WO2021031225A1 - 一种运动矢量导出方法、装置及电子设备 - Google Patents

一种运动矢量导出方法、装置及电子设备 Download PDF

Info

Publication number
WO2021031225A1
WO2021031225A1 PCT/CN2019/102748 CN2019102748W WO2021031225A1 WO 2021031225 A1 WO2021031225 A1 WO 2021031225A1 CN 2019102748 W CN2019102748 W CN 2019102748W WO 2021031225 A1 WO2021031225 A1 WO 2021031225A1
Authority
WO
WIPO (PCT)
Prior art keywords
motion vector
block
current block
coding unit
filtered
Prior art date
Application number
PCT/CN2019/102748
Other languages
English (en)
French (fr)
Inventor
王荣刚
谢熙
范逵
高文
Original Assignee
北京大学深圳研究生院
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 北京大学深圳研究生院 filed Critical 北京大学深圳研究生院
Publication of WO2021031225A1 publication Critical patent/WO2021031225A1/zh
Priority to US17/645,698 priority Critical patent/US11997284B2/en

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
    • H04N19/117—Filters, e.g. for pre-processing or post-processing
    • 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
    • H04N19/102—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
    • H04N19/109—Selection of coding mode or of prediction mode among a plurality of temporal predictive coding modes
    • 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
    • H04N19/50—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
    • H04N19/513—Processing of motion vectors
    • 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
    • H04N19/80—Details of filtering operations specially adapted for video compression, e.g. for pixel interpolation
    • 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
    • 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
    • H04N19/157—Assigned coding mode, i.e. the coding mode being predefined or preselected to be further used for selection of another element or parameter
    • H04N19/159—Prediction type, e.g. intra-frame, inter-frame or bidirectional frame prediction

