1所示；
一个64x64像素块的左上角32x32像素块的左上角16x16像素块的右下角8x8像素块划分成4＝2x2个4x4像素的最小子块，这4个最小子块的序号分别为12，13，14，15，如图1所示。
按照以上规律，可对各个层次的左上，右上，左下，右下的块中的最小子块进行排序，最后得出图1所示的所有256个最小子块的序号。
从以上规律，可知上下左右相邻2x2＝4个最小子块组成一个8x8像素块。上下左右相邻2x2＝4个8x8像素块组成一个16x16像素块。上下左右相邻2x2＝4个16x16像素块组成一个32x32像素块。上下左右相邻2x2＝4个32x32像素块组成一个64x64像素块(LCU)。
一个彩色像素通常有3个分量(component)组成。最常用的两种像素色彩格式(pixel color format)是由绿色分量、蓝色分量、红色分量组成的GBR色彩格式和由一个亮度(luma)分量及两个色度(chroma)分量组成的YUV色彩格式，通称为YUV的色彩格式实际包括多种色彩格式，如YCbCr色彩格式。因此，对一个CU进行编码时，可以把一个CU分成3个分量平面(G平面、B平面、R平面或Y平面、U平面、V平面)，对3个分量平面分别进行编码；也可以把一个像素的3个分量捆绑组合成一个3元组，对由这些3元组组成的CU整体进行编码。前一种像素及其分量的排列方式称为图像(及其CU)的平面格式(planar format)，而后一种像素及其分量的排列方式称为图像(及其CU)的叠包格式(packed format)。像素的GBR色彩格式和YUV色彩格式都是像素的3分量表现格式。
除了像素的3分量表现格式，像素的另一种常用的现有技术的表现格式是调色板索引表 现格式。在调色板索引表现格式中，一个像素的数值也可以用调色板的索引来表现。调色板空间中存储了需要被表现的像素的3个分量的数值或近似数值，调色板的地址被称为这个地址中存储的像素的索引。一个索引可以表现像素的一个分量，一个索引也可以表现像素的3个分量。调色板可以是一个，也可以是多个。在多个调色板的情形，一个完整的索引实际上由调色板编号和该编号的调色板的索引两部分组成。像素的索引表现格式就是用索引来表现这个像素。像素的索引表现格式在现有技术中也被称为像素的索引颜色(indexed color)或仿颜色(pseudo color)表现格式，或者常常被直接称为索引像素(indexed pixel)或仿像素(pseudo pixel)或像素索引或索引。索引有时也被称为指数。把像素用其索引表现格式来表现也称为索引化或指数化。
其他的常用的现有技术的像素表现格式包括CMYK表现格式和灰度表现格式。
YUV色彩格式又可根据是否对色度分量进行下采样再细分成若干种子格式：1个像素由1个Y分量、1个U分量、1个V分量组成的YUV4:4:4像素色彩格式；左右相邻的2个像素由2个Y分量、1个U分量、1个V分量组成的YUV4:2:2像素色彩格式；左右上下相邻按2x2空间位置排列的4个像素由4个Y分量、1个U分量、1个V分量组成的YUV4:2:0像素色彩格式。一个分量一般用1个8～16比特的数字来表示。YUV4:2:2像素色彩格式和YUV4:2:0像素色彩格式都是对YUV4:4:4像素色彩格式施行色度分量的下采样得到。一个像素分量也称为一个像素样值(pixel sample)或简单地称为一个样值(sample)。
编码或解码时的最基本元素可以是一个像素，也可以是一个像素分量，也可以是一个像素索引(即索引像素)。作为编码或解码的最基本元素的一个像素或一个像素分量或一个索引像素统称为一个像素样值(sample)，有时也通称为一个像素值，或简单地称为一个样值。
在本发明和本发明专利申请中，“像素样值”、“像素值”、“样值”、“索引像素”、“像素索引”是同义词，根据上下文，可以明确是表示“像素”还是表示“一个像素分量”还是表示“索引像素”或者同时表示三者之任一。如果从上下文不能明确，那么就是同时表示三者之任一。
在对计算机屏幕图像和视频数据进行压缩时，帧内块匹配(intra block matching，也称帧内运动补偿intra motion compensation或称帧内块复制intra block copying)以及把块(如8x8像素样值)分成更精细的微块(如4x2像素样值或8x2像素样值或2x4像素样值或2x8像素样值)或线条(即高度为1或宽度为1的微块，如4x1像素样值或8x1像素样值或1x4像素样值或1x8像素样值)或把块内的像素样值排列成一个长度远大于宽度的串(如宽度为1个像素样值而长度为64个像素样值的串或宽度为2个像素样值而长度为32个像素样值的串)后进行以微块或线条或串内可变长度的子串为最小匹配单位的微块匹配(也称帧内微块复制)或线条匹配(也称条匹配或帧内线条复制或帧内条复制)或串匹配(也称串复制或帧内串复制)都是有效的技术。
在本发明和本发明专利申请中，编码块或解码块是指一帧图像中对其施行编码或解码的 一个区域。编码块和解码块统称为块。
因此，在本发明和本发明专利申请中：
块包括但不限于通常所说的块、微块、线条(条)和串；
块匹配包括但不限于通常所说的块匹配、块复制、微块匹配、微块复制、线条匹配、条匹配、线条复制、条复制、串匹配、串复制；
匹配块包括但不限于通常所说的匹配块、匹配微块、匹配线条、匹配条、匹配串；
被匹配块包括但不限于通常所说的被匹配块、被匹配微块、被匹配线条、被匹配条、被匹配串；
块是由若干像素值组成的一个区域。一个块，可以由“像素”组成，也可以由“像素的分量”组成，也可以由“索引像素”组成，也可以由这3者混合组成，也可以由这3者中之任意2种混合组成。
块匹配编码就是在对一个编码块进行编码时，在已重构参考像素样值集之中的某个预定的搜索范围(search range)内搜索与所述编码块(即被匹配块)之间匹配误差最小的一个匹配块(称为最优匹配块)，然后把被匹配块与最优匹配块之间的相对位置(称为移动矢量即motion vector简称MV)写入视频压缩码流中。
块匹配解码就是在对一个解码块的压缩码流段进行解码时，根据从视频压缩码流中解析出来的MV，来确定已重构参考像素样值集之中的匹配块的位置，然后把所述匹配块复制和粘贴到所述解码块(即被匹配块)的位置，即把所述解码块的数值直接或间接设置为等于所述匹配块的数值。
现有块匹配技术中，为了完整地计算匹配误差和把整个匹配块复制和粘贴到正在编码、解码以及重构过程中的被匹配块的位置，匹配块必须完整地在已重构参考像素样值集之中，即匹配块都是原始完整已重构匹配块。特别是，匹配块与被匹配块不能有重叠部分，也就是，一个块不能部分地自己与自己匹配(简称为部分自匹配)。
以前面阐述的LCU的大小为64x64像素而最小子块的大小为4x4像素的情形为例，在对任何一个CU进行编码或解码时(这个CU简称为当前CU，其所在的LCU则简称为当前LCU)，可以按照前面阐述的256个最小子块的排序编号规律，确定已重构参考像素样值集包括：
1)当前LCU中满足下列条件的所有最小子块：其序号小于当前CU所含最小子块的序号；
2)按照预定的编码或解码顺序已经完成编码或解码以及重构的所有LCU，通常至少包括位于当前LCU左侧的LCU(左LCU)，位于当前LCU上侧的LCU(上LCU)，位于当前LCU左上侧的LCU(左上LCU)。
图2是当前CU及其已重构参考像素样值集的一个例。当前CU是1个16x16像素CU，由序号为192～207的16个最小子块组成。对于这个当前CU来说，其已重构参考像素样值集(图2中用斜线阴影表示的部分)包括了其序号小于192的所有最小子块。图2也例示了被匹配块即当前CU与匹配块的位置。匹配块整体全部在已重构参考像素样值集之中。被匹配块与匹配块不会相交，即不会有重叠部分。
图3是当前CU及其已重构参考像素样值集的第二例。当前CU是1个8x8像素CU，由序号为244～247的4个最小子块组成。对于这个当前CU来说，其已重构参考像素样值集(图3中用斜线阴影表示的部分)包括了其序号小于244的所有最小子块。图3也例示了被匹配块即当前CU与匹配块的位置。匹配块整体全部在已重构参考像素样值集之中。被匹配块与匹配块不会相交，即不会有重叠部分。
图4是当前CU及其已重构参考像素样值集的第三例。当前CU是1个16x16像素CU，由序号为80～95的16个最小子块组成。对于这个当前CU来说，其已重构参考像素样值集(图4中用斜线阴影表示的部分)包括了其序号小于80的所有最小子块。图4也例示了被匹配块即当前CU与匹配块的位置。匹配块整体全部在已重构参考像素样值集之中。被匹配块与匹配块不会相交，即不会有重叠部分。
图5是当前CU及其已重构参考像素样值集的第四例。当前CU是1个8x8像素CU，由序号为36～39的4个最小子块组成。对于这个当前CU来说，其已重构参考像素样值集(图5中用斜线阴影表示的部分)包括了其序号小于36的所有最小子块。图5也例示了被匹配块即当前CU与匹配块的位置。匹配块整体全部在已重构参考像素样值集之中。被匹配块与匹配块不会相交，即不会有重叠部分。
图6是当前CU及其已重构参考像素样值集的第五例。当前CU是1个8x8像素CU，由序号为168～171的4个最小子块组成。对于这个当前CU来说，其已重构参考像素样值集(图6中用斜线阴影表示的部分)包括了其序号小于168的所有最小子块。图6也例示了被匹配块即当前CU与匹配块的位置。匹配块整体全部在已重构参考像素样值集之中。被匹配块与匹配块不会相交，即不会有重叠部分。
一帧图像中任意一个像素样值通常(但不限于)用一个相对于图像中一个预先设定的参考点(即原点，通常是但不限于图像的最左上像素样值的点)的坐标(X,Y)来表示，而坐标为(X,Y)的像素样值P的数值则用P(X,Y)来表示。X的增大方向通常(但不限于)是向右，而Y的增大方向通常(但不限于)是向下。设(Xc,Yc)是一个宽度为Nx高度为Ny的被匹配块的最左上像素样值的坐标，而(Xr,Yr)是一个匹配块(必定有与被匹配块完全相同的宽度和高度)的最左上像素样值的坐标。根据平面解析几何常识，匹配块整体在已重构参考像素样值集内，即匹配块的全部像素样值都是已重构像素样值的一个必要条件是匹配块与被匹配块不能有任何相交(即重叠)部分。这个必要条件用坐标，宽度和高度之间的关系来表示就是：
|Xr－Xc|≥Nx或者|Yr－Yc|≥Ny。
在匹配块的全部像素样值都是已重构像素样值的情形，把匹配块复制和粘贴到被匹配块的位置的运算，即把匹配块的所有像素样值的数值直接或间接赋值予被匹配块的运算，可以用(但不限于)下列赋值语句之任一或其组合来完成：
