[go: up one dir, main page]

CN107426573B - Self-adaptive rapid prediction unit partitioning method and device based on motion homogeneity - Google Patents

Self-adaptive rapid prediction unit partitioning method and device based on motion homogeneity Download PDF

Info

Publication number
CN107426573B
CN107426573B CN201710670690.5A CN201710670690A CN107426573B CN 107426573 B CN107426573 B CN 107426573B CN 201710670690 A CN201710670690 A CN 201710670690A CN 107426573 B CN107426573 B CN 107426573B
Authority
CN
China
Prior art keywords
motion
prediction unit
minimum
homogeneity
prediction
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201710670690.5A
Other languages
Chinese (zh)
Other versions
CN107426573A (en
Inventor
马丽
宋建斌
刘慧�
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.)
Ordos Institute of Technology
Original Assignee
Ordos Institute of Technology
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 Ordos Institute of Technology filed Critical Ordos Institute of Technology
Priority to CN201710670690.5A priority Critical patent/CN107426573B/en
Publication of CN107426573A publication Critical patent/CN107426573A/en
Application granted granted Critical
Publication of CN107426573B publication Critical patent/CN107426573B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

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
    • H04N19/169—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
    • H04N19/102—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
    • 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
    • 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
    • H04N19/567—Motion estimation based on rate distortion criteria

Landscapes

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

Abstract

The invention discloses a self-adaptive rapid prediction unit partitioning method and device based on motion homogeneity. According to the adaptive fast prediction unit division method, firstly, a current maximum coding unit is divided into a plurality of minimum coding units; then, motion estimation is carried out on each minimum coding unit to obtain a motion vector; secondly, respectively calculating motion homogeneity values of different types of prediction units based on the obtained motion vectors; and finally, selecting the optimal prediction unit for coding according to the calculated motion homogeneity values of the prediction units of different types. Compared with the prior art, the method shortens the selection range of the prediction unit in the current maximum coding unit in advance, reduces the search amount of the prediction unit, can effectively reduce the calculation complexity of video coding, improves the video coding efficiency, and has the characteristics of high coding quality, strong adaptability and the like.