Definitions

  • This specification relates to the technical field of video coding and decoding, and in particular to a method, device and electronic equipment for deriving a motion vector.
  • inter-frame prediction technology can be used to reduce the redundancy between video frames by using the time correlation between each video frame.
  • the Skip/Direct mode directly derives the current coding unit by borrowing the motion information of the neighboring blocks that have been encoded in the spatial domain and the motion information in the encoded image in the time domain. Sports information. Therefore, although the Skip/Direct mode improves the coding efficiency to a certain extent, this method of deriving motion information using only the motion information in the coded neighboring blocks or images reduces the accuracy of inter-frame prediction.
  • Skip The /Direct mode ultimately derives the motion information of the entire current coding unit, which further reduces the accuracy of inter-frame prediction and affects coding efficiency.
  • the object of the present invention is to provide a method, device and electronic device for deriving a motion vector to solve the existing method of deriving motion information only by using the motion information in the coded neighboring blocks or images, which reduces The accuracy of inter-frame prediction affects coding efficiency.
  • An embodiment of this specification provides a method for deriving a motion vector, which includes:
  • a predetermined inter prediction mode using the spatial motion vector and temporal motion vector prediction of the filtered neighboring block and the coordinate position of the current block in the coding unit, it is determined that the current block is on its four sides Reference motion vector in the side direction;
  • An embodiment of this specification provides a motion vector derivation device, the device includes:
  • An acquisition module for acquiring spatial motion vector and temporal motion vector prediction of neighboring blocks of the coding unit in a predetermined direction
  • a filtering module configured to perform a filtering operation on the spatial motion vector and time domain motion vector prediction to obtain the filtered spatial motion vector and time domain motion vector prediction of the neighboring block;
  • the determining module is configured to determine the current block by using the spatial motion vector and time domain motion vector prediction of the filtered neighboring block and the coordinate position of the current block in the coding unit according to a predetermined inter prediction mode Reference motion vectors in its four side directions;
  • the derivation module is configured to derive the motion vector of the current block according to the reference motion vector and the coordinate position of the current block in the coding unit.
  • An electronic device provided by an embodiment of this specification includes a memory, a processor, and a computer program stored in the memory and capable of running on the processor, and the processor implements the aforementioned motion vector derivation method when the program is executed.
  • the present invention obtains the spatial motion vector and time domain motion vector prediction of the adjacent block of the coding unit in the predetermined direction; performs filtering operation on the spatial motion vector and the time domain motion vector prediction to obtain the filtered spatial motion vector of the adjacent block And time-domain motion vector prediction; according to a predetermined inter-frame prediction mode, using the spatial motion vector and time-domain motion vector prediction of the filtered neighboring block and the coordinate position of the current block in the coding unit, it is determined that the current block is in its Reference motion vectors in the four side directions; according to the reference motion vector and the coordinate position of the current block in the coding unit, the motion vector of the current block is derived. Based on the solution of the present invention, the accuracy of inter-frame prediction can be improved, and the coding efficiency can be further improved.
  • FIG. 1 is a schematic flowchart of a method for deriving a motion vector provided by an embodiment of this specification
  • FIG. 2 is a schematic diagram of the position structure of neighboring blocks for obtaining spatial motion vector and temporal motion vector prediction according to an embodiment of this specification;
  • FIG. 3 is a schematic diagram of the principle of using five inter-frame prediction modes to determine the reference motion vector of the current block in its four side directions according to an embodiment of the present specification
  • Fig. 4 is a schematic structural diagram of a motion vector derivation device provided by an embodiment of this specification.
  • Commonly used methods of compressing video data include HEVC, AVS, H.264/MPEG-4AVC, etc. According to these methods, a picture is divided into multiple macroblocks to encode the image, and prediction blocks are generated using inter-frame prediction or intra-frame prediction , So as to encode the corresponding macro block.
  • the difference between the initial block and the prediction block is transformed to generate a transformed block, and the transformed block is quantized using a quantization parameter and a quantization matrix.
  • the quantized coefficients of the quantized block are scanned by a predetermined scanning mode and then entropy coding is performed.
  • the quantization parameter is adjusted for each macroblock, and the previous quantization parameter is used to encode it.
  • the motion vector of the current coded block It can be derived with reference to the motion vectors of neighboring blocks in the time domain and the space domain to save the bit rate overhead of directly transmitting the motion vectors and improve the coding efficiency.
  • the current coding unit is coded in multiple modes, such as Skip/Direct mode, merge mode, etc. These modes borrow motion information from neighboring blocks that have been coded in the spatial domain and the coded image in the time domain.
  • the motion information directly derives the motion information of the current coding unit instead of obtaining motion information through motion estimation, thus reducing the accuracy of inter-frame prediction; in addition, the existing model ultimately derives the motion information of the entire current coding unit. It cannot be accurate to the sub-blocks in each coding unit, which further reduces the accuracy of inter-frame prediction, thereby affecting coding efficiency.
  • FIG. 1 is a schematic flowchart of a method for deriving a motion vector according to an embodiment of the present invention.
  • the method may specifically include the following steps:
  • step S110 the spatial motion vector and temporal motion vector prediction of neighboring blocks of the coding unit in a predetermined direction are obtained.
  • a coding unit may refer to dividing a frame of image into several rectangular blocks of a certain size that do not overlap with each other. Each block is the largest coding unit, and each largest Coding units can be divided into coding units of different sizes from 64*64 to 8*8; coding units have separate prediction mode information, coefficients, etc.
  • obtaining the spatial motion vector of the neighboring block of the coding unit in a predetermined direction may include the following process:
  • Scaling the motion vectors of adjacent blocks on the left and upper sides of the coding unit to obtain the spatial motion vectors of the adjacent blocks specifically, scaling the forward motion vector of the adjacent block to the first frame of the forward reference frame list , Scale the backward motion vector of the adjacent block to the first frame of the backward reference frame list, and use the scaled motion vector as the spatial motion vector of the adjacent block.
  • neighboring blocks refer to coding blocks adjacent to the coding unit in the four directions of left, right, up and down.
  • the neighboring block does not belong to the coding block inside the current coding unit, and the size of the neighboring block is the same as the size of the current block in the coding unit. the same.
  • the neighboring block from which the spatial motion vector is derived can be each neighboring block on the left and upper side of the current coding unit.
  • the spatial motion vector of the neighboring block is obtained by comparing the motion vector of the neighboring block.
  • the result of scaling that is, the spatial motion vector is the result of the scaling of the motion vector.
  • the motion vector here refers to the motion vector of the neighboring blocks that have been encoded on the left and above, and the motion vector can also be represented by motion information.
  • the specific scaling process may include scaling the forward motion vector of the adjacent block to the first frame of the forward reference frame list, scaling the backward motion vector of the adjacent block to the first frame of the backward reference frame list, and
  • the spatial motion vector of neighboring blocks can be obtained by calculating the scaling results of the forward motion vector and the backward motion vector; for the neighboring blocks that cannot obtain the motion vector, find the space from the neighboring blocks available from the spatial motion vector.
  • the neighboring block with the nearest position uses its spatial motion vector after scaling as its own spatial motion vector.
  • the video coding technology includes two reference frame lists, namely the forward reference frame list and the backward reference frame list. These two reference frame lists store the encoded and reconstructed images in a certain order. These images may be The inter prediction mode of the current coding unit is selected as the reference frame.
  • the first frame may refer to an image whose reference frame index is 0 in the reference frame list.
  • obtaining the temporal motion vector prediction of neighboring blocks of a coding unit in a predetermined direction may include the following process:
  • the neighboring block from which the time domain motion vector prediction is derived may be each neighboring block on the right and the lower side of the current coding unit.
  • the time domain motion vector prediction of the neighboring block is obtained from other blocks in the time domain.
  • the motion information of the block at the same coordinate position as the coding unit where the adjacent block is located is obtained, and the result of the scaling of the motion information is used as the temporal motion vector prediction of the adjacent block.
  • the block with the same coordinate position as the current neighboring block in other encoded images in the time domain is called the co-located block
  • the temporal motion vector prediction is to derive the motion of the neighboring block by scaling the motion information of the co-located block. information.
  • the prediction is its own time domain motion vector prediction.
  • the figure shows a schematic diagram of the location structure of neighboring blocks for obtaining spatial motion vectors and temporal motion vector predictions; among them, the left side of Figure 2 is the neighboring blocks and coding units for obtaining spatial motion vectors It can be seen that the neighboring blocks for obtaining the spatial motion vector are the neighboring blocks on the left and upper side of the coding unit.
  • the right side in Figure 2 is the positional relationship between the neighboring blocks for obtaining the temporal motion vector prediction and the coding unit.
  • the neighboring blocks for obtaining the temporal motion vector prediction are the neighboring blocks located on the right and lower sides of the coding unit.
  • step S120 a filtering operation is performed on the spatial motion vector and time domain motion vector prediction to obtain the filtered spatial motion vector and time domain motion vector prediction of the neighboring block.
  • filtering the spatial motion vector and temporal motion vector prediction of the neighboring blocks derived in step S110 may specifically include the following two steps:
  • the first step is to perform a filling operation on both ends of the spatial motion vector and time domain motion vector prediction.
  • the specific filling operation includes the following:
  • MvR and MvB respectively represent the temporal motion vector prediction of the right and lower adjacent blocks before filtering
  • MvL and MvT respectively represent the spatial motion vectors of the left and upper adjacent blocks before filtering
  • i and j represent The column coordinates in units of blocks
  • k and l represent row coordinates in units of sub-blocks
  • M and N represent the width and height of the coding unit in units of sub-blocks.
  • the motion vectors other than the outermost adjacent blocks at the two ends may be used during the filtering operation for the spatial motion vector and temporal motion vector prediction of adjacent blocks, and the motion vectors of these blocks may not Therefore, the motion vector of the neighboring block on the edge can be used as filling.
  • the second step is to filter the spatial motion vector and temporal motion vector prediction according to the following formula:
  • Mvt[x] (3*MvT[i 0 ]+8*MvT[i 1 ]+10*MvT[i 2 ]+8*MvT[i 3 ]+3*MvT[i 4 ])>>5
  • Mvl[y] (3*MvL[j 0 ]+8*MvL[j 1 ]+10*MvL[j 2 ]+8*MvL[j 3 ]+3*MvL[j 4 ])>>5
  • Mvb[x] (3*MvB[i 0 ]+8*MvB[i 1 ]+10*MvB[i 2 ]+8*MvB[i 3 ]+3*MvB[i 4 ])>>5
  • Mvr[y] (3*MvR[j 0 ]+8*MvR[j 1 ]+10*MvR[j 2 ]+8*MvR[j 3 ]+3*MvR[j 4 ])>>5
  • MvR and MvB respectively represent the temporal motion vector prediction of the right and lower adjacent blocks before filtering
  • Mvr, Mvb respectively represent the temporal motion vector prediction of the right and lower adjacent blocks after filtering
  • MvL, MvT respectively Represents the spatial motion vector of the adjacent block on the left and upper side before filtering
  • Mvl, Mvt represent the spatial motion vector of the adjacent block on the left and upper side after filtering, respectively
  • x and y represent the sub-block of the current block in the coding unit
  • y 0 represents the first row sub-block in the current coding unit
  • i and j represent the column coordinates in the unit of sub-block
  • M, N Indicates the width and height of the coding unit in units of sub-blocks.
  • step S130 according to a predetermined inter prediction mode, the current block is determined by using the spatial motion vector and time domain motion vector prediction of the filtered neighboring block and the coordinate position of the current block in the coding unit The reference motion vector in its four side directions.
  • the embodiments of this specification propose 5 new inter prediction modes, including: the first inter prediction mode, the second inter prediction mode, the third inter prediction mode, the fourth inter prediction mode, and the fifth inter prediction mode mode.
  • these five inter-frame prediction modes can be called SLTCRB mode, SLCB mode, STCR mode, SLT mode, and CRB mode, respectively.
  • this figure shows a schematic diagram of the principle of using five inter-frame prediction modes to determine the reference motion vector of the current block in its four side directions in the embodiment of this specification; see FIG. 3,
  • determining the reference motion vector of the current block in its four side directions may specifically include the following content:
  • the left neighboring block and the right neighboring block in the abscissa direction of the current block are selected, and the upper neighboring block in the ordinate direction of the current block is selected
  • the lower neighboring block, the spatial motion vector of the filtered left neighboring block and the upper neighboring block, and the filtered time domain motion vector of the right neighboring block and the lower neighboring block are respectively used as the current Reference motion vectors of the block in its four side directions;
  • the left neighboring block in the abscissa direction of the current block and the lower neighboring block in the ordinate direction are selected, and the filtered left neighboring block
  • the spatial motion vector and the filtered temporal motion vector prediction of the lower neighboring block are respectively used as the reference motion vector of the current block in the left and lower directions; and the filtered coding unit’s lower right neighboring block’s
  • the temporal motion vector prediction and the spatial motion vector of the uppermost adjacent block on the left side of the coding unit after filtering are used as the reference motion vector of the current block in the right and upper directions respectively;
  • the third inter-frame prediction mode (STCR mode)
  • the right adjacent block in the abscissa direction of the current block and the upper adjacent block in the ordinate direction are selected, and the The temporal motion vector prediction and the spatial motion vector of the upper neighboring block after filtering are used as the reference motion vector of the current block in the right and upper direction respectively;
  • the spatial motion vector and the temporal motion vector prediction of the lowermost adjacent block on the right side of the filtered coding unit are respectively used as the reference motion vector of the current block in the left and lower directions;
  • the left adjacent block in the abscissa direction of the current block and the upper adjacent block in the ordinate direction are selected, and the filtered left adjacent block
  • the spatial motion vector and the spatial motion vector of the upper neighboring block are used as the reference motion vector of the current block in the left and upper directions respectively;
  • the spatial motion vector of the uppermost right neighboring block of the filtered coding unit and the filtering The spatial motion vector of the lowermost adjacent block on the left side of the subsequent coding unit is used as the reference motion vector of the current block in the right and lower directions respectively;
  • the right adjacent block in the abscissa direction of the current block and the lower adjacent block in the ordinate direction are selected, and the filtered right adjacent block
  • the time domain motion vector prediction and the time domain motion vector prediction of the lower neighboring block are respectively used as the reference motion vector of the current block in the right and lower directions; and the time domain of the lower leftmost neighboring block of the filtered coding unit
  • the domain motion vector prediction and the temporal motion vector prediction of the uppermost adjacent block on the right side of the coding unit after filtering are respectively used as the reference motion vector of the current block in the left and upper directions.
  • these five new inter-frame prediction modes all use bilinear interpolation to combine spatial motion information and temporal motion information in the neighborhood of the current coding unit to derive each sub-block in the current coding unit
  • the reference motion vector Therefore, in practical applications, taking the current block with coordinates (x, y) as an example, for each of the above modes, four motion vectors can be selected as reference motion vectors in the left, right, up, and down directions according to the following formula:
  • vl, vr, vt and vb respectively represent the reference motion vectors in the left, right, up and down directions of the current block in the coding unit;
  • Mvr, Mvb respectively represent the temporal motion vector prediction of the right and lower adjacent blocks after filtering;
  • Mvl and Mvt respectively represent the spatial motion vectors of the adjacent blocks on the left and upper side after filtering;
  • x and y represent the coordinates of the current block in the coding unit in units of sub-blocks;
  • M and N indicate the width and height of the coding unit in units of sub-blocks.
  • step S140 the motion vector of the current block is derived according to the reference motion vector and the coordinate position of the current block in the coding unit.
  • the following formula may be used to derive the motion vector of the current block according to the reference motion vector determined in step S130 in combination with the coordinate position of the current block in the coding unit, specifically:
  • vl, vr, vt, and vb respectively represent the reference motion vector of the current block in the left, right, up, and down directions;
  • x and y represent the coordinates of the current block in the coding unit in units of sub-blocks;
  • M , N represents the width and height of the coding unit in units of sub-blocks;
  • v h represents the horizontal motion vector of the current block;
  • v v represents the vertical motion vector of the current block;
  • v[x][y] represents the motion vector of the current block.
  • the above inter prediction mode can also be marked in the code stream.
  • the specific marking process may include the following :
  • the 5 new inter-frame prediction modes proposed in this specification form a mode candidate list.
  • a 1-bit flag needs to be transmitted to the code stream to identify whether the above-mentioned inter-frame prediction method is selected.
  • the decoder needs to decode the 1-bit flag to determine whether the above inter prediction method is used. If the above inter prediction method is used, the decoder also needs a decoding index. The index is used to determine which of the above 5 new inter prediction modes belongs to One kind.
  • the embodiment of this specification also provides a process for making a rate-distortion optimization decision on each coding unit at the encoding end, which specifically includes the following content:
  • step S110 for each inter prediction direction, derive the spatial motion vector of each adjacent block on the left and above of the current coding unit and the temporal motion vector prediction of each adjacent block on the right and below of the current coding unit .
  • step S120 the spatial motion vector and the temporal motion vector prediction are filtered.
  • each mode derives the motion vector of each sub-block in the current coding unit, and each sub-block is derived from the motion vector to predict the block.
  • step S110 for each inter-frame prediction direction, the spatial motion vector of each adjacent block on the left and above of the current coding unit and the temporal motion vector prediction of each adjacent block on the right and below of the current coding unit are derived .
  • step S120 the spatial motion vector and the temporal motion vector prediction are filtered.
  • step S130-step 140 select the corresponding new inter prediction mode, derive the motion vector of each sub-block in the current coding unit, and derive the prediction block for each sub-block according to the motion vector.
  • Fig. 4 is a motion vector derivation device provided by the embodiment of this specification.
  • the device mainly includes:
  • the obtaining module 401 is configured to obtain the spatial motion vector and the temporal motion vector prediction of the neighboring blocks of the coding unit in a predetermined direction;
  • the filtering module 402 is configured to perform a filtering operation on the spatial motion vector and the temporal motion vector prediction to obtain the filtered spatial motion vector and the temporal motion vector prediction of the adjacent block;
  • the determining module 403 is configured to use the spatial motion vector and temporal motion vector prediction of the filtered neighboring block and the coordinate position of the current block in the coding unit according to a predetermined inter-frame prediction mode to determine the current Reference motion vectors of the block in its four side directions;
  • the derivation module 404 is configured to derive the motion vector of the current block according to the reference motion vector and the coordinate position of the current block in the coding unit.
  • the obtaining module 401 is further configured to:
  • the forward motion vector of the adjacent block is scaled to the first frame of the forward reference frame list
  • the backward motion vector of the adjacent block is scaled to the first frame of the backward reference frame list
  • the scaled The motion vector is used as the spatial motion vector of the adjacent block.
  • the obtaining module 401 is further configured to:
  • the filtering module 402 is specifically configured to: fill the two ends of the spatial motion vector and time domain motion vector prediction, and predict the spatial motion vector and time domain motion vector according to the following formula Filtering:
  • Mvt[x] (3*MvT[i 0 ]+8*MvT[i 1 ]+10*MvT[i 2 ]+8*MvT[i 3 ]+3*MvT[i 4 ])>>5
  • Mvl[y] (3*MvL[j 0 ]+8*MvL[j 1 ]+10*MvL[j 2 ]+8*MvL[j 3 ]+3*MvL[j 4 ])>>5
  • Mvb[x] (3*MvB[i 0 ]+8*MvB[i 1 ]+10*MvB[i 2 ]+8*MvB[i 3 ]+3*MvB[i 4 ])>>5
  • Mvr[y] (3*MvR[j 0 ]+8*MvR[j 1 ]+10*MvR[j 2 ]+8*MvR[j 3 ]+3*MvR[j 4 ])>>5
  • MvR and MvB respectively represent the temporal motion vector prediction of the right and lower adjacent blocks before filtering
  • Mvr, Mvb respectively represent the temporal motion vector prediction of the right and lower adjacent blocks after filtering
  • MvL, MvT respectively Represents the spatial motion vector of the adjacent block on the left and upper side before filtering
  • Mvl, Mvt represent the spatial motion vector of the adjacent block on the left and upper side after filtering, respectively
  • x and y represent the sub-block of the current block in the coding unit Is the coordinate in the unit
  • i and j represent the column coordinates in the unit of sub-block
  • M and N represent the width and height of the coding unit in the unit of sub-block.
  • the inter prediction mode includes a first inter prediction mode, a second inter prediction mode, a third inter prediction mode, a fourth inter prediction mode, and a fifth inter prediction mode, so
  • the determining module 403 is specifically used for:
  • the left neighboring block and the right neighboring block in the abscissa direction of the current block are selected, and the upper neighboring block in the ordinate direction of the current block is selected
  • the lower neighboring block, the filtered spatial motion vector of the left neighboring block and the upper neighboring block, and the filtered time domain motion vector of the right neighboring block and the lower neighboring block Prediction respectively as the reference motion vector of the current block in its four side directions;
  • the left neighboring block in the abscissa direction and the lower neighboring block in the ordinate direction of the current block are selected, and the filtered left neighboring block
  • the spatial motion vector and the filtered temporal motion vector prediction of the lower neighboring block are respectively used as the reference motion vector of the current block in the left and lower directions; and the filtered coding unit lower side
  • the temporal motion vector prediction of the rightmost neighboring block and the filtered spatial motion vector of the uppermost neighboring block on the left side of the coding unit are respectively used as the reference motion vector of the current block in the right and upper directions;
  • the right adjacent block in the abscissa direction and the upper adjacent block in the ordinate direction of the current block are selected, and the filtered right adjacent block
  • the temporal motion vector prediction and the filtered spatial motion vector of the upper neighboring block are respectively used as the reference motion vector of the current block in the right and upper directions
  • the filtered coding unit upper side The spatial motion vector of the leftmost adjacent block and the filtered temporal motion vector prediction of the lowermost adjacent block on the right side of the coding unit are respectively used as the reference motion vectors of the current block in the left and lower directions;
  • the left neighboring block in the abscissa direction and the upper neighboring block in the ordinate direction of the current block are selected, and the filtered left neighboring block
  • the spatial motion vector and the spatial motion vector of the upper neighboring block are respectively used as the reference motion vectors of the current block in the left and upper directions; and the filtered spatial domain of the uppermost neighboring block of the coding unit
  • the motion vector and the filtered spatial motion vector of the neighboring block on the left side of the coding unit are respectively used as the reference motion vector of the current block in the right and lower directions;
  • the right adjacent block in the abscissa direction and the lower adjacent block in the ordinate direction of the current block are selected, and the filtered right adjacent block
  • the temporal motion vector prediction and the temporal motion vector prediction of the lower neighboring block are respectively used as the reference motion vectors of the current block in the right and lower directions; and the lower leftmost side of the coding unit after filtering is compared
  • the temporal motion vector prediction of the neighboring block and the filtered temporal motion vector prediction of the uppermost neighboring block on the right side of the coding unit are respectively used as the reference motion vector of the current block in the left and upper directions.
  • the deriving module 404 is specifically configured to derive the motion vector of the current block according to the following formula:
  • vl, vr, vt, and vb respectively represent the reference motion vector of the current block in the left, right, up, and down directions;
  • x and y represent the coordinates of the current block in the coding unit in units of sub-blocks;
  • M , N represents the width and height of the coding unit in units of sub-blocks;
  • v h represents the horizontal motion vector of the current block;
  • v v represents the vertical motion vector of the current block;
  • v[x][y] represents the motion vector of the current block.
  • the embodiment of the present specification also provides an electronic device including a memory, a processor, and a computer program stored in the memory and capable of running on the processor, and the processor implements the aforementioned method for deriving a motion vector when the program is executed.
  • the device, electronic device, and method provided in the embodiments of this specification are corresponding. Therefore, the device and electronic device also have beneficial technical effects similar to the corresponding method. Since the beneficial technical effects of the method have been described in detail above, here The beneficial technical effects of the corresponding devices and electronic equipment will not be repeated.
  • program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types.
  • the instructions can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network.
  • program modules can be located in local and remote computer storage media including storage devices.