原始完整已重构匹配块的赋值语句1
for(y＝0；y<Ny；++y)
for(x＝0；x<Nx；++x)
P(Xc+x,Yc+y)＝P(Xr+x,Yr+y)
原始完整已重构匹配块的赋值语句2
for(y＝0；y<Ny；++y)
for(x＝0；x<Nx；++x)
P(Xc+Nx–1–x,Yc+y)＝P(Xr+Nx–1–x,Yr+y)
原始完整已重构匹配块的赋值语句3
for(y＝0；y<Ny；++y)
for(x＝0；x<Nx；++x)
P(Xc+x,Yc+Ny–1–y)＝P(Xr+x,Yr+Ny–1–y)
原始完整已重构匹配块的赋值语句4
for(y＝0；y<Ny；++y)
for(x＝0；x<Nx；++x)
P(Xc+Nx–1–x,Yc+Ny–1–y)＝P(Xr+Nx–1–x,Yr+Ny–1–y)
原始完整已重构匹配块的赋值语句5
for(x＝0；x<Nx；++x)
for(y＝0；y<Ny；++y)
P(Xc+x,Yc+y)＝P(Xr+x,Yr+y)
原始完整已重构匹配块的赋值语句6
for(x＝0；x<Nx；++x)
for(y＝0；y<Ny；++y)
P(Xc+Nx–1–x,Yc+y)＝P(Xr+Nx–1–x,Yr+y)
原始完整已重构匹配块的赋值语句7
for(x＝0；x<Nx；++x)
for(y＝0；y<Ny；++y)
P(Xc+x,Yc+Ny–1–y)＝P(Xr+x,Yr+Ny–1–y)
原始完整已重构匹配块的赋值语句8
for(x＝0；x<Nx；++x)
for(y＝0；y<Ny；++y)
P(Xc+Nx–1–x,Yc+Ny–1–y)＝P(Xr+Nx–1–x,Yr+Ny–1–y)
由于匹配块整体在已重构参考像素样值集内，匹配块与被匹配块必不相交，以上8种赋值语句是完全等价的。
当匹配块是垂直匹配条(即宽度Nx＝1的匹配块)的时候，原始完整已重构匹配条的赋值语句是(但不限于)下列赋值语句之任一或其组合：
垂直的原始完整已重构匹配条的赋值语句1
for(y＝0；y<Ny；++y)
P(Xc,Yc+y)＝P(Xr,Yr+y)
垂直的原始完整已重构匹配条的赋值语句2
for(y＝0；y<Ny；++y)
P(Xc,Yc+Ny–1–y)＝P(Xr,Yr+Ny–1–y)
以上2种赋值语句是完全等价的。
在匹配块是水平匹配条(即高度Ny＝1的匹配块)的时候，原始完整已重构匹配条的赋值语句是(但不限于)下列赋值语句之任一或其组合：
水平的原始完整已重构匹配条的赋值语句1
for(x＝0；x<Nx；++x)
P(Xc+x,Yc)＝P(Xr+x,Yr)
水平的原始完整已重构匹配条的赋值语句2
for(x＝0；x<Nx；++x)
P(Xc+Nx–1–x,Yc)＝P(Xr+Nx–1–x,Yr)
以上2种赋值语句是完全等价的。
当匹配块是匹配串(即宽度为一个较小的非独立的参数W而长度为一个独立可变的参数L的匹配块)的时候，像素样值的坐标(X,Y)成为一个1维地址K，而地址为K的像素样值P的数值则用P(K)来表示。设Kc是一个长度为L的被匹配串的第一个像素样值的地址，而Kr是一个匹配串(必定有与被匹配串完全相同的长度)的第一个像素样值的地址。原始完 整已重构匹配串的赋值语句是(但不限于)下列赋值语句之任一或其组合：
原始完整已重构匹配串的赋值语句1
for(k＝0；k<L；++k)
P(Kc+k)＝P(Kr+k)
原始完整已重构匹配串的赋值语句2
for(k＝0；k<L；++k)
P(Kc+L–1–k)＝P(Kr+L–1–k)
以上2种赋值语句是完全等价的。
匹配块的同义词包括但不限于参考块、预测块。在编码过程中，被匹配块的同义词包括但不限于原始块、当前块、当前编码块。在解码和重构过程中，被匹配块的同义词包括但不限于重构块、重建块、当前块、当前解码块。
匹配条的同义词包括但不限于参考条、预测条。在编码过程中，被匹配条的同义词包括但不限于原始条、当前条、当前编码条。在解码和重构过程中，被匹配条的同义词包括但不限于重构条、重建条、当前条、当前解码条。
匹配串的同义词包括但不限于参考串、预测串。在编码过程中，被匹配串的同义词包括但不限于原始串、当前串、当前编码串。在解码和重构过程中，被匹配串的同义词包括但不限于重构串、重建串、当前串、当前解码串。
在本发明和本发明专利申请中，匹配编码的同义词包括但不限于复制编码和广义的预测编码，匹配解码的同义词包括但不限于复制解码和广义的预测解码。在本发明和本发明专利申请中，预测编码是广义的预测编码的简称，预测解码是广义的预测解码的简称。因此，块匹配编码、微块匹配编码、线条匹配编码、条匹配编码、串匹配编码、任意形状匹配编码的同义词各自分别包括但不限于块复制编码和块预测编码、微块复制编码和微块预测编码、线条复制编码和线条预测编码、条复制编码和条预测编码、串复制编码和串预测编码、任意形状复制编码和任意形状预测编码，同样，块匹配解码、微块匹配解码、线条匹配解码、条匹配解码、串匹配解码、任意形状匹配解码的同义词各自分别包括但不限于块复制解码和块预测解码、微块复制解码和微块预测解码、线条复制解码和线条预测解码、条复制解码和条预测解码、串复制解码和串预测解码、任意形状复制解码和任意形状预测解码。
在本发明和本发明专利申请中，复制编码包括但不限于块复制编码、帧内块复制编码、 微块复制编码、帧内微块复制编码、线条复制编码、帧内线条复制编码、条复制编码、帧内条复制编码、串复制编码、帧内串复制编码、任意形状复制编码、帧内任意形状复制编码，复制解码包括但不限于块复制解码、帧内块复制解码、微块复制解码、帧内微块复制解码、线条复制解码、帧内线条复制解码、条复制解码、帧内条复制解码、串复制解码、帧内串复制解码、任意形状复制解码、帧内任意形状复制解码。
在本发明和本发明专利申请中，预测编码包括但不限于块预测编码、帧内块预测编码、微块预测编码、帧内微块预测编码、线条预测编码、帧内线条预测编码、条预测编码、帧内条预测编码、串预测编码、帧内串预测编码、任意形状预测编码、帧内任意形状预测编码，预测解码包括但不限于块预测解码、帧内块预测解码、微块预测解码、帧内微块预测解码、线条预测解码、帧内线条预测解码、条预测解码、帧内条预测解码、串预测解码、帧内串预测解码、任意形状预测解码、帧内任意形状预测解码。
现有块匹配技术中，由于匹配块整体被限制在已重构参考像素样值集之中，被匹配块与匹配块不得相交，就不能有效地找到图像中可能大量存在的近距离的匹配块，对这类图像和图案的编码效率很低。
发明内容
为了解决图像视频编码和解码的现有技术中的这一问题，本发明提供了一种匹配块仅需部分在已重构参考像素样值集之中，特别是被匹配块与匹配块可相交(即有重叠部分，也称为部分自匹配)的图像编码与解码的方法和装置。仅有一部分而非完整地在已重构参考像素样值集之中的匹配块称为部分已重构匹配块。另一方面，完整地在已重构参考像素样值集之中的匹配块称为原始完整已重构匹配块。
本发明的主要技术特征如图7和图8所示。与当前块(被匹配块)对应的匹配块，不一定需要全部都位于已重构参考像素样值集之中，只要有至少一个像素样值位于已重构参考像素样值集之中即可，因而允许被匹配块与匹配块有重叠部分。图7中例示了2个16x16像素CU(被匹配块)和3个8x8像素CU(被匹配块)。图8中例示了2个16x16像素CU(被匹配块)和5个8x8像素CU(被匹配块)。他们分别对应的匹配块都是部分已重构匹配块，即都只有一部分(图中用斜线阴影表示的部分)在已重构参考像素样值集之中。一个匹配块的在已重构参考像素样值集之中的部分简称为该匹配块的已重构部分，而其余部分则简称为该匹配块的未重构部分。从图中可以看出(也可以严格证明)，一个部分已重构匹配块的已重构部分(斜线阴影表示的部分)有4种情形：
1)一个部分已重构匹配块的已重构部分是该匹配块的上部；
2)一个部分已重构匹配块的已重构部分是该匹配块的左部；
3)一个部分已重构匹配块的已重构部分是该匹配块的左上部；
4)一个部分已重构匹配块的已重构部分是该匹配块的除去右下部的部分，即未重构部分是右下部。在这种情形，已重构部分呈Г形状，由左右两个矩形构成，分别称为左矩形和右矩形。
在块是线条(即高度或宽度为1个样值的块)的时候，以上4种情形变成2种情形，即情形1)和情形2)，而情形3)和情形4)不可能存在。
在块是串(即块内的像素样值被排列成一个长度远大于宽度的串)的时候，以上4种情形变成1种情形，即一个部分已重构匹配串的已重构部分是该匹配串的前部。
图7和图8中表示的匹配块和被匹配块可以是叠包格式的匹配块和被匹配块，也可以是平面格式的一个分量(样值)的匹配块和被匹配块。因而本发明的方法和装置，既可以适用于对叠包格式的LCU和CU的像素的编码、解码以及重构，也可以适用于对平面格式的LCU和CU的一个平面的像素样值的编码、解码以及重构。
本发明的编码方法和装置中，最基本的特有技术特征就是在对当前编码块进行块匹配编码的最优匹配块搜索时，候选的用作参考的匹配块(即位于预定的搜索范围内的所有可能成为最优匹配块的匹配块)，不一定需要全部都位于已重构参考像素样值集之中，即可以含有未重构部分。对于含有未重构部分的匹配块，在计算候选的参考的匹配块与当前编码中的被匹配块之间的匹配误差时，必须首先用所述匹配块的已重构部分的某些像素样值或已重构部分的邻近的某些像素样值或其他像素样值来填充补全所述匹配块的未重构部分，然后再进行匹配误差的计算。等价地，所述填充补全和匹配误差计算这两个运算也可以合并成一步来完成，即把已重构部分或其邻近部分或其他部分的那些用来填充补全的像素样值直接用来计算与被匹配块之间的匹配误差。这两种方式在运算逻辑上和最终实现效果上完全等价。因此，本发明的阐述仅按照第一种方式展开以避免赘述。
本发明的解码方法和装置中，最基本的特有技术特征就是在对当前解码块进行块匹配解码的匹配块复制和粘贴(即把所述解码块的数值直接或间接设置为等于所述匹配块的数值)时，所述匹配块不一定需要全部都位于已重构参考像素样值集之中，即可以含有未重构部分。对于含有未重构部分的匹配块，必须首先用所述匹配块的已重构部分的某些像素样值或已重构部分的邻近的某些像素样值或其他像素样值来填充补全所述匹配块的未重构部分，然后再将填充补全后的完整的匹配块复制和粘贴到所述当前CU(即被匹配块)的位置。等价地，所 述填充补全和复制粘贴这两个操作步骤也可以合并成一个操作步骤来完成，即把已重构部分或其邻近部分或其他部分的那些用来填充补全的像素样值用一个操作步骤复制粘贴到(即直接或间接赋值予)被匹配块。这两种方式在运算逻辑上和最终实现效果上完全等价。因此，本发明的阐述仅按照第一种方式展开以避免赘述。