Description

Self-adaptive rapid prediction unit partitioning method and device based on motion homogeneity
Technical Field
The invention relates to a prediction unit partitioning method, in particular to a self-adaptive fast prediction unit partitioning method based on motion homogeneity, and also relates to a corresponding self-adaptive fast prediction unit partitioning device, belonging to the technical field of video coding.
Background
With the rapid development of information technology, the requirements of video applications such as digital televisions, high definition cinemas, wireless multimedia communication and the like on the resolution, definition, color and the like of videos in the video applications are higher and higher. Accordingly, the amount of video data required has increased dramatically and has far exceeded the growth rate of channel bandwidth and storage capacity, presenting significant difficulties in video signal processing, storage, transmission, display, and the like. Therefore, the video data is efficiently compressed, the real-time requirement of a communication system is met, the high-fidelity decoding quality is provided, and the problem that the research field and the application field cannot be avoided is solved.
The h.265/MPEG-HEVC standard is a high performance video coding standard for high resolution video evolution, with the goal of increasing compression rate and reducing network bandwidth. In the h.265/MPEG-HEVC standard, although advanced technologies such as flexible coding unit division and more various inter-frame motion estimation modes can greatly improve the coding quality of video, meet the requirements of various types of video, and meet the requirements of the current high-resolution video development. The high complexity that follows is a bottleneck for its widespread use. Under the circumstances, a fast algorithm for improving the coding speed under the h.265/MPEG-HEVC standard is being actively researched, aiming at paving the way for the wide application of a new coding mode.
In the h.265/MPEG-HEVC standard, as shown in fig. 1, for one coding unit, Inter prediction modes are further divided into Part _2Nx2N, Part _2xN, Part _2NxN, Part _ Nx2N, Part _2NxnU, Part _2NxnD, Part _ nRx2N, and Part _ nLx2N 8. Therefore, it is necessary to sequentially encode each prediction mode, calculate the rate-distortion cost of each prediction mode, and select the optimal encoding mode by comparing the rate-distortion cost values of each prediction mode. In the process, the prediction modes under all the coding unit partitions are exhaustively traversed to select the optimal coding mode, so that the calculation amount is very large, the calculation complexity of video coding is increased, and the efficiency of video coding is reduced.
Disclosure of Invention
The invention provides a self-adaptive rapid prediction unit partitioning method based on motion homogeneity.
Another objective of the present invention is to provide an adaptive fast prediction unit partitioning apparatus based on motion homogeneity.
In order to achieve the purpose, the invention adopts the following technical scheme:
according to a first aspect of the embodiments of the present invention, there is provided an adaptive fast prediction unit partitioning method based on motion homogeneity, including the following steps:
step S1: dividing the current maximum coding unit into a plurality of minimum coding units;
step S2: performing motion estimation on each minimum coding unit to obtain a motion vector;
step S3: respectively calculating motion homogeneity values of different kinds of prediction units based on the motion vectors;
step S4: and selecting an optimal prediction unit for coding according to the motion homogeneity value.
Preferably, in step S2, when motion estimation is performed on each minimum coding unit, multiple reference frames located before the current frame are selected to perform integer pixel motion and fractional pixel motion estimation, respectively, and the motion vector of each minimum coding unit is recorded.
Preferably, in step S4, the calculated motion homogeneity values of different prediction units are compared, rate distortion costs of two types of prediction units with the smallest selected motion homogeneity value and the second smallest selected motion homogeneity value are respectively calculated, and the prediction unit with the smallest rate distortion cost value is selected as the optimal partition of the current largest coding unit.
According to a second aspect of the embodiments of the present invention, there is provided an adaptive fast prediction unit partitioning apparatus based on motion homogeneity, including a partitioning module, a motion estimation module, a storage module, a calculation module, and a determination module;
the dividing module receives a current maximum coding unit and divides the maximum coding unit into a plurality of minimum coding units;
the motion estimation module respectively carries out motion estimation on the plurality of minimum coding units divided by the division module;
the storage module stores a plurality of motion vectors generated by the motion estimation module;
the calculation module is used for calculating motion homogeneity values of different types of prediction units respectively based on the motion vectors stored by the storage module;
and the judging module selects the optimal prediction unit for coding according to the motion homogeneity values of different prediction units calculated by the calculating module.