Landscapes

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

Abstract

一种运动矢量导出方法、装置及电子设备。所述方法包括：获取编码单元在预定方向上的相邻块的空域运动矢量和时域运动矢量预测(S110)；对所述空域运动矢量和时域运动矢量预测执行滤波操作，得到滤波后的所述相邻块的空域运动矢量和时域运动矢量预测(S120)；根据预定的帧间预测模式，利用所述滤波后的相邻块的空域运动矢量和时域运动矢量预测以及当前块在所述编码单元内的坐标位置，确定所述当前块在其四个侧边方向上的参考运动矢量(S130)；根据所述参考运动矢量以及所述当前块在所述编码单元内的坐标位置，导出所述当前块的运动矢量(S140)。能够提高帧间预测的准确度，进一步提升编码效率。

Description

一种运动矢量导出方法、装置及电子设备
本申请要求享有2019年8月19日提交的名称为“一种运动矢量导出方法、装置及电子设备”的中国专利申请CN201910766509.X的优先权，其全部内容通过引用并入本文中。
技术领域
本说明书涉及视频编解码技术领域，尤其涉及一种运动矢量导出方法、装置及电子设备。
背景技术
由于视频中连续的若干帧图像之间存在较强时间相关性，所以在进行视频编码时，可以采用帧间预测技术，利用各个视频帧之间的时间相关性来减少视频帧间的冗余，以达到有效去除视频时域冗余的目的。在现有的帧间预测技术中，以Skip/Direct模式为例，Skip/Direct模式是通过借用空域上已编码邻块的运动信息和时域上已编码图像内的运动信息直接导出当前编码单元的运动信息。因此，Skip/Direct模式虽然在一定程度上提高了编码效率，但是这种仅仅利用已编码邻块或图像内的运动信息来导出运动信息的方式，降低了帧间预测的准确度，另外，Skip/Direct模式最终导出的是整个当前编码单元的运动信息，从而进一步降低了帧间预测的准确度，进而影响编码效率。
基于现有技术，需要提供一种能提高帧间预测的准确度，进一步提升编码效率的运动矢量导出方案。
发明内容
有鉴于此，本发明的目的在于提供一种运动矢量导出方法、装置及电子设备，以解决现有技术存在的仅仅利用已编码邻块或图像内的运动信息来导出运动信息的方式，降低了帧间预测的准确度，影响编码效率的问题。
为解决上述技术问题，本说明书实施例是这样实现的：
本说明书实施例提供的一种运动矢量导出方法，所述方法包括：
获取编码单元在预定方向上的相邻块的空域运动矢量和时域运动矢量预测；
对所述空域运动矢量和时域运动矢量预测执行滤波操作，得到滤波后的所述相邻块的空域运动矢量和时域运动矢量预测；
根据预定的帧间预测模式，利用所述滤波后的相邻块的空域运动矢量和时域运动矢量预测以及当前块在所述编码单元内的坐标位置，确定所述当前块在其四个侧边方向上的参考运动矢量；
根据所述参考运动矢量以及所述当前块在所述编码单元内的坐标位置，导出所述当前块的运动矢量。
本说明书实施例提供的一种运动矢量导出装置，所述装置包括：
获取模块，用于获取编码单元在预定方向上的相邻块的空域运动矢量和时域运动矢量预测；
滤波模块，用于对所述空域运动矢量和时域运动矢量预测执行滤波操作，得到滤波后的所述相邻块的空域运动矢量和时域运动矢量预测；
确定模块，用于根据预定的帧间预测模式，利用所述滤波后的相邻块的空域运动矢量和时域运动矢量预测以及当前块在所述编码单元内的坐标位置，确定所述当前块在其四个侧边方向上的参考运动矢量；
导出模块，用于根据所述参考运动矢量以及所述当前块在所述编码单元内的坐标位置，导出所述当前块的运动矢量。
本说明书实施例提供的一种电子设备，包括存储器，处理器及存储在存储 器上并可在处理器上运行的计算机程序，所述处理器执行所述程序时实现上述一种运动矢量导出方法。
本说明书实施例采用的上述至少一个技术方案能够达到以下有益效果：
本发明通过获取编码单元在预定方向上的相邻块的空域运动矢量和时域运动矢量预测；对空域运动矢量和时域运动矢量预测执行滤波操作，得到滤波后的相邻块的空域运动矢量和时域运动矢量预测；根据预定的帧间预测模式，利用所述滤波后的相邻块的空域运动矢量和时域运动矢量预测以及当前块在编码单元内的坐标位置，确定当前块在其四个侧边方向上的参考运动矢量；根据参考运动矢量以及当前块在编码单元内的坐标位置，导出当前块的运动矢量。基于本发明的方案，能够提高帧间预测的准确度，进一步提升编码效率。
附图说明
为了更清楚的说明本发明实施例或现有技术中的技术方案，下面将对实施例或现有技术描述中所需要使用的附图作简单的介绍，显而易见的下面描述中的附图仅仅是本发明的实施例，对于本领域普通人员来讲，在不付出创造性劳动的前提下，还可以根据提供的附图获得其他的附图。
图1是本说明书实施例提供的一种运动矢量导出方法的流程示意图；
图2是本说明书实施例提供的一种获取空域运动矢量和时域运动矢量预测的相邻块的位置结构示意图；
图3是本说明书实施例提供的分别采用5种帧间预测模式确定当前块在其四个侧边方向上的参考运动矢量的原理示意图；
图4是本说明书实施例提供的一种运动矢量导出装置的结构示意图。
具体实施方式
为了使本技术领域的人员更好地理解本说明书中的技术方案，下面将结合 本说明书实施例中的附图，对本说明书实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本申请一部分实施例，而不是全部的实施例。基于本说明书实施例，本领域普通技术人员在没有作出创造性劳动前提下所获得的所有其他实施例，都应当属于本申请保护的范围。
常用的压缩视频数据的方法包括HEVC、AVS和H.264/MPEG-4AVC等，根据这些方法，一幅画面被分成多个宏块以对图像编码，利用帧间预测或帧内预测产生预测块，从而对相应宏块编码。变换初始块和预测块之间的差异以产生变换块，利用量化参数和量化矩阵对变换块进行量化。通过预定扫描模式扫描量化块的量化系数并随后进行熵编码。针对每个宏块调节量化参数，并利用先前的量化参数对其进行编码。
在视频中，时域相邻块和空域相邻块与当前编码块之间在内容上有很强的相关性，并且它们的运动矢量也有很强的相关性，因此，当前编码块的运动矢量可以参考时域和空域相邻块的运动矢量进行导出以节省直接传输运动矢量的码率开销，提高编码效率。
在上述视屏编码方法中，包括采用多种模式编码当前的编码单元，例如Skip/Direct模式、merge模式等，这些模式通过借用空域上已编码邻块的运动信息和时域上已编码图像内的运动信息直接导出当前编码单元的运动信息，而不是通过做运动估计来获取运动信息，因而降低了帧间预测的准确度；另外，现有的模式最终导出的是整个当前编码单元的运动信息，而无法精确到每个编码单元内的子块，从而进一步降低了帧间预测的准确度，进而影响编码效率。
本说明书以下实施例是以视屏编码，尤其是以视屏编码的帧间预测过程作为应用场景来说的，但是本说明书实施例不限于以下应用场景，以下场景不构成对本说明书实施例应用范围的限定。
图1为本发明实施例提供的一种运动矢量导出方法的流程示意图。该方法具体可以包括以下步骤：
在步骤S110中，获取编码单元在预定方向上的相邻块的空域运动矢量和时 域运动矢量预测。
在本说明书一个或多个实施例中，编码单元(CU,coding unit)可以是指将一帧图像分成若干个一定大小互不重叠的矩形块，每一个块即为最大编码单元，每个最大编码单元又可以分为从64*64到8*8不同大小的编码单元；编码单元拥有单独的预测模式信息，系数等。
在本说明书一个或多个实施例中，获取编码单元在预定方向上的相邻块的空域运动矢量，可以包括以下过程：
对位于编码单元左侧和上侧的相邻块的运动矢量进行缩放得到相邻块的空域运动矢量；具体地，将相邻块的前向运动矢量缩放到前向参考帧列表的第一帧，将相邻块的后向运动矢量缩放到后向参考帧列表的第一帧，将缩放后的运动矢量作为相邻块的空域运动矢量。
进一步地，相邻块是指编码单元在左右上下四个方向上相邻的编码块，相邻块不属于当前编码单元内部的编码块，相邻块的大小与编码单元内的当前块的大小相同。在实际应用中，导出空域运动矢量的相邻块可以是当前编码单元在其左侧和上侧方位上的每个相邻块，相邻块的空域运动矢量是通过对相邻块的运动矢量进行缩放得到的，也就是说，空域运动矢量就是运动矢量缩放后的结果。这里的运动矢量是指左侧和上方已经完成编码的相邻块的运动矢量，运动矢量也可以用运动信息来表示。
具体缩放过程可以包括，将相邻块的前向运动矢量缩放到前向参考帧列表的第一帧，将相邻块的后向运动矢量缩放到后向参考帧列表的第一帧，通过对前向运动矢量和后向运动矢量的缩放结果进行计算就可以得到相邻块的空域运动矢量；对于获取不到运动矢量的相邻块，从空域运动矢量可得的相邻块中，寻找空间位置最近的相邻块，将其缩放后的空域运动矢量作为自身的空域运动矢量。
在视频编码技术中包括两个参考帧列表，分别为前向参考帧列表和后向参 考帧列表，这两个参考帧列表按一定顺序存放着已完成编码重建的图像，这些图像有可能会被当前编码单元的帧间预测模式选中作为参考帧。第一帧可以指参考帧列表中参考帧索引为0的图像。
在本说明书一个或多个实施例中，获取编码单元在预定方向上的相邻块的时域运动矢量预测，可以包括以下过程：
获取位于编码单元右侧和下侧的相邻块的时域运动矢量预测；具体地，从前向参考帧列表的第一帧或后向参考帧列表的第一帧中获取与相邻块具有相同坐标位置的编码块的运动矢量，对编码块的运动矢量进行缩放得到相邻块的时域运动矢量预测。
进一步地，导出时域运动矢量预测的相邻块可以是当前编码单元在其右侧和下侧方位上的每个相邻块，相邻块的时域运动矢量预测是通过从时域上其它已编码图像中，获取与相邻块所在编码单元相同坐标位置的块的运动信息，将该运动信息缩放后的结果作为相邻块的时域运动矢量预测。其中，时域上其它已编码图像中与当前相邻块具有相同坐标位置的块被称为co-located块，时域运动矢量预测是通过缩放co-located块的运动信息导出相邻块的运动信息。对于获取不到时域运动矢量预测的相邻块，从与当前编码单元相邻且时域运动矢量预测可得的相邻块中，寻找空间位置最近的相邻块，将其时域运动矢量预测作为自身的时域运动矢量预测。