综合本发明的编码与解码的方法和装置，其基本的特有技术特征就是在进行块匹配编码和解码时，匹配块可以含有未重构部分。对于含有未重构部分的匹配块即部分已重构匹配块，必须首先用所述匹配块的已重构部分的某些像素样值或已重构部分的邻近的某些像素样值或其他像素样值来填充补全所述匹配块的未重构部分，即将所述匹配块的已重构部分的某些像素样值或已重构部分的邻近的某些像素样值或其他像素样值的数值直接或间接赋值予所述匹配块的未重构部分，然后再进行编码和解码的其他后续运算。
如前面所述，一个匹配块的已重构部分有4种情形。在这4种情形，都可以用某种合适的方式来填充补全(即直接或间接赋值予)所述匹配块的未重构部分。在块是线条(即高度或宽度为1个样值的块)的时候，匹配块成为匹配条，已重构部分的4种情形也变成2种情形。在块是串(即块内的像素样值被排列成一个长度远大于宽度的串)的时候，匹配块成为匹配串，已重构部分的4种情形也变成1种情形。
在匹配块或宽度为1的匹配条的已重构部分是该匹配块或宽度为1的匹配条的上部的情形(图9的情形1)，如果已重构部分大于未重构部分，则可以从上部已重构部分取出若干行像素样值填充补全(即直接或间接赋值予)下部未重构部分。从上部取出的若干行可以是任意但预先约定(以保证编码器与解码器的一致性和正确性)的若干行，如已重构部分中最上面的若干行，或者已重构部分中最下面的若干行。如果已重构部分小于未重构部分，则可以多次从上部已重构部分取出若干行像素样值，重复填充补全(即直接或间接赋值予)下部未重构部分。所述从上部已重构部分取出的若干行，可以是上部已重构部分的全部行，也可以是上部已重构部分中的一部分行。必须预先约定怎么取的规则，以保证编码器与解码器的一致性和正确性。简言之，所述填充补全方式就是从匹配块或宽度为1的匹配条的已重构部分复制部分或全部像素样值，从上到下(即以垂直方向)粘贴填满匹配块或宽度为1的匹配条的未重构部分，即把匹配块或宽度为1的匹配条的已重构部分的部分或全部像素样值的数值从上到下(即以垂直方向)直接或间接赋值予匹配块或宽度为1的匹配条的未重构部分。
在匹配块或高度为1的匹配条的已重构部分是该匹配块或高度为1的匹配条的左部的情形(图9的情形2)，如果已重构部分大于未重构部分，则可以从左部已重构部分取出若干列像素样值填充补全(即直接或间接赋值予)右部未重构部分。从左部取出的若干列可以是任意但预先约定(以保证编码器与解码器的一致性和正确性)的若干列，如已重构部分中最左面的若干列，或者已重构部分中最右面的若干列。如果已重构部分小于未重构部分，则可以多次从左部已重构部分取出若干列像素样值，重复填充补全(即直接或间接赋值予)右部未重构部分。所述从左部已重构部分取出的若干列，可以是左部已重构部分的全部列，也可以 是左部已重构部分中的一部分列。必须预先约定怎么取的规则，以保证编码器与解码器的一致性和正确性。简言之，所述填充补全方式就是从匹配块或高度为1的匹配条的已重构部分复制部分或全部像素样值，从左到右(即以水平方向)粘贴填满匹配块或高度为1的匹配条的未重构部分，即把匹配块或高度为1的匹配条的已重构部分的部分或全部像素样值的数值从从左到右(即以水平方向)直接或间接赋值予匹配块或高度为1的匹配条的未重构部分。
在匹配块的已重构部分是该匹配块的左上部的情形(图10的情形3)，填充补全(即赋值)的过程可以分两步来进行。第一步，一次或多次从左上部已重构部分取出若干行已重构像素样值填充补全(即直接或间接赋值予)左下部未重构部分。第二步，一次或多次从左部已重构和已填充补全(即已赋值)部分取出若干列像素样值填充补全(即直接或间接赋值予)右部未重构部分。也可以用另一种等价的方式。即，第一步，一次或多次从左上部已重构部分取出若干列已重构像素样值填充补全(即直接或间接赋值予)右上部未重构部分。第二步，一次或多次从上部已重构和已填充补全(即已赋值)部分取出若干行像素样值填充补全(即直接或间接赋值予)下部未重构部分。简言之，所述填充补全方式就是从匹配块的已重构部分复制部分像素样值，先从上到下(即以垂直方向)再从左到右(即以水平方向)粘贴填满匹配块的未重构部分，或者等价地，先从左到右(即以水平方向)再从上到下(即以垂直方向)粘贴填满匹配块的未重构部分，即把匹配块的已重构部分的部分或全部像素样值的数值，先从上到下(即以垂直方向)再从左到右(即以水平方向)直接或间接赋值予匹配块的未重构部分，或者等价地，先从左到右(即以水平方向)再从上到下(即以垂直方向)直接或间接赋值予匹配块的未重构部分。
在匹配块的已重构部分是该匹配块的除去右下部的部分，即未重构部分是右下部的情形(图10的情形4)，已重构部分呈Г形状，由左右两个矩形构成，分别称为左矩形和右矩形。填充补全(即赋值)的一般方式是：一次或多次从Г形已重构部分中取出也呈Г形状的若干像素样值，用来填充补全(即直接或间接赋值予)右下部未重构部分。如果一次取出的Г形像素样值不足以补全(即赋值予)右下部未重构部分的话，就要多次取出，重复填充补全(即赋值予)。所述一般方式的一个特例是：取出的Г形像素样值可退化成位于左矩形内的一个矩形，用来填充补全(即直接或间接赋值予)右下部未重构部分。所述一般方式的另一个特例是：取出的Г形像素样值可退化成位于右矩形内的一个矩形，用来填充补全(即直接或间接赋值予)右下部未重构部分。简言之，所述填充补全方式就是从匹配块的已重构部分复制部分像素样值，从左上到右下(即以45°方向)或者从左到右(即以水平方向)或者从上到下(即以垂直方向)粘贴填满匹配块的未重构部分，即把匹配块的已重构部分的部分或全部像素样值的数值，从左上到右下(即以45°方向)或者从左到右(即以水平方向)或者从上到下(即以垂直方向)直接或间接赋值予匹配块的未重构部分。
一个部分已重构匹配串的已重构部分通常是该匹配串的前部，如果已重构部分大于未重构部分，则可以从前部已重构部分取出若干个像素样值填充补全(即直接或间接赋值予)后部未重构部分。从前部取出的若干个像素样值可以是任意但预先约定(以保证编码器与解码 器的一致性和正确性)的若干个像素样值，如已重构部分中最前面的若干个，或者已重构部分中最后面的若干个。如果已重构部分小于未重构部分，则可以多次从前部已重构部分取出若干个像素样值，重复填充补全(即直接或间接赋值予)后部未重构部分。所述从前部已重构部分取出的若干个，可以是前部已重构部分的全部像素样值，也可以是前部已重构部分中的一部分像素样值。必须预先约定怎么取的规则，以保证编码器与解码器的一致性和正确性。简言之，所述填充补全方式就是从匹配串的已重构部分复制部分或全部像素样值，从前到后粘贴填满匹配串的未重构部分，即把匹配串的已重构部分的部分或全部像素样值的数值从前到后直接或间接赋值予匹配串的未重构部分。
部分已重构匹配块中有一大类是匹配块与被匹配在位置上有重叠部分的匹配块。这一大类特殊的部分已重构匹配块称为与被匹配块相交的匹配块，也称为部分地自己与自己匹配的匹配块，简称为部分自匹配的匹配块或部分自匹配块。图7所示的5对匹配块和被匹配块的例中有4对是部分自匹配的例，有1对不是部分自匹配的例。图8所示的6对匹配块和被匹配块的例中有5对是部分自匹配的例，有1对不是部分自匹配的例。
设(Xc,Yc)是一个宽度为Nx高度为Ny的被匹配块的最左上像素样值的坐标，而(Xr,Yr)是一个匹配块(必定有与被匹配块相同的宽度和高度)的最左上像素样值的坐标。根据平面解析几何常识，与被匹配块相交的匹配块等价于满足下列关系的匹配块：|Xr－Xc|<Nx并且|Yr－Yc|<Ny。另外，根据编码和解码的顺序，Xr≥Xc和Yr≥Yc两式永远不会同时成立。
坐标(Xc,Yc)和(Xr,Yr)与以上4种情形存在下列关系，可用作判别匹配块的已重构部分属于哪种情形的充分条件：
1)如果Xr≥Xc和Yc–Ny<Yr<Yc两式同时成立，并且当前被匹配块的右上区域在已重构参考像素样值集之中，则属于情形1。
2)如果Xc–Nx<Xr<Xc和Yr≥Yc两式同时成立，并且当前被匹配块的左下区域在已重构参考像素样值集之中，则属于情形2。
3)如果Xr≥Xc和Yc–Ny<Yr<Yc两式同时成立但当前被匹配块的右上区域不在已重构参考像素样值集之中或者Xc–Nx<Xr<Xc和Yr≥Yc两式同时成立但当前被匹配块的左下区域不在已重构参考像素样值集之中，则属于情形3。
4)如果Xc–Nx<Xr<Xc和Yc–Ny<Yr<Yc两式同时成立，则属于情形4。
以上通过若干特定的具体实例说明本发明的技术特征。本领域技术人员可由本说明书所揭示的内容轻易地了解本发明的其他优点与功效。本发明还可以通过另外不同的具体实施方式加以实施或应用，本说明书中的各项细节也可以基于不同观点与应用，在不背离本发明的精神下进行各种修饰或改变。本发明中的匹配块的其他名称包括但不限于参考块、预测块。 在编码过程中，本发明中的被匹配块的其他名称包括但不限于原始块、当前块、当前编码块。在解码和重构过程中，本发明中的被匹配块的其他名称包括但不限于重构块、重建块、当前块、当前解码块。
本发明的编码方法的流程示意图如图11所示。本发明的编码方法包括但不限于如下步骤的部分或全部：
1)输入一个当前被匹配块的位置和大小，同时输入若干个在预定的搜索范围内的候选的匹配块的位置和大小，对于所述被匹配块和所述若干个候选的匹配块中的每一个匹配块，执行下列步骤：从所述被匹配块的位置和大小以及所述匹配块的位置和大小判断出所述匹配块是否整体在重构参考像素样值暂存区内，如果是，顺序执行下一步骤，否则，跳至步骤3)；
2)从所述重构参考像素样值暂存区内取出原始完整已重构匹配块，并选择所述原始完整已重构匹配块作为后续步骤5)中移动矢量搜索的输入匹配块；跳至步骤5)；
3)从所述重构参考像素样值暂存区内取出部分已重构匹配块，用所述部分已重构匹配块中的全部或部分已重构像素样值和/或所述部分已重构匹配块的邻近的部分已重构像素样值填充补全所述部分已重构匹配块中的未重构部分，即把所述部分已重构匹配块中的全部或部分已重构像素样值和/或所述部分已重构匹配块的邻近的部分已重构像素样值的数值直接或间接赋值予所述部分已重构匹配块中的未重构部分，产生填充补全已重构匹配块；
4)选择所述填充补全已重构匹配块作为后续步骤5)中移动矢量搜索的输入匹配块；