Preferably, the motion homogeneity value of a certain prediction unit is expressed as:
H＝∑i≤n(Hi)
wherein H represents the motion homogeneity value of a certain prediction unit, n represents that the prediction unit comprises n parts, HiAnd (3) representing the motion homogeneity value of the ith part, wherein n and i are positive integers.
Preferably, the motion homogeneity value of the i-th part is expressed as:
Hi＝Sx+Sy)
wherein S isxRepresenting the variance of the horizontal vector, SyRepresenting the variance of the vertical vector.
Wherein preferably, the variance of the horizontal vector and the variance of the vertical vector are respectively expressed as:
Figure BDA0001373021510000031
Figure BDA0001373021510000032
and the number of the first and second electrodes,
Figure BDA0001373021510000033
Figure BDA0001373021510000034
where m indicates that the i-th part of a certain PU contains m minimum coding units, MVXmHorizontal vector representing the M-th smallest coding unit, MxAn average value of horizontal vectors representing m minimum coding units included in the ith part of the prediction unit; MVY (multifunction vehicle)mRepresents the vertical direction of the m-th minimum coding unitAmount, MyAnd represents the average value of the vertical vectors of m minimum coding units contained in the ith part of the prediction unit, wherein m is a positive integer.
Preferably, the motion homogeneity value of the i-th part is expressed as:
Hi＝(Dx+Dy)
wherein D isxDenotes the standard deviation, D, of the horizontal vectoryThe standard deviation of the vertical vector is indicated.
The self-adaptive rapid prediction unit division method provided by the invention makes full use of the motion homogeneity of the video image and carries out rapid decision on the division of the optimal prediction unit based on the result of motion estimation. Compared with the prior art, the method shortens the selection range of the prediction unit in the current maximum coding unit in advance, reduces the search amount of the prediction unit, can effectively reduce the calculation complexity of video coding, improves the video coding efficiency, and has the characteristics of high coding quality, strong adaptability and the like.
Drawings
FIG. 1 is a diagram illustrating the partitioning of a prediction unit;
FIG. 2 is a flowchart illustrating a method for partitioning an adaptive fast prediction unit according to the present invention;
FIG. 3 is a diagram illustrating the division of a maximum coding unit in the prior art;
FIG. 4 is a schematic diagram illustrating a sequence number of a minimum coding unit in the adaptive fast prediction unit partition method according to the present invention;
FIG. 5 is a diagram illustrating forward multi-reference frame prediction of a current reference frame in the adaptive fast prediction unit partition method according to the present invention;
FIG. 6 is a schematic structural diagram of an adaptive fast prediction unit partitioning apparatus according to the present invention.
Detailed Description
The technical contents of the present invention will be further described in detail with reference to the accompanying drawings and specific embodiments.
The self-adaptive rapid prediction unit division method based on the motion homogeneity fully utilizes the motion homogeneity of the video image and carries out rapid decision on the division self-adaptation of the optimal prediction unit based on the result of motion estimation. Where block-based motion estimation assumes that all pixels within a block have the same motion. This assumption can be satisfied for most cases, and is reasonable under the condition of high video frame rate. Conversely, if the motion vectors of all the pixels in the frame can be obtained, and the sizes and directions of the motion vectors of the pixels in a certain region are the same or similar, all the pixels in the region can be combined into a block, and the motion vectors of the pixels are represented by the motion vectors of the block, so that the purpose of compression coding is achieved. Pixels are said to have "motion homogeneity" if the motion is similar between pixels. Using the motion homogeneity of the pixels, it is possible to intuitively explain how to perform mode selection, and since the h.265/MPEG-HEVC standard employs block-based predictive coding, it is not possible to obtain motion vectors of pixels, but motion vectors of small blocks. And selecting a large block mode by using the small block motion homogeneity, wherein the principle is similar to the pixel point motion homogeneity mode selection principle.
As shown in fig. 2, the adaptive fast prediction unit partitioning method provided by the present invention includes the following steps:
step S1: dividing the current maximum coding unit into a plurality of minimum coding units;
when performing the Inter coding mode, the user may divide the maximum coding unit into a plurality of minimum coding units according to the computational complexity.
As shown in FIG. 3, in one embodiment of the present invention, the coding units may have four size levels of 64x64, 32x32, 16x16, and 8x 8. Generally, 64x64 is used as the highest size level (also referred to as the largest coding unit) and 8x8 is used as the lowest size level (also referred to as the smallest coding unit). Thus, as shown in fig. 4, a maximum coding unit of size 64x64 may be divided into 64 minimum coding units of size 8x 8.
Step S2: performing motion estimation on each minimum coding unit to obtain a motion vector;