如图2所示，该图示出了一种获取空域运动矢量和时域运动矢量预测的相邻块的位置结构示意图；其中，图2中左边为获取空域运动矢量的相邻块与编码单元的位置关系，可见获取空域运动矢量的相邻块为位于编码单元左侧和上侧的相邻块，图2中右边为获取时域运动矢量预测的相邻块与编码单元的位置关系，可见获取时域运动矢量预测的相邻块为位于编码单元右侧和下侧的相邻块。
值得说明的是，由于编码是按照一定顺序进行的，位于当前编码单元左侧和上方的大部分子块已经编码完成，这部分子块的运动信息可得；而右侧和下 方的子块还未进行编码，运动信息不可得。所以在实际操作中，导出的是位于当前编码单元左侧和上方的已完成编码的相邻块的空域运动矢量，而右侧和下方则是通过时域运动矢量预测技术从其它已完成编码的图像中获取时域运动矢量预测。
在步骤S120中，对所述空域运动矢量和时域运动矢量预测执行滤波操作，得到滤波后的所述相邻块的空域运动矢量和时域运动矢量预测。
在本说明书一个或多个实施例中，对步骤S110导出的相邻块的空域运动矢量和时域运动矢量预测进行滤波，具体可以包括以下两个步骤：
第一步，对空域运动矢量和时域运动矢量预测的两端执行填充操作，具体填充操作包括以下内容：
MvT[i]＝MvT[0]，MvT[j]＝MvT[M-1]，-2≤i＜0，M≤j＜M+2
MvB[i]＝MvB[0]，MvB[j]＝MvB[M-1]，-2≤i＜0，M≤j＜M+2
MvL[k]＝MvL[0]，MvL[l]＝MvL[N-1]，-2≤k＜0，N≤l＜N+2
MvR[k]＝MvR[0]，MvR[l]＝MvR[N-1]，-2≤k＜0，N≤l＜N+2
其中，MvR，MvB分别表示滤波前右侧和下侧相邻块的时域运动矢量预测；MvL，MvT分别表示滤波前左侧和上侧相邻块的空域运动矢量；i和j表示以子块为单位的列坐标；k和l表示以子块为单位的行坐标；M、N表示编码单元以子块为单位的宽、高。
在实际应用场景中，由于对相邻块的空域运动矢量和时域运动矢量预测执行滤波操作的过程中会使用最外侧两端相邻块之外的运动矢量，而这些块的运动矢量可能没有获取，因此可以使用最边上相邻块的运动矢量作为填充。
第二步，根据以下公式对所述空域运动矢量和时域运动矢量预测进行滤波：
Mvt[x]＝(3*MvT[i 0]+8*MvT[i 1]+10*MvT[i 2]+8*MvT[i 3]+3*MvT[i 4])＞＞5
Mvl[y]＝(3*MvL[j 0]+8*MvL[j 1]+10*MvL[j 2]+8*MvL[j 3]+3*MvL[j 4])＞＞5
Mvb[x]＝(3*MvB[i 0]+8*MvB[i 1]+10*MvB[i 2]+8*MvB[i 3]+3*MvB[i 4])＞＞5
Mvr[y]＝(3*MvR[j 0]+8*MvR[j 1]+10*MvR[j 2]+8*MvR[j 3]+3*MvR[j 4])＞＞5
0≤x＜M
0≤y＜N
其中，MvR，MvB分别表示滤波前右侧和下侧相邻块的时域运动矢量预测；Mvr，Mvb分别表示滤波后右侧和下侧相邻块的时域运动矢量预测；MvL，MvT分别表示滤波前左侧和上侧相邻块的空域运动矢量；Mvl，Mvt分别表示滤波后左侧和上侧相邻块的空域运动矢量；x、y表示当前块在编码单元内的以子块为单位的坐标；x＝0表示当前编码单元内部的第一列子块，y＝0表示当前编码单元内部的第一行子块；i和j表示以子块为单位的列坐标；M、N表示编码单元以子块为单位的宽、高。
在步骤S130中，根据预定的帧间预测模式，利用所述滤波后的相邻块的空域运动矢量和时域运动矢量预测以及当前块在所述编码单元内的坐标位置，确定所述当前块在其四个侧边方向上的参考运动矢量。
本说明书实施例提出了5种新的帧间预测模式，包括：第一帧间预测模式、第二帧间预测模式、第三帧间预测模式、第四帧间预测模式以及第五帧间预测模式。在实际应用中，这5种帧间预测模式又可以分别称为SLTCRB模式、SLCB模式、STCR模式、SLT模式和CRB模式。
进一步地，如图3所示，该图示出了本说明书实施例中分别采用5种帧间预测模式确定当前块在其四个侧边方向上的参考运动矢量的原理示意图；参见 图3，根据上述5种新的帧间预测模式，确定当前块在其四个侧边方向上的参考运动矢量，具体可以包括以下内容：
当采用第一帧间预测模式(SLTCRB模式)时，选取当前块的横坐标方向上的左侧相邻块和右侧相邻块，并选取当前块的纵坐标方向上的上侧相邻块和下侧相邻块，将滤波后的左侧相邻块和上侧相邻块的空域运动矢量以及滤波后的右侧相邻块和下侧相邻块的时域运动矢量预测分别作为当前块在其四个侧边方向上的参考运动矢量；
当采用第二帧间预测模式(SLCB模式)时，选取当前块的横坐标方向上的左侧相邻块以及纵坐标方向上的下侧相邻块，将滤波后的左侧相邻块的空域运动矢量和滤波后的下侧相邻块的时域运动矢量预测分别作为当前块在左侧和下侧方向上的参考运动矢量；并将滤波后的编码单元下侧最右边相邻块的时域运动矢量预测以及滤波后的编码单元左侧最上方相邻块的空域运动矢量，分别作为当前块在右侧和上侧方向上的参考运动矢量；
当采用第三帧间预测模式(STCR模式)时，选取当前块的横坐标方向上的右侧相邻块以及纵坐标方向上的上侧相邻块，将滤波后的右侧相邻块的时域运动矢量预测和滤波后的上侧相邻块的空域运动矢量分别作为当前块在右侧和上侧方向上的参考运动矢量；并将滤波后的编码单元上侧最左边相邻块的空域运动矢量以及滤波后的编码单元右侧最下方相邻块的时域运动矢量预测，分别作为当前块在左侧和下侧方向上的参考运动矢量；
当采用第四帧间预测模式(SLT模式)时，选取当前块的横坐标方向上的左侧相邻块以及纵坐标方向上的上侧相邻块，将滤波后的左侧相邻块的空域运动矢量和上侧相邻块的空域运动矢量分别作为当前块在左侧和上侧方向上的参考运动矢量；并将滤波后的编码单元上侧最右边相邻块的空域运动矢量以及滤波后的编码单元左侧最下方相邻块的空域运动矢量，分别作为当前块在右侧和下侧方向上的参考运动矢量；
当采用第五帧间预测模式(CRB模式)时，选取当前块的横坐标方向上的 右侧相邻块以及纵坐标方向上的下侧相邻块，将滤波后的右侧相邻块的时域运动矢量预测和下侧相邻块的时域运动矢量预测分别作为当前块在右侧和下侧方向上的参考运动矢量；并将滤波后的编码单元下侧最左边相邻块的时域运动矢量预测以及滤波后的编码单元右侧最上方相邻块的时域运动矢量预测，分别作为当前块在左侧和上侧方向上的参考运动矢量。
在本说明书实施例中，这5种新的帧间预测模式均是使用双线性插值的方式结合当前编码单元邻域的空域运动信息和时域运动信息，从而导出当前编码单元内部每个子块的参考运动矢量。因此，在实际应用中，以坐标为(x，y)的当前块为例，对于上述每种模式，可以根据以下公式来选择四个运动矢量分别作为左右上下四个方向上的参考运动矢量：
a)SLTCRB模式
vl＝Mvl[y]
vt＝Mvt[x]
vr＝Mvr[y]
vb＝Mvb[x]
b)SLCB模式
vl＝Mvl[y]
vt＝Mvl[-1]
vr＝Mvb[M]
vb＝Mvb[x]
c)STCR模式
vl＝Mvt[-1]
vt＝Mvt[x]
vr＝Mvr[y]
vb＝Mvr[N]
d)SLT模式
vl＝Mvl[y]
vt＝Mvt[x]
vr＝Mvt[M]
vb＝Mvl[N]
e)CRB模式
vl＝Mvb[-1]
vt＝Mvr[-1]
vr＝Mvr[y]
vb＝Mvb[x]
其中，vl，vr，vt和vb分别表示编码单元内部当前块的左右上下四个方向上的参考运动矢量；Mvr，Mvb分别表示滤波后右侧和下侧相邻块的时域运动矢量预测；Mvl，Mvt分别表示滤波后左侧和上侧相邻块的空域运动矢量；x、y表示当前块在编码单元内的以子块为单位的坐标；x＝0表示当前编码单元内部的第一列子块，y＝0表示当前编码单元内部的第一行子块；M、N表示编码单元以子块为单位的宽、高。
在步骤S140中，根据所述参考运动矢量以及所述当前块在所述编码单元内 的坐标位置，导出所述当前块的运动矢量。
在本说明书一个或多个实施例中，可以通过以下公式，根据步骤S130确定的参考运动矢量结合当前块在编码单元内的坐标位置，导出当前块的运动矢量，具体地：
v h＝((M-x)*vl+x*vr)/M
v v＝((N-y)*vt+y*vb)/N
v[x][y]＝(v h+v v)/2
其中，vl、vr、vt和vb分别表示当前块在左、右、上、下四个方向上的参考运动矢量；x、y表示当前块在编码单元内的以子块为单位的坐标；M、N表示编码单元以子块为单位的宽、高；v h表示当前块的横向运动矢量；v v表示当前块的纵向运动矢量；v[x][y]表示当前块的运动矢量。
进一步地，基于上述5种新的帧间预测模式，使用双线性插值导出当前块的运动矢量后，还可以在编码码流中对上述帧间预测模式进行标记，具体标记过程可以包括以下内容：
将本说明书提出的5种新帧间预测模式组成一个模式候选列表，对于每一个编码单元，在编码端需要传输1比特标记到码流中用于标识是否选中了上述帧间预测方法，若选中了上述帧间预测方法，还需要传输被选中的帧间预测模式在模式候选列表中的索引，将索引的二值化结果传输到码流中，用于标识选中的帧间预测模式。解码端需要解码1比特标记以确定是否使用了上述帧间预测方法，若使用了上述帧间预测方法，解码端还需要解码索引，索引用于确定属于上述5种新帧间预测模式中的哪一种。
进一步地，基于上述步骤S110-步骤S140提供的运动矢量导出方法，本说明书实施例还提供在编码端对每一个编码单元进行率失真优化决策的流程，具 体包括以下内容：
1)根据步骤S110，对于每个帧间预测方向，导出当前编码单元左侧和上方每个相邻块的空域运动矢量以及当前编码单元右侧和下方每个相邻块的时域运动矢量预测。