5)以所述若干个所述匹配块(即步骤2的所述输入匹配块或者步骤4的所述输入匹配块)作为参考和候选对象，对所述被匹配块进行移动矢量搜索以找到最优匹配块和最优移动矢量并且进行编码的其余步骤。
本发明的解码方法的流程示意图如图12所示。本发明的解码方法包括但不限于如下步骤的部分或全部：
1)根据从输入视频压缩码流中获取的移动矢量(即编码器搜索到的最优移动矢量)，判断出与当前解码位置上被匹配块对应的匹配块(即编码器搜索到的最优匹配块)是否整体在重构参考像素样值暂存区内，如果是，顺序执行下一步骤，否则，跳至步骤3)；
2)从所述重构参考像素样值暂存区内取出原始完整已重构匹配块，并选择所述原始完整已重构匹配块作为后续步骤5)中所要复制的匹配块；跳至步骤5)；
3)从所述重构参考像素样值暂存区内取出部分已重构匹配块，用所述部分已重构匹配块中的全部或部分已重构像素样值和/或所述部分已重构匹配块的邻近的部分已重构像素样值填充补全所述部分已重构匹配块中的未重构部分，即把所述部分已重构匹配块中的全部或部分已重构像素样值和/或所述部分已重构匹配块的邻近的部分已重构像素样值的数值直接或间接赋值予所述部分已重构匹配块中的未重构部分，产生填充补全已重构匹配块；
4)选择所述填充补全已重构匹配块作为后续步骤5)中所要复制的匹配块；
5)复制所述匹配块(即步骤2的所述匹配块或者步骤4的所述匹配块)并粘贴到所述被匹配块的位置，即将所述匹配块的数值直接或间接赋值予所述被匹配块，然后进行解码其余步骤。
本发明的编码装置示意图如图13所示。编码装置包括但不限于以下各模块的部分或全部：
1)匹配块是否整体在重构参考像素样值暂存模块内的判断模块：本模块输入包括一个当前被匹配块的位置和大小，本模块输入还包括在预定的搜索范围内的一个候选的匹配块的位置和大小，本模块从所述被匹配块的位置和大小以及所述匹配块的位置和大小判断出所述匹配块是否整体在重构参考像素样值暂存模块内；
2)重构参考像素样值暂存模块：暂存到当前编码中被匹配块的位置为止的所有以前已重构像素样值，用作当前编码中被匹配块的参考像素样值(即候选的匹配块的像素样值)；
3)用已重构部分填充补全(即赋值予)未重构部分模块(构造未重构部分模块)：从所述重构参考像素样值暂存模块中取出部分已重构匹配块，用所述部分已重构匹配块中的全部或部分已重构像素样值和/或所述部分已重构匹配块的邻近的部分已重构像素样值填充补全所述部分已重构匹配块中的未重构部分，即把所述部分已重构匹配块中的全部或部分已重构像素样值和/或所述部分已重构匹配块的邻近的部分已重构像素样值的数值直接或间接赋值予所述部分已重构匹配块中的未重构部分，产生填充补全的匹配块；
4)移动矢量搜索模块；本模块的功能是对于一个输入的当前被匹配块，在重构参考像素样值集即所述重构参考像素样值暂存模块中，按照预定的评估准则，在一个预定的搜索范围内，搜索得到一个最优匹配块以及对应的最优移动矢量；所述最优匹配块可以是来自所述重构参考像素样值暂存模块的原始完整已重构匹配块，也可以是来自所述用已重构部分填充补全未重构部分模块的填充补全的匹配块；
5)其余编码运算模块：施行对被匹配块，所述被匹配块所在的当前编码CU和当前编 码LCU的其余编码运算。
本发明的解码装置示意图如图14所示。解码装置包括但不限于以下各模块的部分或全部：
1)匹配块是否整体在重构参考像素样值暂存模块内的判断模块：本模块输入包括但不限于从输入视频压缩码流中获取的移动矢量，本模块从所述移动矢量和当前解码中被匹配块的位置与大小判断出与所述移动矢量对应的匹配块是否整体在重构参考像素样值暂存模块内；
2)重构参考像素样值暂存模块：暂存到当前解码中被匹配块的位置为止的所有以前已重构像素样值，用作当前解码中被匹配块的参考像素样值(即所述匹配块的像素样值)；
3)用已重构部分填充补全(即赋值予)未重构部分模块(构造未重构部分模块)：从所述重构参考像素样值暂存模块中由所述移动矢量指定的位置上，取出部分已重构匹配块，用所述部分已重构匹配块中的全部或部分已重构像素样值和/或所述部分已重构匹配块的邻近的部分已重构像素样值填充补全所述部分已重构匹配块中的未重构部分，即把所述部分已重构匹配块中的全部或部分已重构像素样值和/或所述部分已重构匹配块的邻近的部分已重构像素样值的数值直接或间接赋值予所述部分已重构匹配块中的未重构部分，产生填充补全的匹配块；
4)复制匹配块并粘贴到被匹配块的位置(即把匹配块的数值赋值予被匹配块)的模块：本模块的功能是从所述重构参考像素样值暂存模块中由所述移动矢量指定的位置上复制原始完整已重构匹配块，或者复制模块3)产生的填充补全的匹配块，并将所述原始完整已重构匹配块或者所述填充补全的匹配块粘贴到所述当前解码中被匹配块的位置，即将所述原始完整已重构匹配块或者所述填充补全的匹配块的数值直接或间接赋值予所述当前解码中被匹配块；
5)其余解码运算模块：施行对所述当前解码中被匹配块，所述当前解码中被匹配块所在的当前解码CU和当前解码LCU的其余解码运算。
以上所提供的图示仅以示意方式说明本发明的基本构想，图示中仅显示与本发明直接有关的组件而非按照实际实施时的组件数目、形状及尺寸绘制，其实际实施时各组件的型态、数量及比例可为一种随意的改变，且其组件布局型态也可能更为复杂。
以下是本发明的更多的实施细节和变体。
本发明的解码方法可以等价地包括但不限于如下具有相同最终技术效果的步骤的部分或全部：
1)根据从输入视频压缩码流中获取的移动矢量(即编码器搜索到的最优移动矢量)，判断出与当前解码位置上被匹配块对应的匹配块(即编码器搜索到的最优匹配块)是否整体在重构参考像素样值暂存区内，如果是，顺序执行下一步骤，否则，跳至步骤3)；
2)从所述重构参考像素样值暂存区内复制原始完整已重构匹配块，并把所述原始完整已重构匹配块粘贴到所述被匹配块的位置，即把所述原始完整已重构匹配块的数值直接或间接赋值予所述被匹配块；跳至步骤5)；
3)从所述重构参考像素样值暂存区内复制部分已重构匹配块中的已重构像素样值，并把所述已重构像素样值粘贴到所述被匹配块的相应位置，即把所述已重构像素样值的数值直接或间接赋值予所述被匹配块的相应部分；
4)把全部或部分所述已重构像素样值或所述部分已重构匹配块的邻近的部分已重构像素样值直接粘贴到所述被匹配块中未被以上步骤3)粘贴到的位置，填充补全整个被匹配块，即把全部或部分所述已重构像素样值或所述部分已重构匹配块的邻近的部分已重构像素样值的数值直接或间接赋值予所述被匹配块中未被以上步骤3)赋值到的部分，完成整个被匹配块的全部赋值；
5)进行解码其余步骤。
解码装置可以等价地包括但不限于以下具有相同最终技术效果的各模块的部分或全部：
1)匹配块是否整体在重构参考像素样值暂存模块内的判断模块：本模块输入包括但不限于从输入视频压缩码流中获取的移动矢量，本模块从所述移动矢量和当前解码中被匹配块的位置与大小判断出与所述移动矢量对应的匹配块是否整体在重构参考像素样值暂存模块内；
2)重构参考像素样值暂存模块：暂存到当前解码中被匹配块的位置为止的所有以前已重构像素样值，用作当前解码中被匹配块的参考像素样值；
3)复制原始完整已重构匹配块并粘贴到被匹配块的位置(即把原始完整已重构匹配块的数值赋值予被匹配块)的模块：本模块的功能是从所述重构参考像素样值暂存模块中由所述移动矢量指定的位置上复制原始完整已重构匹配块，并将所述原始完整已重构匹配块粘贴到所述当前解码中被匹配块的位置，即把所述原始完整已重构匹配块的数值直接或间接赋值予所述当前解码中被匹配块；
4)复制部分已重构匹配块并粘贴到被匹配块的位置(即把部分已重构匹配块的数值赋值予被匹配块)的模块：本模块的功能是从所述重构参考像素样值暂存模块中由所述移动矢量指定的位置上，复制部分已重构匹配块中的已重构像素样值，首先把所述已重构像素样值粘贴到所述被匹配块的相应位置，然后把全部或部分所述已重构像素样值和/或所述部分已重构匹配块的邻近的部分已重构像素样值直接粘贴到所述被匹配块中尚未被粘贴到的位置，填充补全整个被匹配块，即首先把所述已重构像素样值的数值直接或间接赋值予所述被匹配块的相应部分，然后把全部或部分所述已重构像素样值和/或所述部分已重构匹配块的邻近的部分已重构像素样值的数值直接或间接赋值予所述被匹配块中尚未被赋值到的部分，完成整个被匹配块的全部赋值；
5)其余解码运算模块：施行对所述当前解码中被匹配块，所述当前解码中被匹配块所在的当前解码CU和当前解码LCU的其余解码运算。
部分已重构匹配块包括但不限于：满足|Xr－Xc|<Nx并且|Yr－Yc|<Ny的匹配块，即与被匹配块相交的匹配块，也被称为部分自匹配的匹配块，相应的被匹配块被称为部分自匹配的被匹配块。这是一类最常见的部分已重构匹配块。
在部分自匹配的情形，一种把匹配块的像素样值的数值赋值予被匹配块的方法是：把匹配块的已重构部分的全部的数值，首先直接或间接赋值予被匹配块的相应位置的部分，如果被匹配块中还有未被赋值的部分，则把所述匹配块的已重构部分的全部的数值重复直接或间接赋值予被匹配块的未被赋值的部分，直至被匹配块全部被赋值为止。这种把所述匹配块的已重构部分的全部的数值完全重复直接或间接赋值予被匹配块的赋值方法称为部分自匹配的完全赋值方法。
部分自匹配有一种特别情形，就是部分已重构匹配块的未重构部分都全部地完整地在相应的被匹配块之中。这种特别情形被称为Г形部分自匹配。Г形部分自匹配中又有2种特别情形：上下部分自匹配(即Xr＝Xc并且Yc－Ny<Yr<Yc)的情形和左右部分自匹配(即Xc－Nx<Xr<Xc并且Yr＝Yc)的情形。
Г形部分自匹配等价于：Xc–Nx<Xr≤Xc和Yc–Ny<Yr≤Yc两式同时成立但其中两个等号不同时成立。注意两个等号同时成立(Xr＝Xc并且Yr＝Yc)就是匹配块与被匹配块完全重合，这样的匹配块是不存在的。Г形部分自匹配的完全赋值方法可以用(但不限于)下列赋值语句之任一或其组合来完成：
Г形部分自匹配的完全赋值方法的赋值语句1
for(y＝0；y<Ny；++y)
for(x＝0；x<Nx；++x)
P(Xc+x,Yc+y)＝P(Xr+x,Yr+y)
Г形部分自匹配的完全赋值方法的赋值语句2
for(y＝0；y<Ny；++y)
for(x＝0；x<Nx；++x)
P(Xc+Nx–1–x,Yc+y)＝P(Xr+Nx–1–x,Yr+y)
Г形部分自匹配的完全赋值方法的赋值语句3
for(x＝0；x<Nx；++x)
for(y＝0；y<Ny；++y)
P(Xc+x,Yc+y)＝P(Xr+x,Yr+y)