motion estimation is performed on each of the minimum coding units divided by the current maximum coding unit in step S1, that is, motion estimation is performed on each of the 64 minimum coding units having a size of 8 × 8, so as to obtain and store motion vectors of the minimum coding units. Since motion may be directed to sub-pixel locations, the motion estimation process for each minimum coding unit may be performed in two steps: step one is the motion estimation of integer pixel, it searches for and gets the most similar integer pixel matching block with the current frame in a area; and step two, sub-pixel motion estimation, namely firstly utilizing pixel interpolation of an integral pixel position to obtain a pixel of a sub-pixel position, and then searching in a region to obtain a sub-pixel matching block most similar to the current frame.
As shown in fig. 5, in an embodiment of the present invention, when performing motion estimation on each minimum coding unit, a plurality of reference frames located before a current frame (a current frame with time t) may be selected to perform integer motion estimation and fractional motion estimation, respectively, and a motion vector of each minimum coding unit is recorded. In consideration of the fact that the calculation complexity of motion estimation is increased when a plurality of reference frames are used for motion estimation, the calculation complexity and the precision can be compromised when two reference frames (a reference frame with time t-1 and a reference frame with time t-2) in front of a current frame (with time t) are selected to respectively carry out integer pixel motion and quarter-pixel motion estimation through repeated experiments.
Step S3: respectively calculating motion homogeneity values of different types of prediction units based on the obtained motion vectors;
the present invention represents motion homogeneity by the variance or standard deviation of motion vectors under different kinds of prediction units. For example, the coding unit with the size of 2Nx2N includes Part _2Nx2N, Part _ NxN, Part _2NxN, Part _ Nx2N, Part _2NxnU, Part _2NxnD, Part _ nRx2N, and Part _ nLx2N 8. From the motion vectors of all the minimum coding units obtained in step S2, the variances or standard deviations of the 8 kinds of prediction units are calculated, respectively. Since different prediction units contain different numbers of motion vectors when being coded, the variance or standard deviation of each prediction unit is the variance or standard deviation of all the motion vectors of the minimum coding unit contained in the prediction unit. For example, as shown in fig. 1, in one embodiment of the present invention, since the prediction unit of Part _2NxnD includes a 2NxN/2 prediction block and a 2Nx3N/2 prediction block, and the prediction unit of Part _ NxN includes 4 NxN prediction blocks, the variance of the prediction unit of Part _2NxnD is equal to the sum of the variances of the 2NxN/2 prediction block and the 2Nx3N/2 prediction block; the variance of the prediction unit of Part _ NxN is equal to the sum of the variances of the 4 NxN prediction blocks. Since the motion vector includes a horizontal vector and a vertical vector, and the horizontal vector and the vertical vector are independent of each other. Therefore, the motion homogeneity of the prediction unit may be expressed by the sum of the variance of the horizontal vector and the variance of the vertical vector, and may also be expressed by the sum of the standard deviation of the horizontal component and the standard deviation of the vertical vector.
In summary, the motion homogeneity value of a certain prediction unit can be expressed as:
H＝∑i≤n(Hi) (1)
wherein H represents the motion homogeneity value of a certain prediction unit, n represents that the prediction unit comprises n parts, HiAnd (3) representing the motion homogeneity value of the ith part, wherein n and i are positive integers.
And the motion homogeneity value of the i-th part can be expressed as:
Hi＝Sx+Sy) (2)
wherein S isxRepresenting the variance of the horizontal vector, SyRepresenting the variance of the vertical vector.
The motion homogeneity value of the ith part can also be expressed as:
Hi＝Dx+Dy) (3)
wherein D isxDenotes the standard deviation, D, of the horizontal vectoryThe standard deviation of the vertical vector is indicated.
Then, the variance of the horizontal vector and the variance of the vertical vector can be expressed as:
Figure BDA0001373021510000061
Figure BDA0001373021510000062
where m denotes that the i-th part of a prediction unit includes m minimum coding units, MVXmHorizontal vector representing the M-th smallest coding unit, MxAn average value of horizontal vectors representing m minimum coding units included in the ith part of the prediction unit; MVY (multifunction vehicle)mPerpendicular vector representing the mth smallest coding unit, MyAnd represents the average value of the vertical vectors of m minimum coding units contained in the ith part of the prediction unit, wherein m is a positive integer.
The average value of the horizontal vectors of the m smallest coding units contained in the i-th part of a certain prediction unit can be expressed as:
Figure BDA0001373021510000063
the average value of the vertical vectors of the m smallest coding units contained in the i-th part of a certain prediction unit can be expressed as:
Figure BDA0001373021510000071
the standard deviation of the horizontal vector and the standard deviation of the vertical vector can be expressed as:
Figure BDA0001373021510000072