2)根据步骤S120，对空域运动矢量和时域运动矢量预测进行滤波。
3)根据步骤S130-步骤140，对于本说明书提出的5种新的帧间预测模式，每种模式导出当前编码单元内部每个子块的运动矢量，并且将每个子块根据运动矢量导出预测块。
4)通过RDO决策确定是否使用5种新帧间预测模式中的一种，RDO决策之后编码端对上述帧间预测模式进行标记和索引。
与上述编码端过程相对应的，在解码端对每一个编码单元，则执行如下步骤：
1)根据上述解码帧间预测模式标记和索引的方法，确定是否使用上述5种新帧间预测模式中的一种；若使用上述新帧间预测模式，则继续执行以下步骤，否则结束。
2)根据步骤S110，对于每个帧间预测方向，导出当前编码单元左侧和上方每个相邻块的空域运动矢量以及当前编码单元右侧和下方每个相邻块的时域运动矢量预测。
3)根据步骤S120，对空域运动矢量和时域运动矢量预测进行滤波。
4)根据步骤S130-步骤140，选择对应的新帧间预测模式，导出当前编码单元内部每个子块的运动矢量，每个子块根据运动矢量导出预测块。
5)重建得到重建编码块。
基于同样的思路，本说明书实施例还提供了一种运动矢量导出装置，如图4为本说明书实施例提供的一种运动矢量导出装置，该装置主要包括：
获取模块401，用于获取编码单元在预定方向上的相邻块的空域运动矢量和时域运动矢量预测；
滤波模块402，用于对所述空域运动矢量和时域运动矢量预测执行滤波操作，得到滤波后的所述相邻块的空域运动矢量和时域运动矢量预测；
确定模块403，用于根据预定的帧间预测模式，利用所述滤波后的相邻块的空域运动矢量和时域运动矢量预测以及当前块在所述编码单元内的坐标位置，确定所述当前块在其四个侧边方向上的参考运动矢量；
导出模块404，用于根据所述参考运动矢量以及所述当前块在所述编码单元内的坐标位置，导出所述当前块的运动矢量。
根据本申请的实施例，所述获取模块401还用于：
对位于所述编码单元左侧和上侧的相邻块的运动矢量进行缩放得到所述相邻块的空域运动矢量；具体地，
将所述相邻块的前向运动矢量缩放到前向参考帧列表的第一帧，将所述相邻块的后向运动矢量缩放到后向参考帧列表的第一帧，将缩放后的运动矢量作为所述相邻块的空域运动矢量。
根据本申请的实施例，所述获取模块401还用于：
获取位于所述编码单元右侧和下侧的相邻块的时域运动矢量预测；具体地，
从前向参考帧列表的第一帧或后向参考帧列表的第一帧中获取与所述相邻块具有相同坐标位置的编码块的运动矢量，对所述编码块的运动矢量进行缩放得到所述相邻块的时域运动矢量预测。
根据本申请的实施例，所述滤波模块402具体用于：对所述空域运动矢量和时域运动矢量预测的两端进行填充，并根据以下公式对所述空域运动矢量和时域运动矢量预测进行滤波：
Mvt[x]＝(3*MvT[i 0]+8*MvT[i 1]+10*MvT[i 2]+8*MvT[i 3]+3*MvT[i 4])＞＞5
Mvl[y]＝(3*MvL[j 0]+8*MvL[j 1]+10*MvL[j 2]+8*MvL[j 3]+3*MvL[j 4])＞＞5
Mvb[x]＝(3*MvB[i 0]+8*MvB[i 1]+10*MvB[i 2]+8*MvB[i 3]+3*MvB[i 4])＞＞5
Mvr[y]＝(3*MvR[j 0]+8*MvR[j 1]+10*MvR[j 2]+8*MvR[j 3]+3*MvR[j 4])＞＞5
0≤x＜M
0≤y＜N
其中，MvR，MvB分别表示滤波前右侧和下侧相邻块的时域运动矢量预测；Mvr，Mvb分别表示滤波后右侧和下侧相邻块的时域运动矢量预测；MvL，MvT分别表示滤波前左侧和上侧相邻块的空域运动矢量；Mvl，Mvt分别表示滤波后左侧和上侧相邻块的空域运动矢量；x、y表示当前块在编码单元内的以子块为单位的坐标；i和j表示以子块为单位的列坐标；M、N表示编码单元以子块为单位的宽、高。
根据本申请的实施例，所述帧间预测模式包括第一帧间预测模式、第二帧间预测模式、第三帧间预测模式、第四帧间预测模式以及第五帧间预测模式，所述确定模块403具体用于：
当采用第一帧间预测模式时，选取所述当前块的横坐标方向上的左侧相邻块和右侧相邻块，并选取所述当前块的纵坐标方向上的上侧相邻块和下侧相邻块，将滤波后的所述左侧相邻块和上侧相邻块的空域运动矢量以及滤波后的所述右侧相邻块和下侧相邻块的时域运动矢量预测分别作为所述当前块在其四个侧边方向上的参考运动矢量；
当采用第二帧间预测模式时，选取所述当前块的横坐标方向上的左侧相邻块以及纵坐标方向上的下侧相邻块，将滤波后的所述左侧相邻块的空域运动矢 量和滤波后的所述下侧相邻块的时域运动矢量预测分别作为所述当前块在左侧和下侧方向上的参考运动矢量；并将滤波后的所述编码单元下侧最右边相邻块的时域运动矢量预测以及滤波后的所述编码单元左侧最上方相邻块的空域运动矢量，分别作为所述当前块在右侧和上侧方向上的参考运动矢量；
当采用第三帧间预测模式时，选取所述当前块的横坐标方向上的右侧相邻块以及纵坐标方向上的上侧相邻块，将滤波后的所述右侧相邻块的时域运动矢量预测和滤波后的所述上侧相邻块的空域运动矢量分别作为所述当前块在右侧和上侧方向上的参考运动矢量；并将滤波后的所述编码单元上侧最左边相邻块的空域运动矢量以及滤波后的所述编码单元右侧最下方相邻块的时域运动矢量预测，分别作为所述当前块在左侧和下侧方向上的参考运动矢量；
当采用第四帧间预测模式时，选取所述当前块的横坐标方向上的左侧相邻块以及纵坐标方向上的上侧相邻块，将滤波后的所述左侧相邻块的空域运动矢量和上侧相邻块的空域运动矢量分别作为所述当前块在左侧和上侧方向上的参考运动矢量；并将滤波后的所述编码单元上侧最右边相邻块的空域运动矢量以及滤波后的所述编码单元左侧最下方相邻块的空域运动矢量，分别作为所述当前块在右侧和下侧方向上的参考运动矢量；
当采用第五帧间预测模式时，选取所述当前块的横坐标方向上的右侧相邻块以及纵坐标方向上的下侧相邻块，将滤波后的所述右侧相邻块的时域运动矢量预测和下侧相邻块的时域运动矢量预测分别作为所述当前块在右侧和下侧方向上的参考运动矢量；并将滤波后的所述编码单元下侧最左边相邻块的时域运动矢量预测以及滤波后的所述编码单元右侧最上方相邻块的时域运动矢量预测，分别作为所述当前块在左侧和上侧方向上的参考运动矢量。
根据本申请的实施例，所述导出模块404具体用于：根据以下公式导出所述当前块的运动矢量：
v h＝((M-x)*vl+x*vr)/M
v v＝((N-y)*vt+y*vb)/N
v[x][y]＝(v h+v v)/2
其中，vl、vr、vt和vb分别表示当前块在左、右、上、下四个方向上的参考运动矢量；x、y表示当前块在编码单元内的以子块为单位的坐标；M、N表示编码单元以子块为单位的宽、高；v h表示当前块的横向运动矢量；v v表示当前块的纵向运动矢量；v[x][y]表示当前块的运动矢量。
本说明书实施例还提供一种电子设备，包括存储器，处理器及存储在存储器上并可在处理器上运行的计算机程序，所述处理器执行所述程序时实现上述一种运动矢量导出方法。
上述对本说明书特定实施例进行了描述。其它实施例在所附权利要求书的范围内。在一些情况下，在权利要求书中记载的动作或步骤可以按照不同于实施例中的顺序来执行并且仍然可以实现期望的结果。另外，在附图中描绘的过程不一定要求示出的特定顺序或者连续顺序才能实现期望的结果。在某些实施方式中，多任务处理和并行处理也是可以的或者可能是有利的。
本说明书中的各个实施例均采用递进的方式描述，各个实施例之间相同相似的部分互相参见即可，每个实施例重点说明的都是与其他实施例的不同之处。尤其，对于装置、电子设备实施例而言，由于其基本相似于方法实施例，所以描述的比较简单，相关之处参见方法实施例的部分说明即可。
本说明书实施例提供的装置、电子设备与方法是对应的，因此，装置、电子设备也具有与对应方法类似的有益技术效果，由于上面已经对方法的有益技术效果进行了详细说明，因此，这里不再赘述对应装置、电子设备的有益技术效果。
本说明书是参照根据本说明书实施例的方法、设备(系统)、和计算机程序产品的流程图和/或方框图来描述的。应理解可由计算机程序指令实现流程图 和/或方框图中的每一流程和/或方框、以及流程图和/或方框图中的流程和/或方框的结合。可提供这些计算机程序指令到通用计算机、专用计算机、嵌入式处理机或其他可编程数据处理设备的处理器以产生一个机器，使得通过计算机或其他可编程数据处理设备的处理器执行的指令产生用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的装置。
还需要说明的是，术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含，从而使得包括一系列要素的过程、方法、商品或者设备不仅包括那些要素，而且还包括没有明确列出的其他要素，或者是还包括为这种过程、方法、商品或者设备所固有的要素。在没有更多限制的情况下，由语句“包括一个……”限定的要素，并不排除在包括所述要素的过程、方法、商品或者设备中还存在另外的相同要素。
本说明书可以在由计算机执行的计算机可执行指令的一般上下文中描述，例如程序模块。一般地，程序模块包括执行特定任务或实现特定抽象数据类型的例程、程序、对象、组件、数据结构等等。也可以在分布式计算环境中实践说明书，在这些分布式计算环境中，由通过通信网络而被连接的远程处理设备来执行任务。在分布式计算环境中，程序模块可以位于包括存储设备在内的本地和远程计算机存储介质中。
对所公开的实施例的上述说明，使本领域技术人员能够实现或使用本发明。对这些实施例的多种修改对本领域技术人员来说将是显而易见的，本文中所定义的一般原理可以在不脱离本发明的精神和范围的情况下，在其他实施例中实现。因此，本发明将不会被限制于本文所示的这些实施例，而是要符合与本文所公开的原理和新颖特点相一致的最宽的范围。