Г形部分自匹配的完全赋值方法的赋值语句4
for(x＝0；x<Nx；++x)
for(y＝0；y<Ny；++y)
P(Xc+x,Yc+Ny–1–y)＝P(Xr+x,Yr+Ny–1–y)
由此可见，部分已重构匹配块的Г形部分自匹配的完全赋值方法的赋值语句1、2、3、4与原始完整已重构匹配块的赋值语句1、2、5、7在形式上一致，两者本质上的差别在于两者的适用范围完全不同：在原始完整已重构匹配块的情形，关系式|Xr－Xc|≥Nx或者|Yr－Yc|≥Ny必须被满足，而在部分已重构匹配块的Г形部分自匹配的完全赋值方法的情形，关系式Xc–Nx<Xr≤Xc和Yc–Ny<Yr≤Yc两式必须同时被满足但其中两个等号不能同时成立。由于两者的赋值语句在形式上是一致的，前者显然是后者的同样形式的赋值语句在适用范围上的扩展。
一类特殊的部分已重构匹配块是：满足Xc－Nx<Xr<Xc并且Yc－Ny<Yr<Yc的匹配块，即与被匹配块相交且已重构部分呈严格Г形状的匹配块。
另一类特殊的部分已重构匹配块是：满足Xc－Nx<Xr≤Xc并且Yc－Ny<Yr≤Yc但其中两个等号不能同时成立的匹配块，即与被匹配块相交且已重构部分呈可退化Г形状的匹配块。其中可退化Г形状就是Xc－Nx<Xr<Xc并且Yr＝Yc的情形或者Xr＝Xc并且Yc－Ny<Yr<Yc的情形，即匹配块与被匹配块是水平位移(左右部分自匹配)或者垂直位移(上下部分自匹配)的关系。
还有一种部分已重构匹配块是：满足Xc－Nx<Xr≤Xc－Nx/2并且Yc－Ny<Yr≤Yc－Ny/2的匹配块，即与被匹配块相交且已重构部分大于匹配块的四分之三的匹配块。
在本发明的编码及解码的方法和装置中，在填充补全(即直接或间接赋值予)一帧图像中的各个部分已重构匹配块中的未重构部分时，统一用所述未重构部分的左侧相邻的由已重构像素样值组成的一个矩形来填充补全，且如图15所示，可以允许所述矩形的部分甚至全部已重构像素样值位于所述部分已重构匹配块之外。
在本发明的编码及解码的方法和装置中，在填充补全(即直接或间接赋值予)一帧图像中的各个部分已重构匹配块中的未重构部分时，统一用所述未重构部分的上侧相邻的由已重构像素样值组成的一个矩形来填充补全，且如图15所示，可以允许所述矩形的部分甚至全部已重构像素样值位于所述部分已重构匹配块之外。
在本发明的编码及解码的方法和装置中，在填充补全(即直接或间接赋值予)一帧图像中的各个部分已重构匹配块中的未重构部分时，有的匹配块用所述未重构部分的左侧相邻的由已重构像素样值组成的一个矩形来填充补全，而有的匹配块则用所述未重构部分的上侧相邻的由已重构像素样值组成的一个矩形来填充补全，且都可以允许所述矩形的部分甚至全部已重构像素样值位于所述部分已重构匹配块之外。
在本发明的编码及解码的方法和装置中，构造或填充补全(即直接或间接赋值予)一个部分已重构匹配块中的未重构部分的方式包括但不限于下列方式之一或其组合：
所述未重构部分的全部或部分数值被直接或间接设置为等于所述参考块的所述重构部分内的部分或全部已重构参考像素样值的数值
或者
所述未重构部分的全部或部分数值被直接或间接设置为等于所述参考块的所述重构部分内的部分或全部已重构参考像素样值的数值按照预先规定的方式进行外插运算得到的数值
或者
所述未重构部分的全部或部分数值被直接或间接设置为等于所述参考块的所述重构部分外但邻近的已重构参考像素样值的数值
或者
所述未重构部分的全部或部分数值被直接或间接设置为等于所述参考块的所述重构部分外但邻近的已重构参考像素样值的数值按照预先规定的方式进行外插运算得到的数值
或者
所述未重构部分的全部或部分数值被直接或间接设置为等于预先规定的数值。
在本发明的编码及解码的方法和装置中，对一个解码块的与一个参考块的未重构部分对应的部分进行赋值的方法包括但不限于下列方式之一或其组合：
把所述参考块的所述重构部分内的部分或全部已重构参考像素样值的数值直接或间接赋值予所述解码块的与所述参考块的所述未重构部分对应的部分
或者
把所述参考块的所述重构部分内的部分或全部已重构参考像素样值的数值按照预先规定的方式进行外插运算得到的数值直接或间接赋值予所述解码块的与所述参考块的所述未重构部分对应的部分
或者
把所述参考块的所述重构部分外但邻近的已重构参考像素样值的数值直接或间接赋值予所述解码块的与所述参考块的所述未重构部分对应的部分
或者
把所述参考块的所述重构部分外但邻近的已重构参考像素样值的数值按照预先规定的方式进行外插运算得到的数值直接或间接赋值予所述解码块的与所述参考块的所述未重构部分对应的部分
或者
把预先规定的数值直接或间接赋值予所述解码块的与所述参考块的所述未重构部分对应的部分。
在本发明的编码及解码的方法和装置中，填充补全(即直接或间接赋值予)一帧图像中的所有和各个部分已重构匹配块中的未重构部分的操作，包括但不限于用扩展已重构参考像素样值集来实现。扩展已重构参考像素样值集的方式包括但不限于下列方式之一或其组合：
扩展部分的像素样值的数值被直接或间接设置为等于所述已重构参考像素样值集内的一部分已重构参考像素样值的数值；
或者
扩展部分的像素样值的数值被直接或间接设置为等于所述已重构参考像素样值集内的与扩展部分邻接的一部分已重构参考像素样值的数值；
或者
扩展部分的像素样值的数值被直接或间接设置为等于对所述已重构参考像素样值集内的一部分已重构参考像素样值的数值按照预先规定的方式进行外插运算得到的数值；
或者
扩展部分的像素样值的数值被直接或间接设置为等于对所述已重构参考像素样值集内的与扩展部分邻接的一部分已重构参考像素样值的数值按照预先规定的方式进行外插运算得到的数值；
或者
扩展部分的像素样值的数值被直接或间接设置为等于预先规定的数值；
或者
所有可能的扩展部分用预先规定的数值来填充。
扩展可以是动态的，不断更新的，也可以是静态的。扩展部分足够大到包含任何一个可能的部分已重构匹配块的可能的未重构部分。这样，任何一个部分已重构匹配块的未重构部分的数值就自动被设置为等于扩展部分的像素样值的数值。
如果采用扩展已重构参考像素样值集构成扩展部分的方式，在把一个部分已重构匹配块直接或间接赋值予一个被匹配块的过程中，首先把所述匹配块的已重构部分直接或间接赋值予所述被匹配块的相应部分，然后把所述匹配块的在所述扩展部分内的像素样值的数值直接或间接赋值予所述被匹配块的尚未被赋值的部分这两项操作实际上可以合并完成：把所述匹配块(包括已重构部分和在所述扩展部分内的像素样值部分)直接或间接赋值予所述被匹配块
在本发明的编码及解码的方法和装置中，当匹配块是匹配条(即宽度Nx＝1或高度Ny＝1的匹配块)的时候，部分自匹配只有两种情形：
1)上下部分自匹配，即Xr＝Xc并且Yc－Ny<Yr<Yc的情形；
2)左右部分自匹配，即Xc－Nx<Xr<Xc并且Yr＝Yc的情形；
这两种情形的完全赋值方法的赋值语句分别是(但不限于)：
上下部分自匹配的完全赋值方法的赋值语句
for(y＝0；y<Ny；++y)
P(Xc,Yc+y)＝P(Xr,Yr+y)
左右部分自匹配的完全赋值方法的赋值语句
for(x＝0；x<Nx；++x)
P(Xc+x,Yc)＝P(Xr+x,Yr)
由此可见：
1)部分已重构匹配条的上下部分自匹配的完全赋值方法的赋值语句与垂直的原始完整已重构匹配条的赋值语句1在形式上一致，两者本质上的差别在于两者的适用范围完全不同：前者适用于Xr＝Xc并且Yc–Ny<Yr<Yc的情形而后者适用于Xr不等于Xc或者|Yr－Yc|≥Ny的情形；由于两者的赋值语句在形式上是一致的，前者显然是后者的同样形式的赋值语句在适用范围上的扩展；
2)部分已重构匹配条的左右部分自匹配的完全赋值方法的赋值语句与水平的原始完整已重构匹配条的赋值语句1在形式上一致，，两者本质上的差别在于两者的适用范围完全不同：前者适用于Xc–Nx<Xr<Xc并且Yr＝Yc的情形而后者适用于|Xr－Xc|≥Nx或者|Yr不等于Yc的情形；由于两者的赋值语句在形式上是一致的，前者显然是后者的同样形式的赋值语句在适用范围上的扩展。
在本发明的编码及解码的方法和装置中，当匹配块是匹配串的时候，部分自匹配只有一种情形：前后部分自匹配，即Kc－L<Kr<Kc的情形。前后部分自匹配的完全赋值方法的赋值语句是
for(k＝0；k<L；++k)
P(Kc+k)＝P(Kr+k)
由此可见，部分已重构匹配串的前后部分自匹配的完全赋值方法的赋值语句与原始完整已重构匹配串的赋值语句1在形式上一致，两者本质上的差别在于两者的适用范围完全不同：前者适用于Kc－L<Kr<Kc的情形而后者适用于Kc－Kr≥L的情形；由于两者的赋值语句在形式上是一致的，前者显然是后者的同样形式的赋值语句在适用范围上的扩展。
图16是本发明中左右部分自匹配的完全赋值方法的示意图。匹配条和被匹配条都有8个像素样值。在例1中，部分已重构匹配条的已重构部分有5个像素样值，被匹配条是这5个像素样值的周期性重复。在例2中，部分已重构匹配条的已重构部分有3个像素样值，被匹配条是这3个像素样值的周期性重复。在例3中，部分已重构匹配条的已重构部分有1个像素样值，被匹配条是这1个像素样值的周期性重复。
附图说明
图1是国际最新视频压缩标准HEVC中最小子块排序编号的一个例
图2是当前CU及其已重构参考像素样值集以及现有技术中匹配块和被匹配块位置的一个例
图3是当前CU及其已重构参考像素样值集以及现有技术中匹配块和被匹配块位置的第二例
图4是当前CU及其已重构参考像素样值集以及现有技术中匹配块和被匹配块位置的第三例
图5是当前CU及其已重构参考像素样值集以及现有技术中匹配块和被匹配块位置的第四例
图6是当前CU及其已重构参考像素样值集以及现有技术中匹配块和被匹配块位置的第五例
图7是本发明中匹配块只有一部分而非全部落在已重构参考像素样值集之中的4个例
图8是本发明中匹配块只有一部分而非全部落在已重构参考像素样值集之中的6个例
图9是本发明中用匹配块的已重构部分来填充补全未重构部分的示意图(已重构部分是匹配块的上部和左部的情形)
图10是本发明中用匹配块的已重构部分来填充补全未重构部分的示意图(已重构部分是匹配块的左上部和除去右下部的部分的情形)
图11是本发明的编码方法流程示意图
图12是本发明的解码方法流程示意图
图13是本发明的编码装置的模块组成示意图
图14是本发明的解码装置的模块组成示意图
图15是本发明中用已重构像素样值(可能在匹配块之外)来填充补全未重构部分的示意图
图16是本发明中左右部分自匹配的完全赋值方法的示意图