Figure BDA0001373021510000073
step S4: and selecting the optimal prediction unit for coding according to the calculated motion homogeneity values of the different types of prediction units.
Comparing the motion homogeneity values of the different types of prediction units calculated in step S3, selecting two types of prediction units with the smallest and next smallest motion homogeneity values, and calculating the number of coding bits of the two types of prediction units and the corresponding distortion degree of the reconstructed image. According to the following formula:
J(s,c,m|Qp,λm)＝SSD(s,c,m|Qp)+λmR(s,c,m|Qp) (10)
wherein λ ismRepresenting the Lagrangian multiplier, QpRepresenting quantization parameter, m representing prediction mode, s representing coded block data, c representing reconstructed block data, R (s, c, m | Q)p) Representing the number of bits required to encode the current prediction mode, and the sum of squared errors SSD represents the image distortion. And calculating the rate distortion cost J of each prediction unit according to the formula, and selecting the prediction unit with the minimum rate distortion cost value as the optimal coding mode (namely the optimal division) of the current maximum coding unit after comparison.
It is to be emphasized that: and calculating the distortion (SAD) or the sum of squares of the residuals (SSE) of the two prediction units by adopting other corresponding formulas, and selecting the prediction unit with the minimum distortion or sum of squares of the residuals as the optimal coding mode (i.e. the optimal division) of the current maximum coding unit after comparison. Since the existing calculation method is used in step S4 to calculate the rate-distortion cost value, distortion degree or residual sum of squares, the calculation process will not be described in detail.
As shown in fig. 6, the present invention further provides an adaptive fast prediction unit partitioning apparatus based on motion homogeneity. The device comprises a dividing module 1, a motion estimation module 2, a storage module 3, a calculation module 4 and a judgment module 5. The adaptive fast prediction unit dividing device divides the current maximum coding unit into a plurality of minimum coding units, performs motion estimation on each minimum coding unit, and records the motion vector of each minimum coding unit; and calculating motion homogeneity values of various prediction units according to the motion vector of each minimum coding unit, and comparing the rate distortion cost values of the two selected prediction units with the minimum motion homogeneity value and the second minimum motion homogeneity value, thereby quickly determining the optimal coding mode of the current maximum coding unit.
The dividing module 1 receives a current maximum coding unit and divides the current maximum coding unit into a plurality of minimum coding units. In one embodiment of the present invention, as shown in fig. 4, a maximum coding unit of size 64x64 may be divided into 64 minimum coding units of size 8x 8.
The motion estimation module 2 may perform motion estimation on the plurality of minimum coding units divided by the division module 1, respectively. In an embodiment of the present invention, two reference frames (a reference frame at time t-1 and a reference frame at time t-2) before a current frame (time t) may be selected to perform integer pixel motion and quarter pixel motion estimation, respectively, and a motion vector generated after motion estimation is performed on each minimum coding unit is stored in the storage module 3.
The calculation module 4 may calculate the motion homogeneity values of different kinds of prediction units, respectively, based on the plurality of motion vectors stored by the storage module 3. The method for calculating the motion homogeneity values of different types of prediction units by the calculating module 4 is the same as that in step S3, and is not repeated herein.
The judging module 5 may compare the motion homogeneity values of the different types of prediction units calculated by the calculating module 4, select two prediction units with the smallest motion homogeneity value and the second smallest motion homogeneity value, calculate the rate-distortion cost J of each prediction unit, and select the prediction unit with the smallest rate-distortion cost value as the optimal coding mode (i.e., optimal division) of the current maximum coding unit through comparison. The determining module 5 may further calculate a distortion SAD or a sum of squared residuals SSE of each prediction unit, and select the prediction unit with the minimum distortion SAD or sum of squared residuals SSE as the optimal coding mode (i.e., the optimal partition) of the current maximum coding unit after comparison.
Compared with the prior art, the method shortens the selection range of the prediction unit in the current maximum coding unit in advance, reduces the search amount of the prediction unit, can effectively reduce the calculation complexity of video coding, improves the video coding efficiency, and has the characteristics of high coding quality, strong adaptability and the like.
The method and apparatus for partitioning adaptive fast prediction unit based on motion homogeneity provided by the present invention are described in detail above. It will be apparent to those skilled in the art that any obvious modifications thereto can be made without departing from the true spirit of the invention, which is to be accorded the full scope of the claims herein.