Claims (14)

  1. 一种运动矢量导出方法，所述方法包括：
    获取编码单元在预定方向上的相邻块的空域运动矢量和时域运动矢量预测；
    对所述空域运动矢量和时域运动矢量预测执行滤波操作，得到滤波后的所述相邻块的空域运动矢量和时域运动矢量预测；
    根据预定的帧间预测模式，利用所述滤波后的相邻块的空域运动矢量和时域运动矢量预测以及当前块在所述编码单元内的坐标位置，确定所述当前块在其四个侧边方向上的参考运动矢量；
    根据所述参考运动矢量以及所述当前块在所述编码单元内的坐标位置，导出所述当前块的运动矢量。
  2. 如权利要求1所述的方法，所述获取编码单元在预定方向上的相邻块的空域运动矢量，包括：
    对位于所述编码单元左侧和上侧的相邻块的运动矢量进行缩放得到所述相邻块的空域运动矢量；具体地，
    将所述相邻块的前向运动矢量缩放到前向参考帧列表的第一帧，将所述相邻块的后向运动矢量缩放到后向参考帧列表的第一帧，将缩放后的运动矢量作为所述相邻块的空域运动矢量。
  3. 如权利要求1所述的方法，所述获取编码单元在预定方向上的相邻块的时域运动矢量预测，包括：
    获取位于所述编码单元右侧和下侧的相邻块的时域运动矢量预测；具体地，
    从前向参考帧列表的第一帧或后向参考帧列表的第一帧中获取与所述相邻块具有相同坐标位置的编码块的运动矢量，对所述编码块的运动矢量进行缩放得到所述相邻块的时域运动矢量预测。
  4. 如权利要求1所述的方法，所述对所述空域运动矢量和时域运动矢量预测执行滤波操作，具体包括，对所述空域运动矢量和时域运动矢量预测的两端进行填充，并根据以下公式对所述空域运动矢量和时域运动矢量预测进行滤波：
    Mvt[x]＝(3*MvT[i 0]+8*MvT[i 1]+10*MvT[i 2]+8*MvT[i 3]+3*MvT[i 4])＞＞5
    Mvl[y]＝(3*MvL[j 0]+8*MvL[j 1]+10*MvL[j 2]+8*MvL[j 3]+3*MvL[j 4])＞＞5
    Mvb[x]＝(3*MvB[i 0]+8*MvB[i 1]+10*MvB[i 2]+8*MvB[i 3]+3*MvB[i 4])＞＞5
    Mvr[y]＝(3*MvR[j 0]+8*MvR[j 1]+10*MvR[j 2]+8*MvR[j 3]+3*MvR[j 4])＞＞5
    0≤x＜M
    0≤y＜N
    其中，MvR，MvB分别表示滤波前右侧和下侧相邻块的时域运动矢量预测；Mvr，Mvb分别表示滤波后右侧和下侧相邻块的时域运动矢量预测；MvL，MvT分别表示滤波前左侧和上侧相邻块的空域运动矢量；Mvl，Mvt分别表示滤波后左侧和上侧相邻块的空域运动矢量；x、y表示当前块在编码单元内的以子块为单位的坐标；i和j表示以子块为单位的列坐标；M、N表示编码单元以子块为单位的宽、高。
  5. 如权利要求1所述的方法，所述帧间预测模式包括：
    第一帧间预测模式、第二帧间预测模式、第三帧间预测模式、第四帧间预测模式以及第五帧间预测模式。
  6. 如权利要求5所述的方法，所述根据预定的帧间预测模式，利用所述滤波后的相邻块的空域运动矢量和时域运动矢量预测以及当前块在所述编码单元内的坐标位置，确定所述当前块在其四个侧边方向上的参考运动矢量，具体包括：
    当采用第一帧间预测模式时，选取所述当前块的横坐标方向上的左侧相邻块和右侧相邻块，并选取所述当前块的纵坐标方向上的上侧相邻块和下侧相邻块，将滤波后的所述左侧相邻块和上侧相邻块的空域运动矢量以及滤波后的所述右侧相邻块和下侧相邻块的时域运动矢量预测分别作为所述当前块在其四个侧边方向上的参考运动矢量；
    当采用第二帧间预测模式时，选取所述当前块的横坐标方向上的左侧相邻块以及纵坐标方向上的下侧相邻块，将滤波后的所述左侧相邻块的空域运动矢量和滤波后的所述下侧相邻块的时域运动矢量预测分别作为所述当前块在左侧和下侧方向上的参考运动矢量；并将滤波后的所述编码单元下侧最右边相邻块的时域运动矢量预测以及滤波后的所述编码单元左侧最上方相邻块的空域运动矢量，分别作为所述当前块在右侧和上侧方向上的参考运动矢量；
    当采用第三帧间预测模式时，选取所述当前块的横坐标方向上的右侧相邻块以及纵坐标方向上的上侧相邻块，将滤波后的所述右侧相邻块的时域运动矢量预测和滤波后的所述上侧相邻块的空域运动矢量分别作为所述当前块在右侧和上侧方向上的参考运动矢量；并将滤波后的所述编码单元上侧最左边相邻块的空域运动矢量以及滤波后的所述编码单元右侧最下方相邻块的时域运动矢量预测，分别作为所述当前块在左侧和下侧方向上的参考运动矢量；
    当采用第四帧间预测模式时，选取所述当前块的横坐标方向上的左侧相邻块以及纵坐标方向上的上侧相邻块，将滤波后的所述左侧相邻块的空域运动矢量和上侧相邻块的空域运动矢量分别作为所述当前块在左侧和上侧方向上的参考运动矢量；并将滤波后的所述编码单元上侧最右边相邻块的空域运动矢量以及滤波后的所述编码单元左侧最下方相邻块的空域运动矢量，分别作为所述当前块在右侧和下侧方向上的参考运动矢量；
    当采用第五帧间预测模式时，选取所述当前块的横坐标方向上的右侧相邻块以及纵坐标方向上的下侧相邻块，将滤波后的所述右侧相邻块的时域运动矢量预测和下侧相邻块的时域运动矢量预测分别作为所述当前块在右侧和下侧方 向上的参考运动矢量；并将滤波后的所述编码单元下侧最左边相邻块的时域运动矢量预测以及滤波后的所述编码单元右侧最上方相邻块的时域运动矢量预测，分别作为所述当前块在左侧和上侧方向上的参考运动矢量。
  7. 如权利要求1所述的方法，所述根据所述参考运动矢量以及所述当前块在所述编码单元内的坐标位置，导出所述当前块的运动矢量，具体包括，根据以下公式导出所述当前块的运动矢量：
    v h＝((M-x)*vl+x*vr)/M
    v v＝((N-y)*vt+y*vb)/N
    v[x][y]＝(v h+v v)/2
    其中，vl、vr、vt和vb分别表示当前块在左、右、上、下四个方向上的参考运动矢量；x、y表示当前块在编码单元内的以子块为单位的坐标；M、N表示编码单元以子块为单位的宽、高；v h表示当前块的横向运动矢量；v v表示当前块的纵向运动矢量；v[x][y]表示当前块的运动矢量。
  8. 一种运动矢量导出装置，所述装置包括：
    获取模块，用于获取编码单元在预定方向上的相邻块的空域运动矢量和时域运动矢量预测；
    滤波模块，用于对所述空域运动矢量和时域运动矢量预测执行滤波操作，得到滤波后的所述相邻块的空域运动矢量和时域运动矢量预测；
    确定模块，用于根据预定的帧间预测模式，利用所述滤波后的相邻块的空域运动矢量和时域运动矢量预测以及当前块在所述编码单元内的坐标位置，确定所述当前块在其四个侧边方向上的参考运动矢量；
    导出模块，用于根据所述参考运动矢量以及所述当前块在所述编码单元内的坐标位置，导出所述当前块的运动矢量。
  9. 如权利要求8所述的装置，所述获取模块还用于：
    对位于所述编码单元左侧和上侧的相邻块的运动矢量进行缩放得到所述相邻块的空域运动矢量；具体地，
    将所述相邻块的前向运动矢量缩放到前向参考帧列表的第一帧，将所述相邻块的后向运动矢量缩放到后向参考帧列表的第一帧，将缩放后的运动矢量作为所述相邻块的空域运动矢量。
  10. 如权利要求8所述的装置，所述获取模块还用于：
    获取位于所述编码单元右侧和下侧的相邻块的时域运动矢量预测；具体地，
    从前向参考帧列表的第一帧或后向参考帧列表的第一帧中获取与所述相邻块具有相同坐标位置的编码块的运动矢量，对所述编码块的运动矢量进行缩放得到所述相邻块的时域运动矢量预测。
  11. 如权利要求8所述的装置，所述滤波模块具体用于：对所述空域运动矢量和时域运动矢量预测的两端进行填充，并根据以下公式对所述空域运动矢量和时域运动矢量预测进行滤波：
    Mvt[x]＝(3*MvT[i 0]+8*MvT[i 1]+10*MvT[i 2]+8*MvT[i 3]+3*MvT[i 4])＞＞5
    Mvl[y]＝(3*MvL[j 0]+8*MvL[j 1]+10*MvL[j 2]+8*MvL[j 3]+3*MvL[j 4])＞＞5
    Mvb[x]＝(3*MvB[i 0]+8*MvB[i 1]+10*MvB[i 2]+8*MvB[i 3]+3*MvB[i 4])＞＞5
    Mvr[y]＝(3*MvR[j 0]+8*MvR[j 1]+10*MvR[j 2]+8*MvR[j 3]+3*MvR[j 4])＞＞5
    0≤x＜M
    0≤y＜N
    其中，MvR，MvB分别表示滤波前右侧和下侧相邻块的时域运动矢量预测；Mvr，Mvb分别表示滤波后右侧和下侧相邻块的时域运动矢量预测；MvL，MvT分 别表示滤波前左侧和上侧相邻块的空域运动矢量；Mvl，Mvt分别表示滤波后左侧和上侧相邻块的空域运动矢量；x、y表示当前块在编码单元内的以子块为单位的坐标；i和j表示以子块为单位的列坐标；M、N表示编码单元以子块为单位的宽、高。
  12. 如权利要求8所述的装置，所述帧间预测模式包括第一帧间预测模式、第二帧间预测模式、第三帧间预测模式、第四帧间预测模式以及第五帧间预测模式，所述确定模块具体用于：
    当采用第一帧间预测模式时，选取所述当前块的横坐标方向上的左侧相邻块和右侧相邻块，并选取所述当前块的纵坐标方向上的上侧相邻块和下侧相邻块，将滤波后的所述左侧相邻块和上侧相邻块的空域运动矢量以及滤波后的所述右侧相邻块和下侧相邻块的时域运动矢量预测分别作为所述当前块在其四个侧边方向上的参考运动矢量；
    当采用第二帧间预测模式时，选取所述当前块的横坐标方向上的左侧相邻块以及纵坐标方向上的下侧相邻块，将滤波后的所述左侧相邻块的空域运动矢量和滤波后的所述下侧相邻块的时域运动矢量预测分别作为所述当前块在左侧和下侧方向上的参考运动矢量；并将滤波后的所述编码单元下侧最右边相邻块的时域运动矢量预测以及滤波后的所述编码单元左侧最上方相邻块的空域运动矢量，分别作为所述当前块在右侧和上侧方向上的参考运动矢量；
    当采用第三帧间预测模式时，选取所述当前块的横坐标方向上的右侧相邻块以及纵坐标方向上的上侧相邻块，将滤波后的所述右侧相邻块的时域运动矢量预测和滤波后的所述上侧相邻块的空域运动矢量分别作为所述当前块在右侧和上侧方向上的参考运动矢量；并将滤波后的所述编码单元上侧最左边相邻块的空域运动矢量以及滤波后的所述编码单元右侧最下方相邻块的时域运动矢量预测，分别作为所述当前块在左侧和下侧方向上的参考运动矢量；
    当采用第四帧间预测模式时，选取所述当前块的横坐标方向上的左侧相邻 块以及纵坐标方向上的上侧相邻块，将滤波后的所述左侧相邻块的空域运动矢量和上侧相邻块的空域运动矢量分别作为所述当前块在左侧和上侧方向上的参考运动矢量；并将滤波后的所述编码单元上侧最右边相邻块的空域运动矢量以及滤波后的所述编码单元左侧最下方相邻块的空域运动矢量，分别作为所述当前块在右侧和下侧方向上的参考运动矢量；
    当采用第五帧间预测模式时，选取所述当前块的横坐标方向上的右侧相邻块以及纵坐标方向上的下侧相邻块，将滤波后的所述右侧相邻块的时域运动矢量预测和下侧相邻块的时域运动矢量预测分别作为所述当前块在右侧和下侧方向上的参考运动矢量；并将滤波后的所述编码单元下侧最左边相邻块的时域运动矢量预测以及滤波后的所述编码单元右侧最上方相邻块的时域运动矢量预测，分别作为所述当前块在左侧和上侧方向上的参考运动矢量。
  13. 如权利要求8所述的装置，所述导出模块具体用于：根据以下公式导出所述当前块的运动矢量：
    v h＝((M-x)*vl+x*vr)/M
    v v＝((N-y)*vt+y*vb)/N
    v[x][y]＝(v h+v v)/2
    其中，vl、vr、vt和vb分别表示当前块在左、右、上、下四个方向上的参考运动矢量；x、y表示当前块在编码单元内的以子块为单位的坐标；M、N表示编码单元以子块为单位的宽、高；v h表示当前块的横向运动矢量；v v表示当前块的纵向运动矢量；v[x][y]表示当前块的运动矢量。
  14. 一种电子设备，包括存储器，处理器及存储在存储器上并可在处理器上运行的计算机程序，其特征在于，所述处理器执行所述程序时实现权利要求1至7中任一项所述的方法。
PCT/CN2019/102748 2019-08-19 2019-08-27 一种运动矢量导出方法、装置及电子设备 WO2021031225A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US17/645,698 US11997284B2 (en) 2019-08-19 2021-12-22 Method for deriving motion vector, and electronic device of current block in coding unit

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910766509.XA CN110475116B (zh) 2019-08-19 2019-08-19 一种运动矢量导出方法、装置及电子设备
CN201910766509.X 2019-08-19

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US17/645,698 Continuation-In-Part US11997284B2 (en) 2019-08-19 2021-12-22 Method for deriving motion vector, and electronic device of current block in coding unit

Publications (1)

Publication Number Publication Date
WO2021031225A1 true WO2021031225A1 (zh) 2021-02-25

Family

ID=68511213

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/102748 WO2021031225A1 (zh) 2019-08-19 2019-08-27 一种运动矢量导出方法、装置及电子设备

Country Status (3)

Country Link
US (1) US11997284B2 (zh)
CN (1) CN110475116B (zh)
WO (1) WO2021031225A1 (zh)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111654708B (zh) * 2020-06-07 2022-08-23 咪咕文化科技有限公司 一种运动矢量获取方法、装置及电子设备
CN111901590B (zh) * 2020-06-29 2023-04-18 北京大学 一种用于帧间预测的细化运动矢量存储方法及装置

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105959699A (zh) * 2016-05-06 2016-09-21 西安电子科技大学 一种基于运动估计和时空域相关性的快速帧间预测方法
CN108347616A (zh) * 2018-03-09 2018-07-31 中南大学 一种基于可选时域运动矢量预测的深度预测方法及装置
CN109005412A (zh) * 2017-06-06 2018-12-14 北京三星通信技术研究有限公司 运动矢量获取的方法及设备

Family Cites Families (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9137544B2 (en) * 2010-11-29 2015-09-15 Mediatek Inc. Method and apparatus for derivation of mv/mvp candidate for inter/skip/merge modes
US8711940B2 (en) * 2010-11-29 2014-04-29 Mediatek Inc. Method and apparatus of motion vector prediction with extended motion vector predictor
US8755437B2 (en) * 2011-03-17 2014-06-17 Mediatek Inc. Method and apparatus for derivation of spatial motion vector candidate and motion vector prediction candidate
WO2012122927A1 (en) * 2011-03-14 2012-09-20 Mediatek Inc. Method and apparatus for derivation of motion vector candidate and motion vector prediction candidate
CN102883161B (zh) * 2012-09-19 2015-09-30 华为技术有限公司 视频编码和解码的处理方法和装置
CN104427345B (zh) * 2013-09-11 2019-01-08 华为技术有限公司 运动矢量的获取方法、获取装置、视频编解码器及其方法
WO2016008161A1 (en) * 2014-07-18 2016-01-21 Mediatek Singapore Pte. Ltd. Temporal derived bi-directional motion vector predictor
KR101676791B1 (ko) * 2015-04-14 2016-11-16 삼성전자주식회사 영상 복호화 방법
US10136155B2 (en) * 2016-07-27 2018-11-20 Cisco Technology, Inc. Motion compensation using a patchwork motion field
CN109089119B (zh) * 2017-06-13 2021-08-13 浙江大学 一种运动矢量预测的方法及设备

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105959699A (zh) * 2016-05-06 2016-09-21 西安电子科技大学 一种基于运动估计和时空域相关性的快速帧间预测方法
CN109005412A (zh) * 2017-06-06 2018-12-14 北京三星通信技术研究有限公司 运动矢量获取的方法及设备
CN108347616A (zh) * 2018-03-09 2018-07-31 中南大学 一种基于可选时域运动矢量预测的深度预测方法及装置

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
THIRUMALAI, VIJAYARAGHAVAN ET AL.: "INTER-VIEW MOTION VECTOR PREDICTION FOR DEPTH CODING", HTTPS://WWW.RESEARCHGATE.NET/PUBLICATION/262972574, 31 July 2014 (2014-07-31), XP055783381 *

Also Published As

Publication number Publication date
US11997284B2 (en) 2024-05-28
CN110475116B (zh) 2021-09-21
CN110475116A (zh) 2019-11-19
US20220191503A1 (en) 2022-06-16

Similar Documents

Publication Publication Date Title
KR102160564B1 (ko) 픽쳐 예측 방법 및 관련 장치
JP2021022936A (ja) 画像予測方法および関連装置
RU2559740C2 (ru) Способ и устройство для кодирования/декодирования вектора движения
JP5234587B2 (ja) 映像符号化方法及び復号方法、それらの装置、及びそれらのプログラム並びにプログラムを記録した記憶媒体
TWI711297B (zh) 利用量化係數分量與圖框間預測資訊解碼視訊資料的方法
CN108781284A (zh) 具有仿射运动补偿的视频编解码的方法及装置
CN109068142B (zh) 360度视频帧内预测快速决策方法、装置、编解码器和介质
EP2214415A1 (en) A dual prediction video encoding and decoding method and a device
TW202143731A (zh) 在高級運動向量預測模式中的圖像解碼方法
WO2015010319A1 (zh) 一种基于p帧的多假设运动补偿编码方法
CN101888546B (zh) 一种运动估计的方法及装置
US11997284B2 (en) Method for deriving motion vector, and electronic device of current block in coding unit
WO2022227622A1 (zh) 一种权值可配置的帧间帧内联合预测编解码的方法及装置
CN109889838B (zh) 一种基于roi区域的hevc快速编码方法
JP2005348008A (ja) 動画像符号化方法、動画像符号化装置、動画像符号化プログラム及びそのプログラムを記録したコンピュータ読み取り可能な記録媒体
CN116980596A (zh) 一种帧内预测方法、编码器、解码器及存储介质
WO2023044916A1 (zh) 帧内预测的方法、编码器、解码器和编解码系统
WO2023044918A1 (zh) 帧内预测的方法、编码器、解码器和编解码系统
WO2023044917A1 (zh) 帧内预测的方法、编码器、解码器和编解码系统
KR101286071B1 (ko) 부호화기 및 그 인트라 예측 방법
CN113055670B (zh) 一种基于hevc/h.265的视频编码的方法及系统
CN105306953A (zh) 图像编码方法和装置
WO2020181507A1 (zh) 图像处理的方法与装置
WO2021056920A1 (zh) 视频编解码的方法和装置
WO2020258053A1 (zh) 图像分量预测方法、装置及计算机存储介质

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: 19942389

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19942389

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 19942389

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 16/02/2023)