Claims (23)

  1. 一种图像编码方法，其特征在于：对一个编码块，搜索得到一个参考块，所述参考块有下列两种情形之一：
    所述参考块是原始完整已重构参考块，即完整地在已重构参考像素样值集内
    或者
    所述参考块是部分已重构参考块，即由在已重构参考像素样值集内的已重构部分和不在已重构参考像素样值集内的未重构部分这两部分构成；
    所述参考块与所述编码块具有同样形状和大小；使用所述参考块作为预测块来构造所述编码块；把在预测编码过程中产生的与预测解码有关的参数放入压缩码流。
  2. 根据权利要求1所述的编码方法，其特征在于包括但不限于以下步骤的全部或部分：
    步骤1)输入一个当前编码块的位置和大小，同时输入若干个在预定的搜索范围内的候选的参考块的位置和大小，对于所述编码块和所述若干个候选的参考块中的每一个参考块，执行下列步骤：从所述编码块的位置和大小以及所述参考块的位置和大小判断出所述参考块是否整体在重构参考像素样值暂存区(即已重构参考像素样值集)内，如果是，顺序执行下一步骤，否则，跳至步骤3)；
    步骤2)从所述重构参考像素样值暂存区内取出原始完整已重构参考块，并选择所述原始完整已重构参考块作为后续步骤5)中移动矢量搜索的输入参考块；跳至步骤5)；
    步骤3)从所述重构参考像素样值暂存区内取出部分已重构参考块，把所述部分已重构参考块中的全部或部分已重构像素样值和/或所述部分已重构参考块的邻近的部分已重构像素样值的数值赋值予所述部分已重构参考块中的未重构部分，产生填充补全已重构参考块；
    步骤4)选择所述填充补全已重构参考块作为后续步骤5)中移动矢量搜索的输入参考块；
    步骤5)以所述若干个所述参考块(即步骤2的所述输入参考块或者步骤4的所述输入参考块)作为参考和候选对象，对所述编码块进行移动矢量搜索以找到最优参考块和最优移动矢量并且进行编码的其余步骤。
  3. 一种图像解码方法，其特征在于：解析压缩码流，获取与预测解码有关的参数；根据所述参数的部分或全部，获取与一个解码块对应的一个参考块，所述参考块有下列两种情形之一：
    所述参考块是原始完整已重构参考块，即完整地在已重构参考像素样值集内
    或者
    所述参考块是部分已重构参考块，即由在已重构参考像素样值集内的已重构部分和不在已重构参考像素样值集内的未重构部分这两部分构成。
  4. 根据权利要求3所述的解码方法，其特征在于：所述参数包括但不限于有关所述解码块的位置的参数，有关所述解码块的大小的参数，有关所述参考块的位置的参数，有关所述参考块的大小的参数。
  5. 根据权利要求3所述的解码方法，其特征在于：在部分已重构参考块的情形，构造所述参考块的所述未重构部分，构造方式包括但不限于下列方式之一或其组合：
    所述未重构部分的全部或部分数值被直接或间接设置为等于所述参考块的所述重构部分内的部分或全部已重构参考像素样值的数值
    或者
    所述未重构部分的全部或部分数值被直接或间接设置为等于所述参考块的所述重构部分内的部分或全部已重构参考像素样值的数值按照预先规定的方式进行外插运算得到的数值
    或者
    所述未重构部分的全部或部分数值被直接或间接设置为等于所述参考块的所述重构部分外但邻近的已重构参考像素样值的数值
    或者
    所述未重构部分的全部或部分数值被直接或间接设置为等于所述参考块的所述重构部分外但邻近的已重构参考像素样值的数值按照预先规定的方式进行外插运算得到的数值
    或者
    所述未重构部分的全部或部分数值被直接或间接设置为等于预先规定的数值。
  6. 根据权利要求5所述的解码方法，其特征在于：在部分已重构参考块的情形，把所述参考块的所述重构部分和构造后的所述参考块的所述未重构部分的数值直接或间接赋值予所述解码块。
  7. 根据权利要求3所述的解码方法，其特征在于：对所述已重构参考像素样值集进行扩展， 构成所述已重构参考像素样值集的扩展部分；扩展已重构参考像素样值集的方式包括但不限于下列方式之一或其组合：
    扩展部分的像素样值的数值被直接或间接设置为等于所述已重构参考像素样值集内的一部分已重构参考像素样值的数值；
    或者
    扩展部分的像素样值的数值被直接或间接设置为等于所述已重构参考像素样值集内的与扩展部分邻接的一部分已重构参考像素样值的数值；
    或者
    扩展部分的像素样值的数值被直接或间接设置为等于对所述已重构参考像素样值集内的一部分已重构参考像素样值的数值按照预先规定的方式进行外插运算得到的数值；
    或者
    扩展部分的像素样值的数值被直接或间接设置为等于对所述已重构参考像素样值集内的与扩展部分邻接的一部分已重构参考像素样值的数值按照预先规定的方式进行外插运算得到的数值；
    或者
    扩展部分的像素样值的数值被直接或间接设置为等于预先规定的数值；
    或者
    所有可能的扩展部分用预先规定的数值来填充。
  8. 根据权利要求7所述的解码方法，其特征在于：在部分已重构参考块的情形，构造所述参考块的所述未重构部分，构造方式包括但不限于：所述参考块的所述未重构部分的全部或部分数值被直接或间接设置为等于所述参考块的在扩展部分内的像素样值的数值。
  9. 根据权利要求8所述的解码方法，其特征在于：在部分已重构参考块的情形，把所述参考块的所述重构部分和构造后的所述参考块的所述未重构部分的数值直接或间接赋值予所述解码块。
  10. 根据权利要求3所述的解码方法，其特征在于：在部分已重构参考块的情形，把所述参考块的所述已重构部分的数值直接或间接赋值予所述解码块的与所述参考块的所述已重构部分对应的部分，对所述解码块的与所述参考块的所述未重构部分对应的部分进行赋值，赋值的方法包括但不限于下列方式之一或其组合：
    把所述参考块的所述重构部分内的部分或全部已重构参考像素样值的数值直接或间接赋值予所述解码块的与所述参考块的所述未重构部分对应的部分
    或者
    把所述参考块的所述重构部分内的部分或全部已重构参考像素样值的数值按照预先规定的方式进行外插运算得到的数值直接或间接赋值予所述解码块的与所述参考块的所述未重构部分对应的部分
    或者
    把所述参考块的所述重构部分外但邻近的已重构参考像素样值的数值直接或间接赋值予所述解码块的与所述参考块的所述未重构部分对应的部分
    或者
    把所述参考块的所述重构部分外但邻近的已重构参考像素样值的数值按照预先规定的方式进行外插运算得到的数值直接或间接赋值予所述解码块的与所述参考块的所述未重构部分对应的部分
    或者
    把预先规定的数值直接或间接赋值予所述解码块的与所述参考块的所述未重构部分对应的部分。
  11. 根据权利要求7所述的解码方法，其特征在于：在部分已重构参考块的情形，把所述参考块的所述已重构部分的数值直接或间接赋值予所述解码块的与所述参考块的所述已重构部分对应的部分，把所述参考块的在扩展部分内的像素样值的数值直接或间接赋值予所述解码块的与所述参考块的所述未重构部分对应的部分，即把所述参考块，包括所述已重构部分和在所述扩展部分内的像素样值部分这两部分的数值，直接或间接赋值予所述解码块。
  12. 根据权利要求3所述的解码方法，其特征在于包括但不限于以下步骤的全部或部分：
    步骤1)根据所述参数的部分或全部，判断出与当前解码位置上解码块对应的参考块是否整体在重构参考像素样值暂存区内，如果是，顺序执行下一步骤，否则，跳至步骤3)；
    步骤2)选择所述重构参考像素样值暂存区中根据所述参数的部分或全部确定的原始完整已重构参考块作为后续步骤5)中所使用的参考块；跳至步骤5)；
    步骤3)把所述重构参考像素样值暂存区中根据所述参数的部分或全部确定的部分已重构参考块中的全部或部分已重构像素样值和/或所述部分已重构参考块的邻近的部分 已重构像素样值的数值和/或预先规定的数值直接或间接赋值予所述部分已重构参考块中的未重构部分，产生填充补全已重构参考块；
    步骤4)选择所述填充补全已重构参考块作为后续步骤5)中所要复制的参考块；
    步骤5)把所述参考块，即步骤2的所述参考块或者步骤4的所述参考块，的数值赋值予所述解码块，然后进行解码其余步骤。
  13. 根据权利要求3所述的解码方法，其特征在于包括但不限于以下步骤的全部或部分：
    步骤1)根据所述参数的部分或全部，判断出与当前解码位置上解码块对应的参考块是否整体在重构参考像素样值暂存区内，如果是，顺序执行下一步骤，否则，跳至步骤3)；
    步骤2)把所述重构参考像素样值暂存区中根据所述参数的部分或全部确定的原始完整已重构参考块的数值赋值予所述解码块；跳至步骤5)；
    步骤3)把所述重构参考像素样值暂存区中根据所述参数的部分或全部确定的部分已重构参考块中的已重构像素样值的数值赋值予所述解码块的相应部分；
    步骤4)把全部或部分所述已重构像素样值和/或所述部分已重构参考块的邻近的部分已重构像素样值的数值和/或预先规定的数值直接或间接赋值予所述解码块中未被以上步骤3)赋值到的部分，完成整个解码块的全部赋值；
    步骤5)进行解码其余步骤。
  14. 一种图像编码装置，其特征在于包括但不限于以下各模块的全部或部分：参考块是否整体在重构参考像素样值暂存模块内的判断模块，构造未重构部分模块。
  15. 根据权利要求14所述的编码装置，其特征在于包括但不限于以下各模块的全部或部分：
    模块1)参考块是否整体在重构参考像素样值暂存模块内的判断模块：本模块输入包括一个当前编码块的位置和大小，本模块输入还包括在预定的搜索范围内的一个候选的参考块的位置和大小，本模块从所述编码块的位置和大小以及所述参考块的位置和大小判断出所述参考块是否整体在重构参考像素样值暂存模块内；
    模块2)重构参考像素样值暂存模块：暂存到当前编码中编码块的位置为止的所有以前已重构像素样值，用作当前编码中编码块的参考像素样值(即候选的参考块的像素样值)；
    模块3)构造未重构部分模块：从所述重构参考像素样值暂存模块中取出部分已重构参考块，把所述部分已重构参考块中的全部或部分已重构像素样值和/或所述部分已重 构参考块的邻近的部分已重构像素样值的数值赋值予所述部分已重构参考块中的未重构部分，产生填充补全的参考块；
    模块4)移动矢量搜索模块；本模块的功能是对于一个输入的当前编码块，在已重构参考像素样值集即所述重构参考像素样值暂存模块中，按照预定的评估准则，在一个预定的搜索范围内，搜索得到一个最优参考块以及对应的最优移动矢量；所述最优参考块可以是来自所述重构参考像素样值暂存模块的原始完整已重构参考块，也可以是来自所述用已重构部分赋值予未重构部分模块的填充补全的参考块；
    模块5)其余编码运算模块：施行对编码块，所述编码块所在的当前编码CU和当前编码LCU的其余编码运算。
  16. 一种图像解码装置，其特征在于包括但不限于以下各模块的全部或部分：参考块是否整体在重构参考像素样值暂存模块内的判断模块，构造未重构部分模块，把部分已重构参考块的数值赋值予解码块的模块。
  17. 根据权利要求16所述的解码装置，其特征在于包括但不限于以下各模块的全部或部分：
    模块1)参考块是否整体在重构参考像素样值暂存模块内的判断模块：本模块输入包括但不限于从压缩码流中获取的与预测解码有关的参数的部分或全部，本模块从所述参数的部分或全部和当前解码中解码块的位置与大小判断出对应的参考块是否整体在重构参考像素样值暂存模块内；
    模块2)重构参考像素样值暂存模块：暂存到当前解码中解码块的位置为止的所有以前已重构像素样值，用作当前解码中解码块的参考像素样值(即所述参考块的像素样值)；
    模块3)构造未重构部分模块：把所述重构参考像素样值暂存模块中由所述参数的部分或全部确定的部分已重构参考块中的全部或部分已重构像素样值和/或所述部分已重构参考块的邻近的部分已重构像素样值的数值和/或预先规定的数值直接或间接赋值予所述部分已重构参考块中的未重构部分，产生填充补全的参考块；
    模块4)把参考块的数值赋值予解码块的模块：本模块的功能是把所述重构参考像素样值暂存模块中由所述参数确定的原始完整已重构参考块或者模块3)产生的填充补全的参考块的数值和/或预先规定的数值直接或间接赋值予当前解码中解码块；
    模块5)其余解码运算模块：施行对所述当前解码中解码块，所述当前解码中解码块所在的当前解码CU和当前解码LCU的其余解码运算。
  18. 根据权利要求16所述的解码装置，其特征在于包括但不限于以下各模块的全部或部分：
    模块1)参考块是否整体在重构参考像素样值暂存模块内的判断模块：本模块输入包括但不限于从压缩码流中获取的与预测解码有关的参数的部分或全部，本模块从所述参数的部分或全部和当前解码中解码块的位置与大小判断出对应的参考块是否整体在重构参考像素样值暂存模块内；
    模块2)重构参考像素样值暂存模块：暂存到当前解码中解码块的位置为止的所有以前已重构像素样值，用作当前解码中解码块的参考像素样值；
    模块3)把原始完整已重构参考块的数值赋值予解码块的模块：本模块的功能是把所述重构参考像素样值暂存模块中由所述参数的部分或全部确定的原始完整已重构参考块的数值赋值予所述当前解码中解码块；
    模块4)把部分已重构参考块的数值赋值予解码块的模块：本模块的功能是1)把所述重构参考像素样值暂存模块中由所述参数确定的部分已重构参考块的已重构像素样值的数值赋值予所述解码块的相应部分，2)把全部或部分所述已重构像素样值和/或所述部分已重构参考块的邻近的部分已重构像素样值的数值和/或预先规定的数值直接或间接赋值予所述解码块中尚未被赋值到的部分，完成整个解码块的全部赋值；
    模块5)其余解码运算模块：施行对所述当前解码中解码块，所述当前解码中解码块所在的当前解码CU和当前解码LCU的其余解码运算。
  19. 根据权利要求1至18中任一权利要求所述的编码方法或解码方法或编码装置或解码装置，其特征在于：所述部分已重构匹配块(即参考块或预测块)是满足|Xr－Xc|<Nx并且|Yr－Yc|<Ny的匹配块，其中(Xc,Yc)是所述被匹配块(即编码块或解码块)的最左上像素样值相对于一帧图像的一个预先设定的原点的坐标，(Xr,Yr)是所述匹配块的最左上像素样值相对于所述原点的坐标，Nx是所述被匹配块和所述匹配块的宽度，Ny是所述被匹配块和所述匹配块的高度；即所述匹配块与所述被匹配块在位置上有重叠部分。
  20. 根据权利要求19所述的编码方法或解码方法或编码装置或解码装置，其特征在于：所述部分已重构匹配块是满足Xc－Nx<Xr≤Xc并且Yc－Ny<Yr≤Yc但其中两个等号不能同时成立的匹配块；把所述部分已重构匹配块的数值P(X,Y)赋值予所述被匹配块P(X,Y)的赋值语句包括但不限于下列赋值语句之任一或其组合：
    for(y＝0；y<Ny；++y)
    for(x＝0；x<Nx；++x)
    P(Xc+x,Yc+y)＝P(Xr+x,Yr+y)
    或者
    for(y＝0；y<Ny；++y)
    for(x＝0；x<Nx；++x)
    P(Xc+Nx–1–x,Yc+y)＝P(Xr+Nx–1–x,Yr+y)
    或者
    for(x＝0；x<Nx；++x)
    for(y＝0；y<Ny；++y)
    P(Xc+x,Yc+y)＝P(Xr+x,Yr+y)
    或者
    for(x＝0；x<Nx；++x)
    for(y＝0；y<Ny；++y)
    P(Xc+x,Yc+Ny–1–y)＝P(Xr+x,Yr+Ny–1–y)。
  21. 根据权利要求20所述的编码方法或解码方法或编码装置或解码装置，其特征在于：所述部分已重构匹配块是满足Xr＝Xc并且Yc－Ny<Yr<Yc的上下部分自匹配的垂直匹配条或者满足Xc－Nx<Xr<Xc并且Yr＝Yc的左右部分自匹配的水平匹配条；把所述部分已重构匹配条的数值P(X,Y)赋值予所述被匹配条P(X,Y)的赋值语句包括但不限于：
    上下部分自匹配的赋值语句
    for(y＝0；y<Ny；++y)
    P(Xc,Yc+y)＝P(Xr,Yr+y)
    左右部分自匹配的赋值语句
    for(x＝0；x<Nx；++x)
    P(Xc+x,Yc)＝P(Xr+x,Yr)。
  22. 根据权利要求19所述的编码方法或解码方法或编码装置或解码装置，其特征在于：所述匹配块是匹配串，像素样值的坐标(X,Y)成为一个1维地址K，地址为K的像素样值P的数值是P(K)；一个长度为L的被匹配串的第一个像素样值的地址是Kc，而一个匹配串(必定有与被匹配串完全相同的长度)的第一个像素样值的地址是Kr；所述部分已重构匹配块是满足Kc－L<Kr<Kc的部分已重构匹配串；把所述部分已重构匹配串的数值P(K)赋值予所述被匹配串P(K)的赋值语句包括但不限于：
    for(k＝0；k<L；++k)
    P(Kc+k)＝P(Kr+k)。
  23. 根据权利要求1至18中任一权利要求或权利要求20或21或22所述的编码方法或解码方法或编码装置或解码装置，其特征在于所述编码块或解码块或参考块或预测块是(但不限于)下列情况之一：
    编码单元(Coding Unit简称CU)
    或
    预测单元(Prediction Unit简称PU)
    或
    最大编码单元(Largest Coding Unit简称LCU)
    或
    编码树单元(Coding Tree Unit简称CTU)
    或
    由像素样值组成的一个矩形区域
    或
    由像素样值组成的一个正方形区域
    或
    由像素样值组成的一个三边形区域
    或
    由像素样值组成的一个四边形区域
    或
    由像素样值组成的一个多边形区域
    或
    由像素样值组成的一个垂直条形区域(宽度为1的矩形区域)
    或
    由像素样值组成的一个水平条形区域(高度为1的矩形区域)
    或
    由像素样值组成的一个任意角度条形区域(区域中任一像素样值没有上下相邻的像素样值或者没有左右相邻的像素样值的线条形区域)
    或
    由像素样值组成的一个串形区域
    或
    由像素样值组成的一个由一个或多个条形区域构成的区域。
PCT/CN2014/086054 2013-09-07 2014-09-05 一种使用块匹配的图像压缩方法和装置 Ceased WO2015032350A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US14/917,026 US20170155899A1 (en) 2013-09-07 2014-09-05 Image compression method and apparatus using matching

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201310402757 2013-09-07
CN201310402757.9 2013-09-07

Publications (1)

Publication Number Publication Date
WO2015032350A1 true WO2015032350A1 (zh) 2015-03-12

Family

ID=52627827

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2014/086054 Ceased WO2015032350A1 (zh) 2013-09-07 2014-09-05 一种使用块匹配的图像压缩方法和装置

Country Status (3)

Country Link
US (1) US20170155899A1 (zh)
CN (1) CN104427338B (zh)
WO (1) WO2015032350A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106375769A (zh) * 2016-08-31 2017-02-01 苏睿 图像特征搜索方法和装置

Families Citing this family (21)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP6359101B2 (ja) 2013-10-14 2018-07-18 マイクロソフト テクノロジー ライセンシング，エルエルシー ビデオ及び画像の符号化及び復号のためのイントラブロックコピー予測モードの特徴
US10506254B2 (en) 2013-10-14 2019-12-10 Microsoft Technology Licensing, Llc Features of base color index map mode for video and image coding and decoding
CN105659602B (zh) 2013-10-14 2019-10-08 微软技术许可有限责任公司 用于视频和图像编码的帧内块复制预测模式的编码器侧选项
CN105917650B (zh) 2014-01-03 2019-12-24 微软技术许可有限责任公司 视频和图像编/解码的方法、计算设备及计算机可读介质
US11284103B2 (en) 2014-01-17 2022-03-22 Microsoft Technology Licensing, Llc Intra block copy prediction with asymmetric partitions and encoder-side search patterns, search ranges and approaches to partitioning
US10542274B2 (en) 2014-02-21 2020-01-21 Microsoft Technology Licensing, Llc Dictionary encoding and decoding of screen content
CN105493505B (zh) 2014-06-19 2019-08-06 微软技术许可有限责任公司 统一的帧内块复制和帧间预测模式
CA2959682C (en) * 2014-09-30 2022-12-06 Microsoft Technology Licensing, Llc Rules for intra-picture prediction modes when wavefront parallel processing is enabled
WO2016192055A1 (zh) * 2015-06-03 2016-12-08 富士通株式会社 使用预测信息的图像编码方法、装置以及图像处理设备
EP3308540B1 (en) 2015-06-09 2020-04-15 Microsoft Technology Licensing, LLC Robust encoding/decoding of escape-coded pixels in palette mode
CN106254878B (zh) * 2015-06-14 2020-06-12 同济大学 一种图像编码及解码方法、图像处理设备
KR102856248B1 (ko) * 2016-02-16 2025-09-04 삼성전자주식회사 비디오 부호화 방법 및 장치, 그 복호화 방법 및 장치
CN107770553B (zh) * 2016-08-21 2023-06-27 上海天荷电子信息有限公司 采用多类匹配参数及参数存储地址的数据压缩方法和装置
US10827186B2 (en) * 2016-08-25 2020-11-03 Intel Corporation Method and system of video coding with context decoding and reconstruction bypass
CN106815816A (zh) * 2017-01-15 2017-06-09 四川精目科技有限公司 一种rbf插值高速相机压缩图像重建方法
TWI632527B (zh) 2017-11-22 2018-08-11 東友科技股份有限公司 影像擷取與輸出方法
US11509919B2 (en) * 2018-10-17 2022-11-22 Tencent America Reference sample memory size restrictions for intra block copy
CN111447454B (zh) * 2020-03-30 2022-06-07 浙江大华技术股份有限公司 编码方法及其相关装置
CN113727108B (zh) * 2020-05-26 2024-03-01 腾讯科技（深圳）有限公司 视频解码方法、视频编码方法及相关设备
CN112055219B (zh) * 2020-08-05 2021-08-31 浙江大华技术股份有限公司 一种串匹配预测方法、装置及计算机可读存储介质
US11627328B2 (en) * 2020-10-16 2023-04-11 Tencent America LLC Method and apparatus for video coding

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20030202588A1 (en) * 2002-04-29 2003-10-30 Divio, Inc. Intra-prediction using intra-macroblock motion compensation
JP2006311603A (ja) * 2006-06-26 2006-11-09 Toshiba Corp 動画像符号化方法と装置及び動画像復号化方法と装置
CN102282851A (zh) * 2009-01-15 2011-12-14 瑞萨电子株式会社 图像处理装置、解码方法、帧内解码装置、帧内解码方法以及帧内编码装置
EP2615832A1 (en) * 2012-01-13 2013-07-17 Thomson Licensing Method and device for encoding a block of an image and corresponding reconstructing method and device

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR101432775B1 (ko) * 2008-09-08 2014-08-22 에스케이텔레콤 주식회사 서브블록 내 임의 화소를 이용한 영상 부호화/복호화 방법 및 장치
US20100246675A1 (en) * 2009-03-30 2010-09-30 Sony Corporation Method and apparatus for intra-prediction in a video encoder
KR20120140181A (ko) * 2011-06-20 2012-12-28 한국전자통신연구원 화면내 예측 블록 경계 필터링을 이용한 부호화/복호화 방법 및 그 장치
KR20130049523A (ko) * 2011-11-04 2013-05-14 오수미 인트라 예측 블록 생성 장치
US9491458B2 (en) * 2012-04-12 2016-11-08 Qualcomm Incorporated Scalable video coding prediction with non-causal information

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20030202588A1 (en) * 2002-04-29 2003-10-30 Divio, Inc. Intra-prediction using intra-macroblock motion compensation
JP2006311603A (ja) * 2006-06-26 2006-11-09 Toshiba Corp 動画像符号化方法と装置及び動画像復号化方法と装置
CN102282851A (zh) * 2009-01-15 2011-12-14 瑞萨电子株式会社 图像处理装置、解码方法、帧内解码装置、帧内解码方法以及帧内编码装置
EP2615832A1 (en) * 2012-01-13 2013-07-17 Thomson Licensing Method and device for encoding a block of an image and corresponding reconstructing method and device

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106375769A (zh) * 2016-08-31 2017-02-01 苏睿 图像特征搜索方法和装置
CN106375769B (zh) * 2016-08-31 2019-04-30 西安万像电子科技有限公司 图像特征搜索方法和装置、存储介质及处理器

Also Published As

Publication number Publication date
CN104427338A (zh) 2015-03-18
US20170155899A1 (en) 2017-06-01
CN104427338B (zh) 2019-11-05

Similar Documents

Publication Publication Date Title
WO2015032350A1 (zh) 一种使用块匹配的图像压缩方法和装置
CN105704491B (zh) 图像编码方法、解码方法、编码装置和解码装置
CN112383780B (zh) 点匹配参考集和索引来回扫描串匹配的编解码方法和装置
CN111800641B (zh) 同模式采用不同种类重构像素的图像编码解码方法和装置
CN105491376B (zh) 图像编码、解码方法及装置
US20170006285A1 (en) Method and Device for Coding Image and Method and Device for Decoding Image
CN113784124B (zh) 使用多种形状子块进行精细划分的块匹配编码和解码方法
KR101946598B1 (ko) 이미지 코딩, 디코딩 방법 및 장치
WO2015120823A1 (zh) 使用多种形式的参考像素存储空间的图像压缩方法和装置
CN104935945B (zh) 扩展参考像素样值集的图像编码或解码方法
CN106254865B (zh) 一种图像编码及解码方法、图像处理设备
KR102532391B1 (ko) 영상 부호화 방법과 장치 및 영상 복호화 방법과 장치
EP3310059A1 (en) Image encoding and decoding method, image processing device, and computer storage medium
WO2016202189A1 (zh) 图像编码及解码方法、图像处理设备、计算机存储介质
CN105992003A (zh) 依据排序或频度对调色板颜色编号的图像压缩方法和装置
CN105828080B (zh) 图像编解码方法及装置
CN106303534B (zh) 多种索引串与像素串融合复制方式的图像压缩方法和装置
WO2016197893A1 (zh) 图像编码及解码方法、图像处理设备、计算机存储介质
CN112565750A (zh) 一种视频编码方法、电子设备和存储介质
CN104363459A (zh) 一种适用于hevc标准中帧内预测的参考像素的硬件填充方法
WO2016119666A1 (zh) 图像编解码方法及装置
EP3252706A1 (en) Image processing method and apparatus
WO2016119667A1 (zh) 图像处理方法及装置

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 14842407

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

WWE Wipo information: entry into national phase

Ref document number: 14917026

Country of ref document: US

122 Ep: pct application non-entry in european phase

Ref document number: 14842407

Country of ref document: EP

Kind code of ref document: A1

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 07/09/2016)

122 Ep: pct application non-entry in european phase

Ref document number: 14842407

Country of ref document: EP

Kind code of ref document: A1