Claims (4)

1. A self-adaptive rapid prediction unit partitioning method based on motion homogeneity is characterized by comprising the following steps:
step S1: dividing the current maximum coding unit into a plurality of minimum coding units;
step S2: performing motion estimation on each minimum coding unit to obtain a motion vector;
in step S2, when motion estimation is performed on each minimum coding unit, two reference frames located in front of the current frame are selected to perform integer pixel motion and sub-pixel motion estimation, respectively, and the motion vector of each minimum coding unit is recorded;
step S3: respectively calculating motion homogeneity values of different kinds of prediction units based on the motion vectors;
step S4: selecting an optimal prediction unit for coding according to the motion homogeneity value;
in step S4, comparing the calculated motion homogeneity values of different prediction units, respectively calculating a rate distortion cost, a distortion degree, or a square sum of residuals of two prediction units with the minimum and the second smallest motion homogeneity values, and selecting the prediction unit with the minimum rate distortion cost value, the minimum distortion degree, or the minimum square of residuals as an optimal partition of a current maximum coding unit;
the motion homogeneity value of a certain prediction unit is expressed as:
Figure FFW0000021339240000011
wherein H represents the motion homogeneity value of a certain prediction unit, n represents that the prediction unit comprises n parts, HiRepresenting the motion homogeneity value of the ith part, wherein n and i are positive integers;
the motion homogeneity value of the i-th part is expressed as:
Hi＝(Sx+Sy)
wherein S isxRepresenting the variance of the horizontal vector, SyRepresents the variance of the vertical vector;
or the motion homogeneity value of the i-th part is expressed as:
Hi＝(Dx+Dy)
wherein D isxDenotes the standard deviation, D, of the horizontal vectoryThe standard deviation of the vertical vector is indicated.
2. The method of claim 1, wherein the adaptive fast prediction unit partition based on motion homogeneity comprises:
the variance of the horizontal vector and the variance of the vertical vector are respectively expressed as:
Figure FFW0000021339240000021
Figure FFW0000021339240000022
and the number of the first and second electrodes,
Figure FFW0000021339240000023
Figure FFW0000021339240000024
where m denotes that the i-th part of a prediction unit includes m minimum coding units, MVXmHorizontal vector representing the M-th smallest coding unit, MxAn average value of horizontal vectors representing m minimum coding units included in the ith part of the prediction unit; MVY (multifunction vehicle)mPerpendicular vector representing the mth smallest coding unit, MyAnd represents the average value of the vertical vectors of m minimum coding units contained in the ith part of the prediction unit, wherein m is a positive integer.
3. A self-adaptive rapid prediction unit dividing device based on motion homogeneity is characterized by comprising a dividing module, a motion estimation module, a storage module, a calculation module and a judgment module;
the dividing module receives a current maximum coding unit and divides the maximum coding unit into a plurality of minimum coding units;
the motion estimation module respectively carries out motion estimation on the minimum coding units divided by the division module;
the storage module stores a plurality of motion vectors generated by the motion estimation module;
the calculation module is used for calculating motion homogeneity values of different types of prediction units respectively based on the motion vectors stored by the storage module;
the judging module selects an optimal prediction unit for coding according to the motion homogeneity values of different prediction units calculated by the calculating module; comparing the calculated motion homogeneity values of different prediction units, respectively calculating the rate distortion cost, distortion degree or residual square sum of two selected prediction units with the minimum motion homogeneity value and the second smallest motion homogeneity value, and selecting the prediction unit with the minimum rate distortion cost value, the minimum distortion degree or the minimum residual square as the optimal division of the current maximum coding unit;
when the motion estimation module carries out motion estimation on each minimum coding unit, two reference frames positioned in front of a current frame are selected to carry out integer pixel motion and sub-pixel motion estimation respectively, and a motion vector of each minimum coding unit is recorded;
the motion homogeneity value of a certain prediction unit is expressed as:
Figure FFW0000021339240000031
wherein H represents the motion homogeneity value of a certain prediction unit, n represents that the prediction unit comprises n parts, HiDenotes the ith partThe motion homogeneity value of (1) is that n and i are positive integers;
the motion homogeneity value of the i-th part is expressed as:
Hi＝(Sx+Sy)
wherein S isxRepresenting the variance of the horizontal vector, SyRepresents the variance of the vertical vector;
or the motion homogeneity value of the i-th part is expressed as:
Hi＝(Dx+Dy)
wherein D isxDenotes the standard deviation, D, of the horizontal vectoryThe standard deviation of the vertical vector is indicated.
4. The apparatus for adaptive fast prediction unit partitioning based on motion homogeneity according to claim 3, wherein:
the variance of the horizontal vector and the variance of the vertical vector are respectively expressed as:
Figure FFW0000021339240000041
Figure FFW0000021339240000042
and the number of the first and second electrodes,
Figure FFW0000021339240000043
Figure FFW0000021339240000044
where m denotes that the i-th part of a prediction unit includes m minimum coding units, MVXmHorizontal vector representing the M-th smallest coding unit, MxAn average value of horizontal vectors representing m minimum coding units included in the ith part of the prediction unit; MVY (multifunction vehicle)mPerpendicular vector representing the mth smallest coding unit, MyIndicate the kind of preAnd measuring the average value of the vertical vectors of the m minimum coding units contained in the ith part of the unit, wherein m is a positive integer.
CN201710670690.5A 2017-08-08 2017-08-08 Self-adaptive rapid prediction unit partitioning method and device based on motion homogeneity Active CN107426573B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710670690.5A CN107426573B (en) 2017-08-08 2017-08-08 Self-adaptive rapid prediction unit partitioning method and device based on motion homogeneity

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710670690.5A CN107426573B (en) 2017-08-08 2017-08-08 Self-adaptive rapid prediction unit partitioning method and device based on motion homogeneity

Publications (2)

Publication Number Publication Date
CN107426573A CN107426573A (en) 2017-12-01
CN107426573B true CN107426573B (en) 2020-11-06

Family

ID=60437513

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710670690.5A Active CN107426573B (en) 2017-08-08 2017-08-08 Self-adaptive rapid prediction unit partitioning method and device based on motion homogeneity

Country Status (1)

Country Link
CN (1) CN107426573B (en)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1809167A (en) * 2006-01-26 2006-07-26 天津大学 Quick inter-frame forecast mode selection method
CN101640802B (en) * 2009-08-28 2012-06-20 北京工业大学 Video inter-frame compression coding method based on macroblock features and statistical properties
CN102930559A (en) * 2012-10-23 2013-02-13 华为技术有限公司 Image processing method and device
CN105791856A (en) * 2010-11-23 2016-07-20 Lg电子株式会社 Method for encoding and decoding images, and device using same

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20080130749A1 (en) * 2006-05-23 2008-06-05 Chang-Che Tsai Method for Performing Pattern-Based Block Motion Estimation
CN104065973B (en) * 2013-03-20 2017-11-17 华为技术有限公司 A kind of method and device of high-performance video coding search
CN103491371B (en) * 2013-09-04 2017-04-12 华为技术有限公司 Encoding method, device and equipment based on hierarchy
US10097851B2 (en) * 2014-03-10 2018-10-09 Euclid Discoveries, Llc Perceptual optimization for model-based video encoding
CN104079937B (en) * 2014-07-16 2017-06-23 山东大学 It is a kind of based on motion vector analysis by H.264 to the quick interframe code-transferring method and transcoding device of HEVC
CN104754338A (en) * 2014-12-30 2015-07-01 北京奇艺世纪科技有限公司 Selection method and device for intra-frame predication mode

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1809167A (en) * 2006-01-26 2006-07-26 天津大学 Quick inter-frame forecast mode selection method
CN101640802B (en) * 2009-08-28 2012-06-20 北京工业大学 Video inter-frame compression coding method based on macroblock features and statistical properties
CN105791856A (en) * 2010-11-23 2016-07-20 Lg电子株式会社 Method for encoding and decoding images, and device using same
CN102930559A (en) * 2012-10-23 2013-02-13 华为技术有限公司 Image processing method and device

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
《高效视频编码》;沈燕飞，李锦涛，朱珍民，张勇东;《计算机学报》;20131130;第36卷(第11期);全文 *

Also Published As

Publication number Publication date
CN107426573A (en) 2017-12-01

Similar Documents

Publication Publication Date Title
CN109417631B (en) Video coding with adaptive motion information refinement
EP2772057B1 (en) Method and device for determining parameters for encoding or decoding of an image of a video sequence
US20190208223A1 (en) Method and apparatus for video coding with automatic motion information refinement
US7580456B2 (en) Prediction-based directional fractional pixel motion estimation for video coding
TW201640894A (en) Overlapped motion compensation for video coding
KR20230007313A (en) Parallelized Rate-Distortion Optimized Quantization Using Deep Learning
JP2002330443A (en) Method for efficiently predicting position change in video in digital moving picture system
US8462849B2 (en) Reference picture selection for sub-pixel motion estimation
CN112055203B (en) Inter-frame prediction method, video coding method and related devices
AU2021286043B2 (en) Encoding and decoding method and apparatus, and device therefor
Ma et al. Low complexity rate distortion optimization for HEVC
CN112004097B (en) Inter-frame prediction method, image processing apparatus, and computer-readable storage medium
CN109660811B (en) Rapid HEVC inter-frame coding method
KR20180037042A (en) A motion vector field coding method and a decoding method, and a coding and decoding apparatus
CN113965753A (en) Inter-frame image motion estimation method and system based on code rate control
CN111263144B (en) Motion information determination method and equipment
US20180139451A1 (en) Refinement of a low-pel resolution motion estimation vector
CN107613305B (en) P, B frame rapid motion estimation method in HEVC
CN109547798B (en) Rapid HEVC inter-frame mode selection method
CN107426573B (en) Self-adaptive rapid prediction unit partitioning method and device based on motion homogeneity
KR101545195B1 (en) Apparatus and method for fast Intra Prediction Algorithm
CN117615129A (en) Inter-frame prediction method, inter-frame prediction device, computer equipment and storage medium
CN112601081B (en) Adaptive partition multi-prediction method and device
CA3198095A1 (en) Multi-pass decoder-side motion vector refinement
KR100723840B1 (en) Video motion estimation device

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant