WO2023090894A1

WO2023090894A1 - Method, apparatus, and recording medium for image encoding/decoding

Info

Publication number: WO2023090894A1
Application number: PCT/KR2022/018201
Authority: WO
Inventors: 권형진; 김동현; 김연희; 김종호; 도지훈; 이주영; 임웅; 정세윤; 최진수; 이태진; 김재곤; 박도현; 문기화
Original assignee: 한국전자통신연구원; 한국항공대학교산학협력단
Priority date: 2021-11-17
Filing date: 2022-11-17
Publication date: 2023-05-25

Abstract

A method, apparatus, and recording medium for image encoding/decoding are disclosed. In the method for image encoding/decoding, an intra-prediction mode for a target block is induced, and a reference sample for intra-prediction is generated. The intra-prediction for the target block is executed by using the induced intra-prediction mode and the generated reference sample. A machine learning-based intra-prediction mode inducing method may be used for inducing the intra-prediction mode. In the intra-prediction, machine learning-based prediction, prediction sample blending, prediction sample improvement, and prediction sample scaling may be used.

Description

Method, apparatus and recording medium for video encoding/decoding

The present invention relates to a method, apparatus and recording medium for video encoding/decoding.

The present invention claims the benefit of the filing date of Korean Patent Application No. 10-2021-0158574 filed on November 17, 2021 and the benefit of the filing date of Korean Patent Application No. 10-2022-0154359 filed on November 17, 2022, All of its contents are incorporated herein.

Through the continuous development of the information communication industry, broadcasting services having high definition (HD) resolution have been spread worldwide. Through this proliferation, many users have become accustomed to high-resolution and high-definition images and/or videos.

In order to satisfy users' demand for high image quality, many organizations are accelerating the development of next-generation video devices. Users' interest in Ultra High Definition (UHD) TV, which has four times higher resolution than FHD TV, as well as High Definition TV (HDTV) and Full HD (FHD) TV. has increased, and according to this increase in interest, an image encoding/decoding technology for an image having a higher resolution and quality is required.

As video compression technology, there are various technologies such as inter prediction technology, intra prediction technology, transformation and quantization technology, and entropy encoding technology.

The inter-prediction technique is a technique of predicting a value of a pixel included in a current picture using pictures before and/or after the current picture. Intra-prediction technology is a technology of predicting a value of a pixel included in a current picture by using information about pixels in the current picture. The transformation and quantization technique is a technique for compressing the energy of the residual image. The entropy coding technique assigns short codes to values with a high frequency of occurrence and assigns long codes to values with a low frequency of occurrence.

Using this video compression technology, data for video can be effectively compressed, transmitted, and stored.

An embodiment may provide a device, method, and recording medium for determining prediction information using machine learning.

An embodiment may provide an apparatus, method, and recording medium for determining an intra prediction mode using machine learning.

In one aspect, deriving an intra prediction mode of a target block; constructing a reference sample for the intra prediction; and performing the intra prediction on the target block using the intra prediction mode and the reference sample.

The intra prediction mode may be a prediction mode including intra prediction.

The intra prediction modes include a Position Dependent Intra Prediction Combination (PDPC) mode, a Cross-Component Linear Model (CCLM) mode, and an Intra Sub-Partitions (ISP) mode. mode, Multiple Reference Line (MRL) mode, Matrix weighted Intra Prediction (MIP) mode, Intra Block Copy (IBC) mode, or Machine Learning (ML) mode. can

The ML mode may include one or more of machine learning-based prediction, prediction sample blending, prediction sample improvement, and prediction sample scaling.

The intra prediction mode may be derived using a Most Probable Mode (MPM) list.

The MPM list may include a first MPM list and a second MPM list.

The first MPM list may include an intra prediction mode using machine learning.

The second MPM list may include an intra prediction mode that does not use machine learning.

In constructing the MPM list, an intra prediction mode of a neighboring block of the target block may be replaced with another intra prediction mode, and the replaced intra prediction mode may be added to the MPM list.

The intra prediction mode may be derived using a machine learning-based intra prediction mode derivation method.

The machine learning-based intra prediction mode derivation method includes a single layer perceptron (SLP), a multi-layer perceptron (MLP), a deep neural network (DNN), and a convolutional neural network (CNN). And it may be performed using one or more of a recurrent neural network (RNN).

An output of the machine learning-based intra prediction mode derivation method may be one or more probabilities of one or more intra prediction modes.

The reference sample may be constructed using reference sample filtering.

The reference sample may be constructed using reference sample scaling.

The intra prediction may be performed using machine learning based prediction.

The intra prediction may be performed using predictive sample blending.

The intra prediction may be performed using prediction sample refinement.

The intra prediction mode may be derived using a probability of a transform mode.

The intra prediction mode may be derived using a probability of a transformation mode set.

In another aspect, deriving an intra prediction mode of the target block; constructing a reference sample for the intra prediction; and performing the intra prediction on the target block using the intra prediction mode and the reference sample.

In another aspect, in a computer-readable recording medium storing a bitstream for video decoding, the bitstream includes encoded coding information, and decoding of the encoded coding information is performed to obtain the coding information. is generated, an intra prediction mode of a target block is derived using the coding information, a reference sample for the intra prediction is configured, and the intra prediction of the target block is performed using the intra prediction mode and the reference sample. A computer readable recording medium on which this is performed is provided.

An apparatus, method, and recording medium for determining predictive information using machine learning are provided.

An embodiment provides an apparatus, method, and recording medium for determining an intra prediction mode using machine learning.

1 is a block diagram showing a configuration according to an embodiment of an encoding device to which the present invention is applied.

2 is a block diagram showing a configuration according to an embodiment of a decoding device to which the present invention is applied.

3 is a diagram schematically illustrating a division structure of an image when encoding and decoding an image.

4 is a diagram illustrating a form of a prediction unit that a coding unit may include.

5 is a diagram illustrating a form of a transform unit that may be included in a coding unit.

6 shows division of a block according to an example.

7 is a diagram for explaining an embodiment of an intra prediction process.

8 is a diagram for explaining reference samples used in an intra prediction process.

9 is a diagram for explaining an embodiment of an inter prediction process.

10 shows spatial candidates according to an example.

11 illustrates an order of adding motion information of spatial candidates to a merge list according to an example.

12 illustrates a process of transformation and quantization according to an example.

13 illustrates diagonal scanning according to an example.

14 illustrates horizontal scanning according to an example.

15 illustrates vertical scanning according to an example.

16 is a structural diagram of an encoding device according to an embodiment.

17 is a structural diagram of a decryption device according to an embodiment.

18 is a flowchart of a method of predicting a target block and generating a bitstream according to an embodiment.

19 is a flowchart of a method of predicting a target block using a bitstream according to an embodiment.

20 is a flowchart of a prediction method according to an embodiment.

21 illustrates a plurality of reference sample lines of a target block according to an example.

22 illustrates a reference sample area composed of one area according to an example.

23 illustrates a reference sample area composed of two areas according to an example.

24 illustrates a reference sample area composed of two areas according to an example.

25 illustrates a reference sample area composed of three areas according to an example.

26 illustrates a first method in which an available sample replaces an unavailable sample to be used as a reference sample according to an example.

27 illustrates a second method in which an available sample replaces an unavailable sample to be used as a reference sample according to an example.

28 illustrates first down-scaling using sub-sampling of reference samples of a target block according to an example.

29 illustrates second down-scaling using sub-sampling of reference samples of a target block according to an example.

30 illustrates up-scaling using interpolation of reference samples of a target block according to an example.

31 illustrates an ML mode using one reference sample as an input according to an example.

32 illustrates an ML mode using N reference samples as inputs according to an example.

33 illustrates division in an inner layer of a machine learning-based intra prediction model according to an example.

34 illustrates division into M branches in one layer according to an example.

35 illustrates division into two or more branches in M-1 layers according to an example.

36 illustrates a weight derivation model for performing predictive sample blending according to an embodiment.

37 illustrates one weight output for a prediction block according to an example.

38 illustrates weights output for prediction samples of a prediction block according to an example.

39 illustrates a target block and reference samples adjacent to the target block according to an example.

40 illustrates output generated by processing for a context according to an example.

Since the present invention can make various changes and have various embodiments, specific embodiments will be illustrated in the drawings and described in detail in the detailed description. However, this is not intended to limit the present invention to specific embodiments, and should be understood to include all modifications, equivalents, and substitutes included in the spirit and scope of the present invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS For detailed descriptions of exemplary embodiments described below, reference is made to the accompanying drawings, which illustrate specific embodiments by way of example. These embodiments are described in sufficient detail to enable those skilled in the art to practice the embodiments. It should be understood that the various embodiments are different, but need not be mutually exclusive. For example, specific shapes, structures, and characteristics described herein may be implemented in one embodiment in another embodiment without departing from the spirit and scope of the invention. Additionally, it should be understood that the location or arrangement of individual components within each disclosed embodiment may be changed without departing from the spirit and scope of the embodiment. Accordingly, the detailed description set forth below is not to be taken in a limiting sense, and the scope of the exemplary embodiments, if properly described, is limited only by the appended claims, along with all equivalents as claimed by those claims.

Like reference numbers in the drawings indicate the same or similar function throughout the various aspects. The shapes and sizes of elements in the drawings may be exaggerated for clarity.

In the present invention, terms such as first and second may be used to describe various components, but the components should not be limited by the terms. These terms are only used for the purpose of distinguishing one component from another. For example, a first element may be termed a second element, and similarly, a second element may be termed a first element, without departing from the scope of the present invention. The term "and/or" may include any combination of a plurality of related listed items or any of a plurality of related listed items.

When a component is referred to as being “connected” or “connected” to another component, the two components may be directly connected or connected to each other, but It will be understood that other components may exist in the middle of the two components. On the other hand, when a component is referred to as "directly connected" or "directly connected" to another component, it should be understood that no other component exists in the middle of the above two components. something to do.

Elements appearing in the embodiments are shown independently to represent different characteristic functions, and do not mean that each element is composed of separate hardware or a single software unit. That is, each component is listed and included as each component for convenience of description, and at least two of each component are combined to form one component, or one component is divided into a plurality of components to perform functions. It can be performed, and integrated embodiments and separate embodiments of each of these components are also included in the scope of the present invention as long as they do not depart from the essence of the present invention.

Terms used in the embodiments are only used to describe specific embodiments, and are not intended to limit the present invention. Singular expressions include plural expressions unless the context clearly dictates otherwise. In embodiments, the terms "comprise" or "have" are intended to indicate that there is a feature, number, step, operation, component, part, or combination thereof described in the specification, but one or more other It should be understood that the presence or addition of features, numbers, steps, operations, components, parts, or combinations thereof is not precluded. That is, the description of "including" a specific configuration in the embodiments does not exclude configurations other than the corresponding configuration, and means that additional configurations may also be included in the practice of the present invention or the scope of the technical idea of the present invention. .

In embodiments, the term “at least one” may mean one or more numbers such as 1, 2, 3, and 4. In embodiments, the term “a plurality of” may mean one of two or more numbers, such as 2, 3, and 4.

Some of the components of the embodiments are not essential components that perform essential functions in the present invention, but may be optional components for improving performance. Embodiments may be implemented by including only essential components in implementing the essence of the embodiments, excluding components used for performance improvement. A structure including only essential components excluding optional components used for performance improvement is also included in the scope of the embodiments.

Hereinafter, embodiments will be described in detail with reference to the accompanying drawings so that those skilled in the art can easily practice the embodiments. In describing the embodiments, if it is determined that a detailed description of a related known configuration or function may obscure the gist of the present specification, the detailed description will be omitted. In addition, the same reference numerals are used for the same components in the drawings, and duplicate descriptions of the same components are omitted.

Hereinafter, an image may mean one picture constituting a video, and may also indicate the video itself. For example, "encoding and/or decoding an image" may mean "encoding and/or decoding a video", and may mean "encoding and/or decoding one of images constituting a video". may be

Hereinafter, the terms “video” and “motion picture(s)” may be used interchangeably and may be used interchangeably.

Hereinafter, the target image may be an encoding target image that is an encoding target and/or a decoding target image that is a decoding target. Also, the target image may be an input image input to an encoding device or an input image input to a decoding device. Also, the target image may be a current image that is a target of current encoding and/or decoding. For example, the terms “target image” and “current image” may be used interchangeably and may be used interchangeably.

Hereinafter, the terms “image”, “picture”, “frame” and “screen” may be used interchangeably and interchangeably.

Hereinafter, the target block may be an encoding target block that is an encoding target and/or a decoding target block that is a decoding target. Also, the target block may be a current block that is a target of current encoding and/or decoding. For example, the terms “target block” and “current block” may be used interchangeably and may be used interchangeably. The current block may refer to a coding target block that is an encoding target during encoding and/or a decoding target block that is a decoding target during decoding. Also, the current block may be at least one of a coding block, a prediction block, a residual block, and a transform block.

Hereinafter, the terms “block” and “unit” may be used interchangeably and may be used interchangeably. Or "block" may represent a specific unit.

In the following, the terms "region" and "segment" may be used interchangeably.

In embodiments, each of the specified information, data, flag, index and element, attribute, etc. may have a value. A value "0" of information, data, flags, indexes, elements, and attributes may represent false, logical false, or a first predefined value. That is to say, the value "0", false, logic false and the first predefined value may be used interchangeably. The value "1" of information, data, flags, indexes, elements, and attributes may represent true, logical true, or a second predefined value. In other words, the value "1", true, logically true, and the second predefined value may be used interchangeably.

When a variable such as i or j is used to indicate a row, column, or index, the value of i may be an integer greater than or equal to 0, or may be an integer greater than or equal to 1. That is to say, row, column, index, etc. may be counted from 0, in embodiments, may be counted from 1.

In embodiments, the term “one or more” or the term “at least one” may mean the term “plurality”. “One or more” or “at least one” may be used interchangeably with “plural”.

Below, terms used in the embodiments are explained.

Encoder: An encoder may mean a device that performs encoding. In other words, an encoder may mean an encoding device.

Decoder: A decoder may refer to a device that performs decoding. In other words, a decryptor may mean a decryption device.

Unit: A unit may represent a unit of encoding and/or decoding of an image. The terms "unit" and "block" may be used interchangeably and may be used interchangeably.

- A unit may be an MxN array of samples. M and N may each be a positive integer. A unit may refer to an array of samples in a two-dimensional form.

- In encoding and decoding of an image, a unit may be a region created by dividing one image. In other words, a unit may be a specified area within one image. One image may be divided into a plurality of units. Alternatively, the unit may refer to the divided parts when one image is divided into subdivided parts and encoding or decoding of the divided parts is performed.

- In encoding and decoding of an image, a predefined process for a unit may be performed according to a unit type.

- Depending on the function, the type of unit is a macro unit, a coding unit (CU), a prediction unit (PU), a residual unit, and a transform unit (TU), etc. can be classified as Or, depending on the function, the unit is a block, macroblock, coding tree unit, coding tree block, coding unit, coding block, or prediction unit. It may mean a prediction unit, a prediction block, a residual unit, a residual block, a transform unit, a transform block, and the like. For example, the target unit may be at least one of a CU, a PU, a residual unit, and a TU to be encoded and/or decoded.

- A unit may refer to information including a luma component block, a chroma component block corresponding to the luma component block, and a syntax element for each block, in order to be referred to as a block.

- Units may vary in size and shape. Also, units can have various sizes and shapes. In particular, the shape of the unit may include not only a square but also a two-dimensional geometric figure such as a rectangle, a trapezoid, a triangle, and a pentagon.

- In addition, the unit information may include at least one or more of a unit type, a size of a unit, a depth of a unit, a coding order of a unit, a decoding order of a unit, and the like. For example, the unit type may indicate one of CU, PU, residual unit, and TU.

- One unit can be further divided into sub-units having a smaller size than the unit.

Depth: Depth may mean the degree of division of a unit. Also, the depth of a unit may indicate a level at which the unit exists when the unit(s) are expressed as a tree structure.

- Unit division information may include depth about the depth of the unit. Depth may indicate the number and/or extent to which a unit is divided.

- In the tree structure, the root node has the shallowest depth and the leaf node has the deepest depth. The root node may be the highest node. A leaf node may be the lowest node.

- One unit may be hierarchically divided into a plurality of sub-units while having depth information based on a tree structure. In other words, a unit and a sub-unit generated by division of the unit may correspond to a node and a child node of the node, respectively. Each divided sub-unit may have a depth. Since depth represents the number and/or degree of division of a unit, division information of a sub-unit may include information about the size of the sub-unit.

- In the tree structure, the highest node may correspond to the first non-split unit. The highest node may be referred to as a root node. Also, the highest node may have the smallest depth value. In this case, the highest node may have a depth of level 0.

- A node with a depth of level 1 may represent a unit created as the original unit is split once. A node with a depth of level 2 may represent a unit created as the original unit is split twice.

- A node with a depth of level n may represent a unit created as the initial unit is split n times.

- A leaf node may be the lowest node and may be a node that cannot be further divided. The depth of a leaf node may be the maximum level. For example, the predefined value of the maximum level may be 3.

-QT depth may indicate the depth for quad division. BT depth may indicate a depth for binary partitioning. The TT depth may indicate a depth for ternary division.

Sample: A sample may be a base unit constituting a block. A sample may be represented as values from 0 to 2 ^Bd -1 according to a bit depth (Bd).

- A sample can be a pixel or a pixel value.

- In the following, the terms "pixel", "pixel" and "sample" may be used in the same meaning and may be used interchangeably.

Coding Tree Unit (CTU): A CTU may consist of one luma component (Y) coding tree block and two chroma component (Cb, Cr) coding tree blocks related to the luma component coding tree block. there is. In addition, the CTU may mean including syntax elements for the above blocks and each block of the above blocks.

- Each coding tree unit is a quad tree (QT), binary tree (BT), and ternary tree (TT) to construct sub units such as a coding unit, a prediction unit, and a transform unit. It can be segmented using one or more segmentation schemes. A quad tree may mean a quarternary tree. In addition, each coding tree unit may be split using a MultiType Tree (MTT) using one or more splitting schemes.

- CTU may be used as a term to refer to a pixel block, which is a processing unit in the process of decoding and encoding an image, as in segmentation of an input image.

Coding Tree Block (CTB): A coding tree block may be used as a term to refer to any one of a Y coding tree block, a Cb coding tree block, and a Cr coding tree block.

Neighbor Block: A neighboring block may mean a block adjacent to a target block. A neighboring block may mean a reconstructed neighboring block.

- In the following, the terms "neighboring block" and "adjacent block" may be used in the same meaning and may be used interchangeably.

- A neighboring block may mean a reconstructed neighboring block.

Spatial neighbor block: A spatial neighbor block may be a block that is spatially adjacent to the target block. Neighboring blocks may include spatial neighboring blocks.

- The target block and spatial neighboring blocks may be included in the target picture.

- A spatial neighboring block may mean a block whose boundary meets the target block or a block located within a predetermined distance from the target block.

- A spatial neighboring block may mean a block adjacent to a vertex of a target block. Here, the block adjacent to the vertex of the target block may be a block vertically adjacent to a neighboring block horizontally adjacent to the target block or a block horizontally adjacent to a neighboring block vertically adjacent to the target block.

Temporal neighbor block: A temporal neighbor block may be a block that is temporally adjacent to the target block. Neighboring blocks may include temporal neighboring blocks.

- A temporal neighboring block may include a co-located block (col block).

- A collocated block may be a block in an already reconstructed co-located picture (col picture). A position of a collocated block in a collocated picture may correspond to a position of a target block in a target picture. Alternatively, the position of the collocated block in the collocated picture may be the same as the position of the target block in the target picture. A collocated picture may be a picture included in a reference picture list.

- A temporal neighboring block may be a block temporally adjacent to a spatial neighboring block of a target block.

Prediction mode: The prediction mode may be information indicating a mode used for intra prediction or a mode used for inter prediction.

Prediction unit: A prediction unit may mean a base unit for prediction such as inter prediction, intra prediction, inter compensation, intra compensation, and motion compensation.

- One prediction unit may be divided into a plurality of smaller-sized partitions or sub-prediction units. A plurality of partitions may also be a basis unit in performing prediction or compensation. A partition generated by division of a prediction unit may also be a prediction unit.

Prediction unit partition (prediction unit partition): A prediction unit partition may mean a form in which a prediction unit is divided.

Reconstructed neighboring unit: A reconstructed neighboring unit may be a unit that has already been decoded and reconstructed in a neighbor of a target unit.

- The reconstructed neighbor unit may be a spatial neighbor unit or a temporal neighbor unit to the target unit.

- The reconstructed spatial neighboring unit may be a unit in the target picture and already reconstructed through encoding and/or decoding.

- The reconstructed temporal neighbor unit may be a unit in the reference picture and a unit that has already been reconstructed through encoding and/or decoding. A position of the reconstructed temporal neighboring unit within the reference image may be the same as a position within the target unit's target picture or may correspond to a position within the target unit's target picture. Alternatively, the reconstructed temporal neighboring unit may be a neighboring block of a corresponding block in the reference picture. Here, the position of the corresponding block in the reference image may correspond to the position of the target block in the target image. Here, the correspondence of the positions of the blocks may mean that the positions of the blocks are the same, and may mean that one block is included in another block, and one block occupies a specified position of the other block. can mean doing

Sub-picture: A picture can be divided into one or more sub-pictures. A sub-picture may consist of one or more tile rows and one or more tile columns.

- A sub-picture may be an area having a square shape or a rectangular (ie, non-square) shape within a picture. In addition, a sub-picture may include one or more CTUs. .

- A sub-picture may be a rectangular area of one or more slices within one picture.

- One sub-picture may include one or more tiles, one or more bricks, and/or one or more slices.

Tile: A tile can be a square or rectangular (ie, non-square) area within a picture.

- A tile may contain one or more CTUs.

- A tile can be divided into one or more bricks.

Brick: A brick may mean one or more CTU rows within a tile.

- A tile can be divided into one or more bricks. Each brick may contain one or more CTU rows.

- A tile that is not divided into two or more can also mean a brick.

Slice: A slice can include one or more tiles within a picture. Or, a slice may include one or more bricks within a tile.

- A sub-picture may contain one or more slices that collectively cover a rectangular area within the picture. Accordingly, each sub-picture boundary may always be a slice boundary. Also, each vertical sub-picture boundary may always be a vertical tile boundary.

Parameter set: A parameter set may correspond to header information among structures in a bitstream.

- Parameter sets include a Video Parameter Set (VPS), a Sequence Parameter Set (SPS), a Picture Parameter Set (PPS), an Adaptation Parameter Set (APS), and a decoding parameter It may include at least one of a set (Decoding Parameter Set; DPS), and the like.

Information signaled through a parameter set may be applied to pictures referring to the parameter set. For example, information in the VPS may be applied to pictures referring to the VPS. Information in the SPS may be applied to pictures referring to the SPS. Information in the PPS may be applied to pictures referring to the PPS.

A parameter set may refer to an upper parameter set. For example, PPS may refer to SPS. SPS may refer to VPS.

- In addition, the parameter set may include tile group, slice header information, and tile header information. A tile group may refer to a group including a plurality of tiles. Also, the meaning of a tile group may be the same as that of a slice.

Rate-distortion optimization: The encoding device uses a combination of the size of the coding unit, the prediction mode, the size of the prediction unit, motion information, and the size of the conversion unit to provide high encoding efficiency. Distortion optimization can be used.

- The rate-distortion optimization method may calculate a rate-distortion cost of each combination in order to select an optimal combination among the above combinations. The rate-distortion cost can be calculated using the formula “D+λ*R”. In general, a combination that minimizes the rate-distortion cost according to the equation “D+λ*R” can be selected as an optimal combination in the rate-distortion optimization method.

- D can represent distortion. D may be the mean square error of difference values between the original transform coefficients and the reconstructed transform coefficients within the transform unit.

- R can represent rate. R may represent a bit rate using related context information.

- λ may represent a Lagrangian multiplier. R may include not only coding parameter information such as a prediction mode, motion information, and a coded block flag, but also bits generated by encoding transform coefficients.

- The encoding device may perform processes such as inter prediction, intra prediction, transformation, quantization, entropy encoding, inverse quantization, and/or inverse transformation to calculate accurate D and R. These processes can greatly increase complexity in an encoding device.

Bitstream: A bitstream may mean a string of bits including coded image information.

Parsing: Parsing may mean determining a value of a syntax element by entropy decoding a bitstream. Alternatively, parsing may mean entropy decoding itself.

Symbol: may mean at least one of a syntax element, a coding parameter, and a transform coefficient of a coding target unit and/or a decoding target unit. Also, a symbol may mean an object of entropy encoding or a result of entropy decoding.

Reference picture: A reference picture may refer to an image that a unit refers to for inter prediction or motion compensation. Alternatively, the reference picture may be an image including a reference unit referred to by a target unit for inter prediction or motion compensation.

Hereinafter, the terms “reference picture” and “reference image” may be used interchangeably and may be used interchangeably.

Reference picture list: The reference picture list may be a list including one or more reference pictures used for inter prediction or motion compensation.

- The types of reference picture lists are List Combined (LC), List 0 (L0), List 1 (L1), List 2 (List 2; L2), and List 3 (List 3; L3). ), etc. may be present.

- One or more reference picture lists may be used for inter prediction.

Inter prediction indicator: The inter prediction indicator may indicate the direction of inter prediction for a target unit. Inter prediction can be one of uni-prediction and bi-prediction, etc. Alternatively, the inter prediction indicator may indicate the number of reference pictures used when generating a prediction unit of a target unit. Alternatively, the inter prediction indicator may indicate the number of prediction blocks used for inter prediction or motion compensation of the target unit.

Prediction list utilization flag: The prediction list utilization flag may indicate whether a prediction unit is generated using at least one reference picture in a specific reference picture list.

- An inter prediction indicator can be derived using the prediction list utilization flag. Conversely, the prediction list utilization flag can be derived using the inter prediction indicator. For example, when the prediction list utilization flag indicates a first value of 0, it may indicate that a prediction block is not generated using a reference picture in the reference picture list for the target unit. When the prediction list utilization flag indicates the second value of 1, it may indicate that a prediction unit is generated using the reference picture list for the target unit.

Reference picture index: The reference picture index may be an index indicating a specific reference picture in a reference picture list.

Picture Order Count (POC): The POC of a picture may indicate a display order of pictures.

Motion Vector (MV): A motion vector may be a two-dimensional vector used in inter prediction or motion compensation. A motion vector may mean an offset between a target image and a reference image.

- For example, MV can be expressed as (mv _x , mv _y ). mv _x may represent a horizontal component, and mv _y may represent a vertical component.

Search range: The search range may be a two-dimensional area where MVs are searched during inter prediction. For example, the size of the search area may be MxN. M and N may each be a positive integer.

Motion vector candidate: A motion vector candidate may mean a block as a prediction candidate or a motion vector of a block as a prediction candidate when predicting a motion vector.

- A motion vector candidate may be included in a motion vector candidate list.

Motion vector candidate list: The motion vector candidate list may refer to a list constructed using one or more motion vector candidates.

Motion vector candidate index: The motion vector candidate index may mean an indicator indicating a motion vector candidate in the motion vector candidate list. Alternatively, the motion vector candidate index may be an index of a motion vector predictor.

Motion information: Motion information includes not only motion vectors, reference picture indices and inter prediction indicators, but also reference picture list information, reference pictures, motion vector candidates, motion vector candidate indices, merge candidates and merge indices, etc. It may mean information including at least one of

Merge candidate list: A merge candidate list may refer to a list constructed using one or more merge candidates.

Merge candidate: A merge candidate is a spatial merge candidate, a temporal merge candidate, a combined merge candidate, a combined bi-prediction merge candidate, a candidate based on history, a candidate based on an average of two candidates, and zero It may mean a merge candidate and the like. A merge candidate may include an inter prediction indicator, and may include motion information such as a reference picture index for each list, a motion vector, a prediction list utilization flag, and an inter prediction indicator.

Merge index: A merge index may be an indicator pointing to a merge candidate in a merge candidate list.

- The merge index may indicate a reconstructed unit that derives a merge candidate from among reconstructed units spatially adjacent to the target unit and reconstructed units temporally adjacent to the target unit.

- The merge index may indicate at least one piece of motion information of a merge candidate.

Transform unit: A transform unit may be a basic unit in residual signal encoding and/or residual signal decoding, such as transform, inverse transform, quantization, inverse quantization, transform coefficient encoding, and transform coefficient decoding. One transform unit may be divided into a plurality of sub-transform units having smaller sizes. Here, the transformation may include one or more of a first-order transformation and a second-order transformation, and the inverse transformation may include one or more of a first-order inverse transformation and a second-order inverse transformation.

Scaling: Scaling may refer to a process of multiplying a transform coefficient level by a factor.

- As a result of the scaling to the transform coefficient level, a transform coefficient can be generated. Scaling may be referred to as dequantization.

Quantization Parameter (QP): A quantization parameter may mean a value used when generating a transform coefficient level for a transform coefficient in quantization. Alternatively, the quantization parameter may refer to a value used when generating a transform coefficient by scaling a transform coefficient level in inverse quantization. Alternatively, the quantization parameter may be a value mapped to a quantization step size.

Delta quantization parameter: The delta quantization parameter may refer to a difference value between a predicted quantization parameter and a quantization parameter of a target unit.

Scan: A scan may refer to a method of arranging the order of coefficients within a unit, block or matrix. For example, arranging a 2D array into a 1D array may be referred to as a scan. Alternatively, arranging a one-dimensional array into a two-dimensional array may also be referred to as scan or inverse scan.

Transform coefficient: A transform coefficient may be a coefficient value generated by performing transformation in an encoding device. Alternatively, the transform coefficient may be a coefficient value generated by performing at least one of entropy decoding and inverse quantization in the decoding apparatus.

- A quantized level generated by applying quantization to a transform coefficient or a residual signal or a quantized transform coefficient level may also be included in the meaning of a transform coefficient.

Quantized level: A quantized level may refer to a value generated by performing quantization on a transform coefficient or a residual signal in an encoding device. Alternatively, the quantized level may mean a value to be subjected to inverse quantization when the decoding apparatus performs inverse quantization.

- A quantized transform coefficient level, which is a result of transform and quantization, may also be included in the meaning of the quantized level.

Non-zero transform coefficient: A non-zero transform coefficient may mean a transform coefficient having a non-zero value or a transform coefficient level having a non-zero value. Alternatively, the non-zero transform coefficient may refer to a transform coefficient whose value is not 0 or a transform coefficient level whose value is not 0.

Quantization matrix: A quantization matrix may mean a matrix used in a quantization process or an inverse quantization process to improve subjective or objective picture quality of an image. A quantization matrix may also be referred to as a scaling list.

Quantization matrix coefficient: A quantization matrix coefficient may mean each element in a quantization matrix. Quantization matrix coefficients may also be referred to as matrix coefficients.

Default matrix: The default matrix may be a quantization matrix predefined in an encoding device and a decoding device.

Non-default matrix: The non-default matrix may be a quantization matrix that is not predefined in the encoding device and the decoding device. The non-default matrix may refer to a quantization matrix signaled from an encoding device to a decoding device by a user.

Most Probable Mode (MPM): MPM may indicate an intra prediction mode that is highly likely to be used for intra prediction of a target block.

- The encoding device and the decoding device may determine one or more MPMs based on a coding parameter related to the target block and an attribute of an object related to the target block.

- The encoding device and the decoding device may determine one or more MPMs based on the intra prediction mode of the reference block. Reference blocks may be plural. The plurality of reference blocks may include a spatial neighboring block adjacent to the left side of the target block and a spatial neighboring block adjacent to the top of the target block. In other words, one or more different MPMs may be determined depending on which intra prediction modes are used for reference blocks.

- One or more MPMs may be determined in the same way in the encoding device and the decoding device. In other words, the encoding device and the decoding device may share an MPM list including one or more identical MPMs.

MPM List: An MPM list can be a list containing one or more MPMs. The number of one or more MPMs in the MPM list may be predefined.

MPM indicator: The MPM indicator may indicate an MPM used for intra prediction of a target block among one or more MPMs in the MPM list. For example, the MPM indicator may be an index to an MPM list.

- Since the MPM list is determined in the same way in the encoding device and the decoding device, the MPM list itself may not need to be transmitted from the encoding device to the decoding device.

- The MPM indicator may be signaled from the encoding device to the decoding device. As the MPM indicator is signaled, the decoding apparatus may determine an MPM to be used for intra prediction of the target block among MPMs in the MPM list.

MPM use indicator: The MPM use indicator may indicate whether an MPM use mode is to be used for prediction of a target block. The MPM use mode may be a mode for determining an MPM to be used for intra prediction of a target block by using an MPM list.

- The MPM use indicator may be signaled from the encoding device to the decoding device.

Signaling: Signaling may indicate that information is transmitted from an encoding device to a decoding device. Alternatively, signaling may mean that an encoding device includes information in a bitstream or a recording medium. Information signaled by the encoding device may be used by the decoding device.

- The encoding device may generate encoded information by performing encoding on signaled information. Encoded information may be transmitted from an encoding device to a decoding device. The decoding apparatus may obtain information by decoding the transmitted encoded information. Here, encoding may be entropy encoding, and decoding may be entropy decoding.

Optional Signaling: Information can be selectively signaled. Selective signaling of information may mean that an encoding device selectively includes information in a bitstream or a recording medium (according to specific conditions). Selective signaling of information may mean that a decoding apparatus selectively extracts information from a bitstream (according to a specific condition).

Omission of signaling: Signaling of information may be omitted. Omission of signaling of information about information may mean that an encoding device does not include information in a bitstream or a recording medium (according to a specific condition). Omission of signaling for information may mean that the decoding apparatus does not extract information from the bitstream (according to a specific condition).

Statistical value: Variables, coding parameters and constants, etc. can have values that can be computed. A statistical value may be a value generated by an operation on the values of these specified objects. For example, the statistical value is an average value, a weighted average value, a weighted sum, a minimum value, a maximum value, and a mode for values such as a specified variable, a specified coding parameter, and a specified constant. It can be one or more of a value, a median value, and an interpolated value.

The encoding device 100 may be an encoder, a video encoding device, or an image encoding device. A video may include one or more images. The encoding apparatus 100 may sequentially encode one or more images of a video.

Referring to FIG. 1, an encoding apparatus 100 includes an inter prediction unit 110, an intra prediction unit 120, a switch 115, a subtractor 125, a transform unit 130, a quantization unit 140, and entropy encoding. It may include a unit 150, an inverse quantization unit 160, an inverse transform unit 170, an adder 175, a filter unit 180, and a reference picture buffer 190.

The encoding apparatus 100 may perform encoding on a target image using an intra mode and/or an inter mode. In other words, the prediction mode for the target block may be one of an intra mode and an inter mode.

Hereinafter, the terms “intra mode”, “intra prediction mode”, “in-picture mode” and “in-picture prediction mode” may be used interchangeably and interchangeably.

Hereinafter, the terms “inter mode”, “inter prediction mode”, “inter-screen mode” and “inter-prediction mode” may be used in the same meaning and may be used interchangeably.

Hereinafter, the term “video” may refer to only a part of an image or may refer to a block. Also, processing of “image” may indicate sequential processing of a plurality of blocks.

Also, the encoding device 100 may generate a bitstream including encoded information through encoding of a target image, and output and store the generated bitstream. The generated bitstream may be stored in a computer readable recording medium and may be streamed through a wired and/or wireless transmission medium.

As the prediction mode, if the intra mode is used, the switch 115 can be switched to intra. As the prediction mode, when the inter mode is used, the switch 115 can be switched to inter.

The encoding apparatus 100 may generate a prediction block for the target block. Also, after the prediction block is generated, the encoding apparatus 100 may encode a residual block for the target block by using the target block and the residual of the prediction block.

When the prediction mode is the intra mode, the intra predictor 120 may use a pixel of a block already encoded and/or decoded, which is adjacent to the target block, as a reference sample. The intra predictor 120 may perform spatial prediction on the target block using the reference sample, and generate prediction samples for the target block through the spatial prediction. A prediction sample may mean a sample within a prediction block.

The inter prediction unit 110 may include a motion estimation unit and a motion compensation unit.

When the prediction mode is the inter mode, the motion prediction unit may search for a region that best matches the target block from the reference image in the motion prediction process, and derive motion vectors for the target block and the searched region using the searched region. can do. At this time, the motion prediction unit may use a search area as a search target area.

The reference picture may be stored in the reference picture buffer 190, and when encoding and/or decoding of the reference picture is processed, the encoded and/or decoded reference picture may be stored in the reference picture buffer 190.

As a decoded picture is stored, the reference picture buffer 190 may be a decoded picture buffer (DPB).

The motion compensator may generate a prediction block for the target block by performing motion compensation using a motion vector. Here, the motion vector may be a 2D vector used for inter prediction. Also, the motion vector may indicate an offset between the target image and the reference image.

The motion estimation unit and the motion compensation unit may generate a prediction block by applying an interpolation filter to a partial region in a reference image when a motion vector has a non-integer value. In order to perform inter prediction or motion compensation, methods of motion prediction and motion compensation of PUs included in the CU based on the CU include skip mode, merge mode, and advanced motion vector prediction (Advanced Motion Vector Prediction (AMVP) mode and current picture reference mode may be determined, and inter prediction or motion compensation may be performed according to each mode.

The subtractor 125 may generate a residual block that is a difference between the target block and the predicted block. A residual block may also be referred to as a residual signal.

The residual signal may mean a difference between the original signal and the prediction signal. Alternatively, the residual signal may be a signal generated by transforming, quantizing, or transforming and quantizing the difference between the original signal and the predicted signal. A residual block may be a residual signal for a block unit.

The transform unit 130 may generate transform coefficients by performing transform on the residual block, and may output the generated transform coefficients. Here, the transform coefficient may be a coefficient value generated by performing transform on the residual block.

The conversion unit 130 may use one of a plurality of predefined conversion methods in performing the conversion.

A plurality of predefined transform methods may include discrete cosine transform (DCT), discrete sine transform (DST), and Karhunen-Loeve transform (KLT) based transform. there is.

A transform method used for transforming the residual block may be determined according to at least one of coding parameters for the target block and/or neighboring blocks. For example, the transform method may be determined based on at least one of an inter prediction mode for a PU, an intra prediction mode for a PU, a TU size, and a TU shape. Alternatively, transformation information indicating a transformation method may be signaled from the encoding apparatus 100 to the decoding apparatus 200.

When a transform skip mode is applied, the transform unit 130 may skip transforming the residual block.

A quantized transform coefficient level or a quantized level may be generated by applying quantization to the transform coefficients. Hereinafter, in the embodiments, a quantized transform coefficient level and a quantized level may also be referred to as a transform coefficient.

The quantization unit 140 may generate a quantized transform coefficient level (ie, a quantized level or a quantized coefficient) by quantizing a transform coefficient according to a quantization parameter. The quantization unit 140 may output the generated quantized transform coefficient level. At this time, the quantization unit 140 may quantize the transform coefficient using a quantization matrix.

The entropy encoding unit 150 may generate a bitstream by performing entropy encoding according to a probability distribution based on values calculated by the quantization unit 140 and/or coding parameter values calculated in the encoding process. . The entropy encoding unit 150 may output the generated bitstream.

The entropy encoding unit 150 may perform entropy encoding on information about pixels of an image and information for decoding an image. For example, information for decoding an image may include a syntax element and the like.

When entropy encoding is applied, a small number of bits may be allocated to a symbol having a high probability of occurrence, and a large number of bits may be allocated to a symbol having a low probability of occurrence. As symbols are represented through such allocation, the size of bitstrings for symbols that are encoding targets can be reduced. Therefore, compression performance of image encoding can be improved through entropy encoding.

In addition, the entropy encoding unit 150 uses exponential golomb, context-adaptive variable length coding (CAVLC), and context-adaptive binary arithmetic coding (Context-Adaptive Binary Coding) for entropy encoding. A coding method such as Arithmetic Coding (CABAC) may be used. For example, the entropy encoding unit 150 may perform entropy encoding using a variable length coding/code (VLC) table. For example, the entropy encoding unit 150 may derive a binarization method for a target symbol. Also, the entropy encoding unit 150 may derive a probability model of the target symbol/bin. The entropy encoding unit 150 may perform arithmetic encoding using the derived binarization method, probability model, and context model.

The entropy encoding unit 150 may change coefficients in the form of a 2-dimensional block into a form of a 1-dimensional vector through a transform coefficient scanning method in order to encode the quantized transform coefficient level.

A coding parameter may be information required for encoding and/or decoding. The coding parameter may include information that is encoded in the encoding device 100 and transmitted from the encoding device 100 to the decoding device, and may include information that can be derived in an encoding or decoding process. For example, as information transmitted to the decoding device, there is a syntax element.

A coding parameter may include information derived from an encoding process or a decoding process, as well as information (or flags and indexes, etc.) encoded in an encoding device and signaled from an encoding device to a decoding device, such as a syntax element. there is. Also, the coding parameter may include information required for encoding or decoding an image. For example, the size of the unit/block, the shape of the unit/block, the depth of the unit/block, division information of the unit/block, division structure of the unit/block, information indicating whether the unit/block is divided in the form of a quad tree, Information indicating whether the unit/block is split into a binary tree, the split direction of the binary tree (horizontal or vertical), the split type of the binary tree (symmetric or asymmetric), the unit/block is a ternary tree information indicating whether it is split into ternary tree type split direction (horizontal direction or vertical direction), ternary tree type split type (symmetric split or asymmetric split, etc.), unit/block multi-type tree information indicating whether or not to split into a multi-type tree, combination and direction (horizontal or vertical, etc.) of multi-type tree split, split type of multi-type tree split (symmetric split or asymmetric split), multi-type Split tree in tree form (binary tree or ternary tree), type of prediction mode (intra prediction or inter prediction), intra prediction mode/direction, intra luma prediction mode/direction, intra chroma prediction mode/direction, intra partitioning information, inter prediction Split information, coding block split flag, prediction block split flag, transform block split flag, reference sample filtering method, reference sample filter tap (tap), reference sample filter coefficient, prediction block filtering method, prediction block filter tap, prediction block filter coefficient , prediction block boundary filtering method, prediction block boundary filter tap, prediction block boundary filter coefficient, inter prediction mode, motion information, motion vector, motion vector difference, reference picture index, inter prediction direction, inter prediction indicator, prediction list utilization ) flag, reference picture list, reference picture, POC, motion vector predictor, motion vector prediction index, motion vector prediction candidate, motion vector candidate list, information indicating whether merge mode is used, merge index, merge candidate, merge candidate list , information indicating whether skip mode is used, interpolation filter type, interpolation filter filter tap, interpolation filter filter coefficient, motion vector magnitude, motion vector representation accuracy, transform type, transform size, and primary transform. Information indicating whether or not to use, information indicating whether to use additional (secondary) transform, primary transform selection information (or primary transform index), secondary transform selection information (or secondary transform index), residual Information indicating the presence or absence of a signal, a coded block pattern, a coded block flag, a quantization parameter, a residual quantization parameter, a quantization matrix, information about an intra-loop filter, and an intra-loop filter Information indicating whether to apply, coefficients of intra-loop filter, filter tap of intra-loop, shape/form of intra-loop filter, information indicating whether to apply deblocking filter, information indicating whether to apply deblocking filter, coefficient, filter tap of deblocking filter, strength of deblocking filter, shape/shape of deblocking filter, information indicating whether to apply adaptive sample offset, adaptive sample offset value, adaptive sample offset category, adaptive sample Offset type, information indicating whether the adaptive in-loop filter is applied, coefficients of the adaptive in-loop filter, filter tap of the adaptive in-loop filter, shape/shape of the adaptive in-loop filter Shape, binarization/inverse binarization method, context model, context model determination method, context model update method, information indicating whether regular mode is performed, information indicating whether bypass mode is performed, significant coefficient flags, last Significant coefficient flag, coefficient group unit coding flag, last significant coefficient position, flag indicating whether coefficient value is greater than 1, flag indicating whether coefficient value is greater than 2, flag indicating whether coefficient value is greater than 3 , remaining coefficient value information, sign information, reconstructed luma sample, reconstructed chroma sample, context bin, bypass bin, residual luma sample, residual chroma sample, transform coefficient, luma transform coefficient, chroma transform coefficient, Quantized level, luma quantized level, chroma quantized level, transform coefficient level, luma transform coefficient level, chroma transform coefficient level, transform coefficient level scanning method, size of motion vector search region from the side of a decoding apparatus, decoding apparatus The shape of the motion vector search area on the side, the number of motion vector searches on the side of the decoding device, CTU size, minimum block size, maximum block size, maximum block depth, minimum block depth, image display/output order, slice identification information , slice type, slice division information, tile group identification information, tile group type, tile group division information, tile identification information, tile type, tile division information, picture type, bit depth, input sample bit depth, reconstructed sample bit depth , residual sample bit depth, transform coefficient bit depth, quantized level bit depth, luma signal information, chroma signal information, a color space of a target block, and a value of at least one of a color space of a residual block, A combined form or statistic may be included in the coding parameters. In addition, information related to the aforementioned coding parameters may also be included in the coding parameters. Information used to calculate and/or derive the aforementioned coding parameters may also be included in the coding parameters. Information calculated or derived using the aforementioned coding parameters may also be included in the coding parameters.

The primary transform selection information may indicate a primary transform applied to the target block.

The secondary transform selection information may indicate a secondary transform applied to the target block.

The residual signal may represent a difference between the original signal and the predicted signal. Alternatively, the residual signal may be a signal generated by transforming a difference between the original signal and the predicted signal. Alternatively, the residual signal may be a signal generated by transforming and quantizing a difference between the original signal and the prediction signal. A residual block may be a residual signal for a block.

Here, signaling information may mean that the encoding device 100 includes entropy-encoded information generated by performing entropy encoding on a flag or index in a bitstream, , in the decoding apparatus 200, may mean obtaining information by performing entropy decoding on entropy-encoded information extracted from a bitstream. Here, the information may include flags and indexes.

A signal may refer to signaled information. Hereinafter, information on images and blocks may be referred to as signals. Also, in the following, the terms “information” and “signal” may be used interchangeably and may be used interchangeably. For example, the specific signal may be a signal representing a specific block. An original signal may be a signal representing a target block. A prediction signal may be a signal representing a prediction block. A residual signal may be a signal representing a residual block.

A bitstream may include information according to a specified syntax. The encoding device 100 may generate a bitstream including information according to a specified syntax. The encoding device 200 may obtain information from a bitstream according to a specified syntax.

Since encoding through inter prediction is performed by the encoding apparatus 100, the encoded target image may be used as a reference image for other image(s) to be processed later. Accordingly, the encoding apparatus 100 may reconstruct or decode the encoded target image and store the reconstructed or decoded image in the reference picture buffer 190 as a reference image. Inverse quantization and inverse transformation may be performed on the encoded target image for decoding.

The quantized level may be inversely quantized in the inverse quantization unit 160 and inversely transformed in the inverse transformation unit 170 . The inverse quantization unit 160 may generate inverse quantized coefficients by performing inverse quantization on the quantized level. The inverse transform unit 170 may generate inverse quantized and inverse transformed coefficients by performing an inverse transform on the inverse quantized coefficients.

The inverse quantized and inverse transformed coefficients may be combined with the prediction block through the adder 175. A reconstructed block may be generated by adding the inverse quantized and inverse transformed coefficients and the prediction block. Here, the inverse quantized and/or inverse transformed coefficient may mean a coefficient on which at least one of dequantization and inverse-transformation has been performed, and may mean a reconstructed residual block. Here, the reconstructed block may mean a recovered block or a decoded block.

The reconstructed block may pass through the filter unit 180 . The filter unit 180 may include at least one of a deblocking filter, a sample adaptive offset (SAO), an adaptive loop filter (ALF), and a non-local filter (NLF). One or more may be applied to a reconstructed sample, reconstructed block or reconstructed picture. The filter unit 180 may also be referred to as an in-loop filter.

The deblocking filter may remove block distortion generated at a boundary between blocks in a reconstructed picture. In order to determine whether to apply the deblocking filter, it may be determined whether to apply the deblocking filter to the target block based on pixel(s) included in several columns or rows included in the block.

When a deblocking filter is applied to a target block, the applied filter may vary according to the required strength of deblocking filtering. In other words, among different filters, a filter determined according to the strength of deblocking filtering may be applied to the target block. When a deblocking filter is applied to the target block, a long-tap filter, a strong filter, a weak filter, and a Gaussian filter are selected according to the strength of the deblocking filtering required. ) may be applied to the target block.

Also, when vertical filtering and horizontal filtering are performed on a target block, horizontal filtering and vertical filtering may be processed in parallel.

SAO may add an appropriate offset to a pixel value of a pixel to compensate for a coding error. The SAO may perform correction using an offset on a difference between an original image and an image to which deblocking is applied in units of pixels for an image to which deblocking is applied. In order to perform offset correction on an image, a method of dividing pixels included in the image into a certain number of areas, determining an area to be offset from among the divided areas, and applying the offset to the determined area will be used. Alternatively, a method of applying an offset in consideration of edge information of each pixel of the image may be used.

ALF may perform filtering based on a value obtained by comparing the reconstructed image and the original image. After dividing the pixels included in the image into predetermined groups, a filter to be applied to each divided group may be determined, and filtering may be performed differentially for each group. Information related to whether to apply the adaptive loop filter may be signaled for each CU. This information may be signaled for the luma signal. The shape and filter coefficients of ALF to be applied to each block may be different for each block. Alternatively, a fixed type of ALF may be applied to the block regardless of the characteristics of the block.

The non-local filter may perform filtering based on reconstructed blocks similar to the target block. A region similar to the target block may be selected from the reconstructed image, and filtering of the target block may be performed using statistical properties of the selected similar region. Information related to whether to apply the non-local filter may be signaled to the CU. Also, shapes and filter coefficients of non-local filters to be applied to blocks may be different according to blocks.

A reconstructed block or a reconstructed image that has passed through the filter unit 180 may be stored in the reference picture buffer 190 as a reference picture. A reconstructed block that has passed through the filter unit 180 may be part of a reference picture. In other words, the reference picture may be a reconstructed picture composed of reconstructed blocks that have passed through the filter unit 180. The stored reference picture can then be used for inter prediction or motion compensation.

The decoding device 200 may be a decoder, a video decoding device, or an image decoding device.

Referring to FIG. 2 , the decoding apparatus 200 includes an entropy decoding unit 210, an inverse quantization unit 220, an inverse transform unit 230, an intra prediction unit 240, an inter prediction unit 250, and a switch 245. , an adder 255, a filter unit 260, and a reference picture buffer 270.

The decoding device 200 may receive the bitstream output from the encoding device 100. The decoding apparatus 200 may receive a bitstream stored in a computer readable recording medium or may receive a bitstream streamed through a wired/wireless transmission medium.

The decoding apparatus 200 may perform intra mode and/or inter mode decoding on a bitstream. Also, the decoding apparatus 200 may generate a reconstructed image or a decoded image through decoding, and output the generated reconstructed image or decoded image.

For example, conversion to an intra mode or an inter mode according to a prediction mode used for decoding may be performed by the switch 245 . When the prediction mode used for decoding is an intra mode, the switch 245 may be switched to intra mode. When the prediction mode used for decoding is the inter mode, the switch 245 may be switched to inter mode.

The decoding apparatus 200 may obtain a reconstructed residual block by decoding the input bitstream and generate a prediction block. When the reconstructed residual block and the prediction block are obtained, the decoding apparatus 200 may generate a reconstructed block to be decoded by summing the reconstructed residual block and the prediction block.

The entropy decoding unit 210 may generate symbols by performing entropy decoding on the bitstream based on the probability distribution of the bitstream. The generated symbols may include symbols in the form of quantized transform coefficient levels (ie, quantized levels or quantized coefficients). Here, the entropy decoding method may be similar to the above-described entropy encoding method. For example, the entropy decoding method may be a reverse process of the above-described entropy encoding method.

The entropy decoding unit 210 may change a coefficient in the form of a 1-dimensional vector into a form of a 2-dimensional block through a transform coefficient scanning method in order to decode the quantized transform coefficient level.

For example, the coefficients may be changed into a 2D block form by scanning the coefficients of a block using an upper-right diagonal scan. Alternatively, which scan to be used among the upper-right diagonal scan, vertical scan, and horizontal scan may be determined according to the size of the block and/or the intra prediction mode.

The quantized coefficient may be inversely quantized in the inverse quantization unit 220 . The inverse quantization unit 220 may generate inverse quantized coefficients by performing inverse quantization on the quantized coefficients. In addition, the inverse quantized coefficient may be inversely transformed in the inverse transform unit 230 . The inverse transform unit 230 may generate a reconstructed residual block by performing an inverse transform on the inverse quantized coefficient. As a result of performing inverse quantization and inverse transformation on the quantized coefficients, a reconstructed residual block may be generated. In this case, the inverse quantization unit 220 may apply a quantization matrix to the quantized coefficients in generating the reconstructed residual block.

When the intra mode is used, the intra predictor 240 may generate a prediction block by performing spatial prediction on the target block using pixel values of previously decoded blocks adjacent to the target block.

The inter prediction unit 250 may include a motion compensation unit. Alternatively, the inter prediction unit 250 may be referred to as a motion compensation unit.

When the inter mode is used, the motion compensator may generate a prediction block by performing motion compensation on the target block using a motion vector and a reference image stored in the reference picture buffer 270 .

When the motion vector has a non-integer value, the motion compensator may apply an interpolation filter to a partial region in the reference image and generate a prediction block using the reference image to which the interpolation filter is applied. The motion compensation unit may determine which mode among skip mode, merge mode, AMVP mode, and current picture reference mode is a motion compensation method used for a PU included in a CU based on the CU to perform motion compensation, and the determined mode Accordingly, motion compensation may be performed.

The reconstructed residual block and the prediction block may be added through an adder 255. Adder 255 may produce a reconstructed block by adding the reconstructed residual block and the prediction block.

The reconstructed block may pass through the filter unit 260 . The filter unit 260 may apply at least one of a deblocking filter, an SAO filter, an ALF filter, and a non-local filter to a reconstructed block or a reconstructed image. A reconstructed image may be a picture including a reconstructed block.

The filter unit 260 may output a reconstructed image.

The reconstructed block and/or the reconstructed image that has passed through the filter unit 260 may be stored as a reference picture in the reference picture buffer 270 . A reconstructed block that has passed through the filter unit 260 may be part of a reference picture. In other words, the reference picture may be a reconstructed image composed of reconstructed blocks that have passed through the filter unit 260 . The stored reference picture can then be used for inter prediction and/or motion compensation.

3 may schematically show an example in which one unit is divided into a plurality of sub-units.

In order to efficiently segment an image, a coding unit (CU) may be used in encoding and decoding. A unit may be a term that collectively refers to 1) a block including image samples and 2) a syntax element. For example, “division of a unit” may mean “division of a block corresponding to a unit”.

A CU may be used as a base unit for image encoding and/or decoding. In addition, the CU may be used as a unit to which one selected mode of intra mode and inter mode is applied in video encoding and/or decoding. In other words, in video encoding and/or decoding, it may be determined which mode among the intra mode and the inter mode is to be applied to each CU.

Also, a CU may be a base unit in encoding and/or decoding of prediction, transform, quantization, inverse transform, inverse quantization, and transform coefficients.

Referring to FIG. 3 , an image 300 may be sequentially divided into units of largest coding units (LCUs). For each LCU, a partition structure may be determined. Here, LCU may be used as the same meaning as Coding Tree Unit (CTU).

Division of a unit may mean division of a block corresponding to the unit. The block division information may include depth information about the depth of a unit. Depth information may indicate the number and/or degree of division of a unit. One unit may be hierarchically divided into a plurality of sub-units with depth information based on a tree structure.

Each divided sub-unit may have depth information. Depth information may be information indicating the size of a CU. Depth information may be stored for each CU.

Each CU may have depth information. When a CU is split, CUs generated by splitting may have a depth increased by 1 from the depth of the split CU.

The division structure may refer to a distribution of CUs in the LCU 310 to efficiently encode an image. This distribution may be determined according to whether one CU is to be divided into a plurality of CUs. The number of divided CUs may be a positive integer greater than or equal to 2, including 2, 4, 8 and 16, and the like.

The horizontal and vertical sizes of the CU generated by division may be smaller than the horizontal and vertical sizes of the CU before division, depending on the number of CUs generated by division. For example, the horizontal size and vertical size of the CU generated by division may be half of the horizontal size and half of the vertical size of the CU before division.

A divided CU may be recursively divided into a plurality of CUs in the same way. By recursive division, at least one of the horizontal size and the vertical size of the divided CU may be reduced compared to at least one of the horizontal and vertical sizes of the CU before division.

The division of the CU may be made recursively to a predefined depth or to a predefined size.

For example, the depth of CU may have a value of 0 to 3. The size of the CU may range from 64x64 to 8x8 depending on the depth of the CU.

For example, the depth of the LCU 310 may be 0, and the depth of the smallest coding unit (SCU) may be a predefined maximum depth. Here, the LCU may be a CU having the largest coding unit size as described above, and the SCU may be a CU having the smallest coding unit size.

The division may start from the LCU 310, and the depth of the CU may increase by 1 whenever the horizontal size and/or the vertical size of the CU are reduced by the division.

For example, for each depth, a CU that is not split may have a size of 2Nx2N. In addition, in the case of a CU to be divided, a CU of 2Nx2N size may be divided into 4 CUs having a size of NxN. The size of N can be halved every time the depth increases by 1.

Referring to FIG. 3 , an LCU having a depth of 0 may be 64x64 pixels or a 64x64 block. 0 may be the minimum depth. A SCU with a depth of 3 can be 8x8 pixels or an 8x8 block. 3 may be the maximum depth. In this case, the CU of the 64x64 block, which is the LCU, may be expressed as a depth of 0. A CU of a 32x32 block can be represented with a depth of 1. A CU of a 16x16 block can be represented with a depth of 2. A CU of an 8x8 block, which is an SCU, can be expressed as a depth of 3.

Information on whether the CU is split may be expressed through split information of the CU. The division information may be 1 bit of information. All CUs except for the SCU may include partition information. For example, a value of partition information of a CU that is not split may be a first value, and a value of partition information of a CU that is split may be a second value. When the split information indicates whether the CU splits, the first value may be 0 and the second value may be 1.

For example, when one CU is divided into 4 CUs, the horizontal and vertical sizes of each CU of the 4 CUs generated by the division are half of the horizontal size and half of the vertical size of the CU before the division, respectively. can When a 32x32 CU is divided into 4 CUs, the sizes of the 4 CUs may be 16x16. When one CU is divided into 4 CUs, it can be said that the CU is divided in a quad-tree form. In other words, it can be seen that quad-tree partitioning is applied to the CU.

For example, when one CU is split into two CUs, the horizontal size or vertical size of each CU of the two CUs generated by the split is half of the horizontal size or half of the vertical size of the CU before the split, respectively. can When a CU of 32x32 size is vertically divided into two CUs, the sizes of the two divided CUs may be 16x32. When a CU of 32x32 size is horizontally divided into two CUs, the sizes of the two divided CUs may be 32x16. When one CU is divided into two CUs, it can be said that the CU is divided in a binary-tree form. In other words, it can be seen that binary-tree partitioning has been applied to the CU.

For example, when one CU is divided into three CUs, the three divided CUs may be generated by dividing the horizontal or vertical size of the CU before being divided at a ratio of 1:2:1. For example, when a CU of 16x32 size is divided into 3 CUs in the horizontal direction, the 3 divided CUs may have sizes of 16x8, 16x16 and 16x8, respectively, from the top. For example, when a CU of size 32x32 is divided into three CUs in the vertical direction, the three divided CUs may have sizes of 8x32, 16x32, and 8x32 from the left, respectively. When one CU is divided into three CUs, it can be said that the CU is divided in a ternary-tree form. In other words, it can be seen that ternary-tree partitioning is applied to the CU.

Both quad-tree partitioning and binary-tree partitioning are applied to the LCU 310 of FIG. 3 .

In the encoding apparatus 100, a 64x64 Coding Tree Unit (CTU) may be divided into a plurality of smaller CUs by a recursive quad-tree structure. One CU can be divided into 4 CUs with equal sizes. CUs can be partitioned recursively, and each CU can have a structure of a quad tree.

Through recursive segmentation on the CU, an optimal segmentation method that results in the smallest rate-distortion ratio can be selected.

The CTU 320 of FIG. 3 is an example of a CTU to which quad tree partitioning, binary tree partitioning, and ternary tree partitioning are all applied.

As described above, in order to divide the CTU, at least one of quad tree partitioning, binary tree partitioning, and ternary tree partitioning may be applied to the CTU. Partitions can be applied based on a specified priority order.

For example, quad tree partitioning may be preferentially applied to CTUs. A CU that cannot be further divided into a quad tree may correspond to a leaf node of a quad tree. A CU corresponding to a leaf node of a quad tree may be a root node of a binary tree and/or a ternary tree. That is, a CU corresponding to a leaf node of a quad tree may be split in the form of a binary tree or a ternary tree, or may not be split any further. At this time, quad tree splitting is not applied again to a CU generated by applying binary tree splitting or ternary tree splitting to a CU corresponding to a leaf node of a quad tree, so that block splitting and/or signaling of block splitting information is performed. can be done effectively.

Splitting of a CU corresponding to each node of the quad tree may be signaled using quad splitting information. Quad splitting information having a first value (eg, “1”) may indicate that the CU is split in the form of a quad tree. Quad splitting information having a second value (eg, “0”) may indicate that the CU is not split in the form of a quad tree. The quad division information may be a flag having a specified length (eg, 1 bit).

There may be no priority between binary tree splitting and ternary tree splitting. That is, a CU corresponding to a leaf node of a quad tree may be partitioned in a binary tree form or a ternary tree form. In addition, a CU generated by binary tree splitting or ternary tree splitting may be split again into a binary tree shape or a ternary tree shape, or may not be split any further.

Partitioning in the case where there is no priority between binary tree partitioning and ternary tree partitioning may be referred to as multi-type tree partitioning. That is, a CU corresponding to a leaf node of a quad tree may become a root node of a multi-type tree. Splitting of a CU corresponding to each node of the multi-type tree may be signaled using at least one of information indicating whether or not the multi-type tree is split, splitting direction information, and splitting tree information. For splitting of a CU corresponding to each node of the multi-type tree, information indicating whether to split or not, splitting direction information, and splitting tree information may be signaled sequentially.

For example, information indicating whether a multi-type tree having a first value (eg, “1”) is split may indicate that the corresponding CU is split into a multi-type tree. Information indicating whether the multi-type tree having a second value (eg, “0”) is split may indicate that the corresponding CU is not split into a multi-type tree.

When a CU corresponding to each node of a multi-type tree is split into a multi-type tree, the corresponding CU may further include splitting direction information.

The splitting direction information may indicate a splitting direction of multi-type tree splitting. The division direction information having a first value (eg, “1”) may indicate that the corresponding CU is divided in the vertical direction. The division direction information having a second value (eg, “0”) may indicate that the corresponding CU is divided in the horizontal direction.

When a CU corresponding to each node of a multi-type tree is split into a multi-type tree, the corresponding CU may further include split tree information. Split tree information may indicate a tree used for multi-type tree split.

For example, split tree information having a first value (eg, “1”) may indicate that the corresponding CU is split in the form of a binary tree. Split tree information having a second value (eg, “0”) may indicate that the corresponding CU is split in the form of a ternary tree.

Here, each of the aforementioned information indicating whether to split or not, split tree information, and split direction information may be a flag having a specified length (eg, 1 bit).

At least one of the above-described quad splitting information, information indicating whether to split a multi-type tree, splitting direction information, and splitting tree information may be entropy-encoded and/or entropy-decoded. For entropy encoding/decoding of such information, information of neighboring CUs adjacent to the target CU may be used.

For example, it may be considered that there is a high probability that the division shape of the left CU and/or the upper CU (that is, whether or not to split, the split tree, and/or the split direction) and the split shape of the target CU are similar to each other. Accordingly, context information for entropy encoding and/or entropy decoding of information of a target CU may be derived based on information of a neighboring CU. In this case, the information of the neighboring CU may include at least one of 1) quad split information, 2) information indicating whether the multi-type tree is split, 3) split direction information, and 4) split tree information of the neighboring CU.

As another embodiment, binary tree splitting may be preferentially performed among binary tree splitting and ternary tree splitting. That is, binary tree splitting is applied first, and a CU corresponding to a leaf node of the binary tree may be set as a root node of the ternary tree. In this case, quad tree splitting and binary tree splitting may not be performed on a CU corresponding to a node of a ternary tree.

A CU that is not further split by quad tree splitting, binary tree splitting, and/or ternary tree splitting may become a unit of encoding, prediction, and/or transformation. That is, for prediction and/or transformation, the CU may not be further split. Accordingly, a partitioning structure and partitioning information for dividing a CU into prediction units and/or transform units may not exist in the bitstream.

However, when the size of a CU serving as a unit of division is larger than the size of the maximum transform block, this CU may be recursively split until the size of the CU is less than or equal to the size of the maximum transform block. For example, when the size of a CU is 64x64 and the size of a maximum transform block is 32x32, the CU may be divided into four 32x32 blocks for transform. For example, when the size of a CU is 32x64 and the size of a maximum transform block is 32x32, the CU may be divided into two 32x32 blocks for transform.

In this case, information on whether the CU is split for conversion may not be separately signaled. Without signaling, whether to split the CU may be determined by comparing the horizontal size (and/or vertical size) of the CU and the horizontal size (and/or vertical size) of the largest transform block. For example, when the horizontal size of the CU is greater than the horizontal size of the largest transform block, the CU may be vertically divided into two parts. In addition, when the vertical size of the CU is greater than the vertical size of the largest transform block, the CU may be divided into two horizontally.

Information on the maximum size and/or minimum size of a CU and information on the maximum size and/or minimum size of a transform block may be signaled or determined at a higher level for a CU. For example, the upper level may be a sequence level, a picture level, a tile level, a tile group level, and a slice level. For example, the minimum size of a CU may be determined to be 4x4. For example, the maximum size of a transform block may be determined to be 64x64. For example, the minimum size of a transform block may be determined to be 4x4.

Information about the minimum size of a CU corresponding to a leaf node of a quad tree (ie quad tree minimum size) and/or the maximum depth of a path from the root node of a multi-type tree to a leaf node (ie multi-type tree maximum size). depth) may be signaled or determined at a higher level for the CU. For example, the higher level may be a sequence level, a picture level, a slice level, a tile group level, and a tile level. Information on the minimum size of the quad tree and/or information on the maximum depth of the multi-type tree may be separately signaled or determined for each intra-slice and inter-slice.

Difference information about the size of the CTU and the maximum size of the transform block may be signaled or determined at a higher level for the CU. For example, the higher level may be a sequence level, a picture level, a slice level, a tile group level, and a tile level. Information about the maximum size of the CU corresponding to each node of the binary tree (ie, the maximum size of the binary tree) may be determined based on the size and difference information of the CTU. The maximum size of the CU corresponding to each node of the ternary tree (ie, the maximum size of the ternary tree) may have a different value depending on the slice type. For example, within an intra slice, the maximum size of the ternary tree may be 32x32. Also, for example, within an inter slice, the maximum size of a ternary tree may be 128x128. For example, the minimum size of a CU corresponding to each node of a binary tree (say, the minimum size of a binary tree) and/or the minimum size of a CU corresponding to each node of a ternary tree (say, the minimum size of a ternary tree) is Can be set to a minimum size.

As another example, the maximum size of the binary tree and/or the maximum size of the ternary tree may be signaled or determined at the slice level. Also, the minimum size of the binary tree and/or the minimum size of the ternary tree may be signaled or determined at the slice level.

Based on the aforementioned various block sizes and various depths, quad splitting information, information indicating whether to split a multi-type tree, splitting tree information, and/or splitting direction information may or may not exist in the bitstream.

For example, if the size of the CU is not larger than the quad tree minimum size, the CU may not include quad splitting information, and the quad splitting information for the CU may be inferred as a second value.

For example, if the size (horizontal size and vertical size) of a CU corresponding to a node of a multi-type tree is larger than the binary tree maximum size (horizontal size and vertical size) and/or the ternary tree maximum size (horizontal size and vertical size) If larger, the CU may not be partitioned into binary tree form and/or ternary tree form. According to this determination method, information indicating whether to split the multi-type tree may not be signaled and may be inferred as a second value.

Alternatively, the size (horizontal size and vertical size) of a CU corresponding to a node of a multi-type tree is equal to the minimum size (horizontal size and vertical size) of a binary tree, or the size of a CU (horizontal size and vertical size) is equal to the size of a ternary tree. When equal to twice the minimum size (horizontal size and vertical size), the CU may not be split into a binary tree shape and/or a ternary tree shape. According to this determination method, information indicating whether to split the multi-type tree may not be signaled and may be inferred as a second value. This is because when the CU is divided into a binary tree shape and/or a ternary tree shape, a CU smaller than the minimum size of the binary tree and/or the minimum size of the ternary tree is generated.

Alternatively, binary tree splitting or ternary tree splitting may be limited based on the size of the virtual pipeline data unit (ie, pipeline buffer size). For example, binary tree splitting or ternary tree splitting may be limited when a CU is split into sub-CUs not suitable for a pipeline buffer size by binary tree splitting or ternary tree splitting. The pipeline buffer size may be equal to the size of the largest transform block (eg, 64X64).

For example, when the pipeline buffer size is 64X64, the following partitions can be limited.

- ternary tree split over NxM (N and/or M is 128) CUs

- horizontal binary tree partitioning over 128xN (N <= 64) CUs

- Vertical binary tree partitioning for Nx128 (N <= 64) CUs

Alternatively, when the depth in the multi-type tree of a CU corresponding to a node of the multi-type tree is equal to the maximum depth of the multi-type tree, the CU may not be split into a binary tree form and/or a ternary tree form. According to this determination method, information indicating whether to split the multi-type tree may not be signaled and may be inferred as a second value.

Alternatively, only when at least one of vertical binary tree splitting, horizontal binary tree splitting, vertical ternary tree splitting, and horizontal ternary tree splitting is possible for a CU corresponding to a node of a multi-type tree, a multi-type tree Information indicating whether to divide may be signaled. Otherwise, the CU may not be partitioned in binary tree form and/or ternary tree form. According to this determination method, information indicating whether to split the multi-type tree may not be signaled and may be inferred as a second value.

Alternatively, only when both vertical binary tree splitting and horizontal binary tree splitting are possible for a CU corresponding to a node of a multi-type tree, or both vertical ternary tree splitting and horizontal ternary tree splitting are possible, splitting direction information may be signaled. Otherwise, the division direction information may not be signaled and may be inferred as a value indicating a direction in which the CU may be divided.

Alternatively, only when both vertical binary tree splitting and vertical ternary tree splitting are possible for a CU corresponding to a node of a multi-type tree, or both horizontal binary tree splitting and horizontal ternary tree splitting are possible, split tree information may be signaled. Otherwise, the split tree information may not be signaled and may be inferred as a value indicating a tree applicable to split of a CU.

Among the CUs divided from the LCU, a CU that is not further divided may be divided into one or more prediction units (PUs).

A PU may be a basic unit for prediction. A PU may be coded and decoded in any one of skip mode, inter mode, and intra mode. A PU may be divided into various types according to each mode. For example, the target block described above with reference to FIG. 1 and the target block described above with reference to FIG. 2 may be PUs.

A CU may not be divided into PUs. When a CU is not divided into PUs, the size of a CU and the size of a PU may be the same.

In skip mode, there may not be a split within a CU. In the skip mode, a 2Nx2N mode 410 in which sizes of PU and CU are the same may be supported without division.

In inter mode, 8 divided types can be supported within the CU. For example, in inter mode, 2Nx2N mode 410, 2NxN mode 415, Nx2N mode 420, NxN mode 425, 2NxnU mode 430, 2NxnD mode 435, nLx2N mode 440 and nRx2N Mode 445 may be supported.

In intra mode, 2Nx2N mode 410 and NxN mode 425 may be supported.

In the 2Nx2N mode 410, a PU having a size of 2Nx2N may be encoded. A PU having a size of 2Nx2N may mean a PU having the same size as that of a CU. For example, a 2Nx2N PU may have a size of 64x64, 32x32, 16x16 or 8x8.

In the NxN mode 425, a PU having a size of NxN may be encoded.

For example, in intra prediction, when the size of a PU is 8x8, 4 divided PUs can be coded. The size of the divided PU may be 4x4.

When a PU is coded using an intra mode, the PU may be coded using one intra prediction mode among a plurality of intra prediction modes. For example, High Efficiency Video Coding (HEVC) technology may provide 35 intra prediction modes, and a PU may be coded in one intra prediction mode among the 35 intra prediction modes.

Which mode among the 2Nx2N mode 410 and the NxN mode 425 the PU will be encoded may be determined by a rate-distortion cost.

The encoding apparatus 100 may perform an encoding operation on a PU having a size of 2Nx2N. Here, the encoding operation may be encoding a PU in each of a plurality of intra prediction modes usable by the encoding apparatus 100 . An optimal intra prediction mode for a PU having a size of 2Nx2N may be derived through an encoding operation. An optimal intra prediction mode may be an intra prediction mode that generates a minimum rate-distortion cost for encoding a 2Nx2N PU among a plurality of intra prediction modes usable by the encoding apparatus 100 .

Also, the encoding apparatus 100 may sequentially perform an encoding operation on each PU of NxN-divided PUs. Here, the encoding operation may be encoding a PU in each of a plurality of intra prediction modes usable by the encoding apparatus 100 . An optimal intra prediction mode for an NxN-sized PU may be derived through an encoding operation. An optimal intra-prediction mode may be an intra-prediction mode that generates the minimum rate-distortion cost for encoding an NxN-sized PU among a plurality of intra-prediction modes usable by the encoding apparatus 100.

The encoding apparatus 100 may determine which of the 2Nx2N-sized PU and the NxN-sized PUs to encode based on the comparison of the rate-distortion cost of the 2Nx2N-sized PU and the rate-distortion costs of the NxN-sized PUs.

One CU may be divided into one or more PUs, and a PU may also be divided into a plurality of PUs.

For example, when one PU is divided into 4 PUs, the horizontal size and vertical size of each of the 4 PUs generated by the division are half of the horizontal size and half of the vertical size of the PU before the division, respectively. can When a 32x32 PU is divided into 4 PUs, the sizes of the 4 PUs may be 16x16. When one PU is divided into 4 PUs, it can be said that the PU is divided in a quad-tree shape.

For example, when one PU is split into two PUs, the horizontal size or vertical size of each PU of the two PUs generated by the split is half of the horizontal size or half of the vertical size of the PU before splitting, respectively. can When a 32x32 PU is vertically divided into 2 PUs, the sizes of the 2 PUs may be 16x32. When a 32x32-sized PU is horizontally divided into two PUs, the sizes of the two divided PUs may be 32x16. When one PU is divided into two PUs, it can be said that the PU is divided in the form of a binary-tree.

A transform unit (TU) may be a basic unit used for transformation, quantization, inverse transformation, inverse quantization, entropy encoding, and entropy decoding processes within the CU.

A TU may have a square shape or a rectangular shape. The shape of the TU may be determined depending on the size and/or shape of the CU.

Among the CUs split from the LCU, a CU that is no longer split into CUs may be split into one or more TUs. In this case, the division structure of the TU may be a quad-tree structure. For example, as shown in FIG. 5 , one CU 510 may be divided one or more times according to a quad-tree structure. Through division, one CU 510 may be composed of TUs of various sizes.

When one CU is split two or more times, the CU can be considered to be split recursively. Through splitting, one CU may be composed of TUs having various sizes.

Alternatively, one CU may be divided into one or more TUs based on the number of vertical and/or horizontal lines dividing the CU.

A CU may be divided into symmetric TUs or asymmetric TUs. For division into asymmetric TUs, information on the size and/or shape of a TU may be signaled from the encoding device 100 to the decoding device 200. Alternatively, the size and/or shape of the TU may be derived from information on the size and/or shape of the CU.

A CU may not be divided into TUs. When a CU is not divided into TUs, the size of a CU and the size of a TU may be the same.

One CU may be divided into one or more TUs, and a TU may also be divided into a plurality of TUs.

For example, when one TU is divided into 4 TUs, the horizontal and vertical sizes of each TU of the 4 TUs generated by the division are half of the horizontal size and half of the vertical size of the TU before the division, respectively. can When a TU of 32x32 size is divided into 4 TUs, the sizes of the 4 divided TUs may be 16x16. When one TU is divided into 4 TUs, it can be said that the TU is divided in the form of a quad-tree.

For example, when one TU is split into two TUs, the horizontal size or vertical size of each TU of the two TUs generated by the split is half of the horizontal size or half of the vertical size of the TU before the split, respectively. can When a 32x32 size TU is vertically split into two TUs, the sizes of the two split TUs may be 16x32. When a TU of size 32x32 is horizontally divided into two TUs, the sizes of the two divided TUs may be 32x16. When one TU is split into two TUs, it can be said that the TU is split in a binary-tree format.

CUs may be divided in other ways than shown in FIG. 5 .

For example, one CU may be split into 3 CUs. The horizontal size or vertical size of the three divided CUs may be 1/4, 1/2, and 1/4 of the horizontal or vertical size of the CU before division, respectively.

For example, when a CU of 32x32 size is vertically divided into 3 CUs, the sizes of the 3 divided CUs may be 8x32, 16x32 and 8x32, respectively. In this way, when one CU is split into three CUs, it can be considered that the CU is split in the form of a ternary tree.

One of the exemplified quad tree-type partitioning, binary tree-type partitioning, and ternary tree-type partitioning may be applied for partitioning of the CU, and a plurality of partitioning schemes may be combined together and used for partitioning of the CU. . In this case, a case in which a plurality of partitioning methods are combined and used may be referred to as a composite tree type of partitioning.

6 shows division of a block according to an example.

In the process of encoding and/or decoding an image, a target block may be divided as shown in FIG. 6 . For example, the target block may be a CU.

For division of the target block, an indicator indicating division information may be signaled from the encoding apparatus 100 to the decoding apparatus 200. The division information may be information indicating how the target block is divided.

Splitting information includes a split flag (hereinafter referred to as “split_flag”), a quad-binary flag (hereinafter referred to as “QB_flag”), a quad-tree flag (hereinafter referred to as “quadtree_flag”), and a binary tree flag (hereinafter referred to as “binarytree_flag”). ") and a binary type flag (hereinafter referred to as "Btype_flag").

split_flag may be a flag indicating whether a block is split. For example, a value of 1 of split_flag may indicate that a block is split. A value of 0 of split_flag may indicate that the block is not split.

QB_flag may be a flag indicating in which form a block is divided into a quad tree form and a binary tree form. For example, a value of 0 of QB_flag may indicate that a block is divided in a quad tree form. A value of 1 of QB_flag may indicate that a block is divided in the form of a binary tree. Alternatively, a value of 0 of QB_flag may indicate that the block is divided in the form of a binary tree. A value of 1 for QB_flag may indicate that a block is divided in a quad tree form.

quadtree_flag may be a flag indicating whether a block is split into a quad tree. For example, a value of 1 of quadtree_flag may indicate that a block is split into a quad tree. A value of 0 of quadtree_flag may indicate that a block is not split in the form of a quad tree.

binarytree_flag may be a flag indicating whether a block is split in the form of a binary tree. For example, a value of 1 in binarytree_flag may indicate that a block is split into a binary tree. A value of 0 in binarytree_flag may indicate that a block is not split in the form of a binary tree.

Btype_flag may be a flag indicating whether the block is split vertically or horizontally when the block is split in the form of a binary tree. For example, a value of 0 in Btype_flag may indicate that a block is divided in a horizontal direction. A value of 1 of Btype_flag may indicate that a block is divided in a vertical direction. Alternatively, a value of 0 of Btype_flag may indicate that the block is divided in the vertical direction. A value of 1 of Btype_flag may indicate that the block is divided in the horizontal direction.

For example, partition information for the block of FIG. 6 can be derived by signaling at least one of quadtree_flag, binarytree_flag, and Btype_flag as shown in Table 1 below.

[Table 1]

For example, the split information for the block of FIG. 6 can be derived by signaling at least one of split_flag, QB_flag, and Btype_flag as shown in Table 2 below.

[Table 2]

The partitioning method may be limited to a quad tree or only a binary tree depending on the size and/or shape of the block. When this restriction is applied, split_flag may be a flag indicating whether to split into a quad tree type or a flag indicating whether to split into a binary tree type. The size and shape of the block may be derived according to the depth information of the block, and the depth information may be signaled from the encoding apparatus 100 to the decoding apparatus 200.

When the size of a block falls within a specified range, only quad-tree partitioning may be possible. For example, the specified range may be defined by at least one of a maximum block size and a minimum block size in which quad-tree partitioning is possible.

Information indicating the maximum block size and/or minimum block size in which only quad tree-type division is possible may be signaled from the encoding apparatus 100 to the decoding apparatus 200 through a bitstream. Also, this information may be signaled for at least one unit of video, sequence, picture, parameter, tile group, and slice (or segment).

Alternatively, the maximum block size and/or the minimum block size may be fixed sizes predefined in the encoding apparatus 100 and the decoding apparatus 200. For example, when the size of a block is larger than 64x64 and smaller than 256x256, only quad-tree partitioning may be possible. In this case, split_flag may be a flag indicating whether to split into a quad tree.

If the size of the block is larger than the maximum transform block size, only a quad tree type of splitting may be possible. In this case, the divided block may be at least one of CU and TU.

In this case, split_flag may be a flag indicating whether to split into a quad tree.

When the size of a block falls within a specified range, only binary tree or ternary tree partitioning may be possible. Here, for example, the specified range may be defined by at least one of a maximum block size and a minimum block size in which binary tree or ternary tree partitioning is possible.

Information representing a maximum block size and/or a minimum block size in which only binary tree-type division or ternary tree-type division is possible may be signaled from the encoding apparatus 100 to the decoding apparatus 200 through a bitstream. In addition, this information may be signaled for at least one unit of sequence, picture, and slice (or segment).

Alternatively, the maximum block size and/or the minimum block size may be fixed sizes predefined in the encoding apparatus 100 and the decoding apparatus 200. For example, when the size of a block is greater than 8x8 and less than 16x16, only binary tree partitioning may be possible. In this case, split_flag may be a flag indicating whether to split into a binary tree type or a ternary tree type.

The above-described quad tree partitioning may be equally applied to binary tree and/or ternary tree partitioning.

The division of a block may be limited by the previous division. For example, when a block is divided into a specified binary tree shape to generate a plurality of divided blocks, each divided block may be additionally divided only into the specified tree shape. Here, the specified tree type may be at least one of a binary tree type, a ternary tree type, and a quad tree type.

When the horizontal size or vertical size of a divided block corresponds to a size that cannot be further divided, the aforementioned indicator may not be signaled.

7 is a diagram for explaining an embodiment of an intra prediction process.

Arrows extending from the center to the periphery of the graph of FIG. 7 may represent prediction directions of directional intra prediction modes. Also, numbers displayed adjacent to arrows may represent examples of mode values assigned to an intra prediction mode or a prediction direction of the intra prediction mode.

In FIG. 7 , the number 0 may indicate a planar mode, which is a non-directional intra prediction mode. Number 1 may indicate a DC mode, which is a non-directional intra prediction mode.

Intra coding and/or decoding may be performed using reference samples of neighboring blocks of the target block. A neighboring block may be a reconstructed neighboring block. A reference sample may mean a neighboring sample.

For example, intra encoding and/or decoding may be performed using a value of a reference sample included in a reconstructed neighboring block or a coding parameter.

The encoding apparatus 100 and/or the decoding apparatus 200 may generate a prediction block by performing intra prediction on the target block based on sample information in the target image. When performing intra prediction, the encoding apparatus 100 and/or the decoding apparatus 200 may generate a prediction block for a target block by performing intra prediction based on sample information in a target image. When intra prediction is performed, the encoding device 100 and/or the decoding device 200 may perform directional prediction and/or non-directional prediction based on at least one reconstructed reference sample.

A prediction block may refer to a block generated as a result of performing intra prediction. A prediction block may correspond to at least one of CU, PU, and TU.

A unit of a prediction block may be the size of at least one of CU, PU, and TU. The prediction block may have a square shape with a size of 2Nx2N or NxN. The size of NxN may include 4x4, 8x8, 16x16, 32x32, and 64x64.

Alternatively, the prediction block may be a square block having a size of 2x2, 4x4, 8x8, 16x16, 32x32, or 64x64, or may be a rectangular block having a size of 2x8, 4x8, 2x16, 4x16, or 8x16. there is.

Intra prediction may be performed according to an intra prediction mode for a target block. The number of intra prediction modes that the target block may have may be a predefined fixed value or may be a value determined differently according to properties of the prediction block. For example, the properties of the prediction block may include the size of the prediction block and the type of the prediction block. Also, the properties of the prediction block may indicate coding parameters for the prediction block.

For example, the number of intra prediction modes may be fixed to N regardless of the size of a prediction block. Alternatively, for example, the number of intra prediction modes may be 3, 5, 9, 17, 34, 35, 36, 65, 67, or 95.

The intra prediction mode may be a non-directional mode or a directional mode.

For example, the intra prediction mode may include two non-directional modes and 65 directional modes, corresponding to numbers 0 to 66 shown in FIG. 7 .

For example, when a specified intra prediction method is used, the intra prediction mode may include two non-directional modes and 93 directional modes, corresponding to numbers -14 to 80 shown in FIG. 7 .

The two non-directional modes may include DC mode and Planar mode.

The directional mode may be a prediction mode having a specific direction or a specific angle. The directional mode may also be referred to as an argular mode.

The intra prediction mode may be expressed as at least one of a mode number, a mode value, a mode angle, and a mode direction. That is to say, the terms “(mode) number of an intra prediction mode”, “(mode) value of an intra prediction mode”, “(mode) angle of an intra prediction mode” and “(mode) direction of an intra prediction mode) have the same meaning. and can be used interchangeably.

The number of intra prediction modes may be M. M may be 1 or more. In other words, the number of intra prediction modes may be M including the number of non-directional modes and the number of directional modes.

The number of intra prediction modes may be fixed to M regardless of block sizes and/or color components. For example, the number of intra prediction modes may be fixed to either 35 or 67 regardless of the size of a block.

Alternatively, the number of intra prediction modes may be different according to the shape, size, and/or color component type of a block.

For example, in FIG. 7 , directional prediction modes indicated by dotted lines may be applied only to prediction for non-square blocks.

For example, as the block size increases, the number of intra prediction modes may increase. Alternatively, as the block size increases, the number of intra prediction modes may decrease. When the size of a block is 4x4 or 8x8, the number of intra prediction modes may be 67. When the size of a block is 16x16, the number of intra prediction modes may be 35. When the size of a block is 32x32, the number of intra prediction modes may be 19. When the size of a block is 64x64, the number of intra prediction modes may be 7.

For example, the number of intra prediction modes may be different depending on whether the color component is a luma signal or a chroma signal. Alternatively, the number of intra prediction modes of the luma component block may be greater than the number of intra prediction modes of the chroma component block.

For example, in the case of a vertical mode having a mode value of 50, prediction may be performed in a vertical direction based on a pixel value of a reference sample. For example, in the case of a horizontal mode having a mode value of 18, prediction may be performed in a horizontal direction based on a pixel value of a reference sample.

Even in a directional mode other than the aforementioned mode, the encoding apparatus 100 and the decoding apparatus 200 may perform intra prediction on a target unit using a reference sample according to an angle corresponding to the directional mode.

An intra prediction mode positioned to the right of the vertical mode may be named a vertical-right mode. An intra-prediction mode positioned below the horizontal mode may be named a horizontal-below mode. For example, in FIG. 7 , intra prediction modes having a mode value of one of 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, and 66 are vertical It can be the right modes. Intra prediction modes having a mode value of 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, and 17 may be horizontal bottom modes.

The non-directional mode may include a DC mode and a planar mode. For example, the mode value of the DC mode may be 1. The mode value of the planner mode may be 0.

The directional mode may include an angular mode. Modes other than the DC mode and the planner mode among the plurality of intra prediction modes may be directional modes.

When the intra prediction mode is the DC mode, a prediction block may be generated based on an average of pixel values of a plurality of reference samples. For example, a pixel value of a prediction block may be determined based on an average of pixel values of a plurality of reference samples.

The number of intra prediction modes and the mode value of each intra prediction mode described above may be merely illustrative. The number of the aforementioned intra prediction modes and the mode value of each intra prediction mode may be differently defined according to embodiments, implementations, and/or needs.

In order to perform intra prediction on the target block, a step of checking whether samples included in the reconstructed neighboring block can be used as reference samples of the target block may be performed. If there is a sample that cannot be used as a reference sample of the target block among the samples of the neighboring block, a value generated by copying and/or interpolation using at least one sample value among the samples included in the reconstructed neighboring block This can be replaced with the sample value of a sample that cannot be used as a reference sample. If a value generated by copying and/or interpolation is replaced with a sample value of a sample, the sample may be used as a reference sample of the target block.

When intra prediction is used, a filter may be applied to at least one of a reference sample or a prediction sample based on at least one of an intra prediction mode and a size of a target block.

The type of filter applied to at least one of the reference sample and the prediction sample may differ according to at least one of an intra prediction mode of the target block, a size of the target block, and a shape of the target block. The type of filter may be classified according to one or more of a length of a filter tap, a value of a filter coefficient, and a filter strength. The length of the filter tap may mean the number of filter taps. Also, the number of filter taps may mean the length of a filter.

When the intra prediction mode is the planner mode, in generating the prediction block of the target block, the upper reference sample of the target sample, the left reference sample of the target sample, and the upper right reference sample of the target block according to the location of the prediction target sample in the prediction block. A sample value of the prediction target sample may be generated using a weight-sum of the lower left reference sample of the target block.

When the intra prediction mode is the DC mode, an average value of upper reference samples and left reference samples of the target block may be used to generate a prediction block of the target block. In addition, filtering using values of reference samples may be performed on specified rows or specified columns within the target block. The specified rows may be one or more top rows adjacent to the reference sample. The specified columns may be one or more left columns adjacent to the reference sample.

When the intra prediction mode is a directional mode, a prediction block may be generated using an upper reference sample, a left reference sample, an upper right reference sample, and/or a lower left reference sample of the target block.

Interpolation in units of real numbers may be performed to generate the prediction samples described above.

The intra prediction mode of the target block may be predicted from the intra prediction modes of neighboring blocks of the target block, and information used for prediction may be entropy encoded/decoded.

For example, if the intra prediction modes of the target block and the neighboring block are the same, it may be signaled that the intra prediction modes of the target block and the neighboring block are the same using a predefined flag.

For example, an indicator indicating an intra prediction mode identical to that of a target block among intra prediction modes of a plurality of neighboring blocks may be signaled.

If intra prediction modes of the target block and neighboring blocks are different from each other, information of the intra prediction mode of the target block may be encoded and/or decoded using entropy encoding and/or decoding.

The reconstructed reference samples used for intra prediction of the target block include below-left reference samples, left reference samples, above-left corner reference samples, and above reference samples. s and above-right reference samples, etc.

For example, the left reference samples may refer to reconstructed reference pixels adjacent to the left side of the target block. Top reference samples may refer to reconstructed reference pixels adjacent to the top of the target block. The top left corner reference sample may refer to a reconstructed reference pixel located at the top left corner of the target block. In addition, the lower left reference samples may refer to a reference sample located at the lower end of the left sample line among samples located on the same line as the left sample line composed of the left reference samples. The upper right reference samples may refer to reference samples located to the right of the upper pixel line among samples located on the same line as the upper sample line composed of the upper reference samples.

When the size of the target block is NxN, each of the lower left reference samples, the left reference samples, the upper reference samples, and the upper right reference samples may be N.

A prediction block may be generated through intra prediction of the target block. Generation of the prediction block may include determining values of pixels of the prediction block. The size of the target block and the prediction block may be the same.

Reference samples used for intra prediction of the target block may vary according to the intra prediction mode of the target block. A direction of an intra prediction mode may represent a dependency relationship between reference samples and pixels of a prediction block. For example, a value of a specified reference sample may be used as a value of one or more specified pixels of a prediction block. In this case, the specified reference sample and the specified one or more pixels of the prediction block may be samples and pixels specified as a straight line in the direction of the intra prediction mode. In other words, the value of the specified reference sample may be copied to a value of a pixel located in a direction opposite to the direction of the intra prediction mode. Alternatively, the value of a pixel of the prediction block may be a value of a reference sample located in the direction of the intra prediction mode based on the position of the pixel.

For example, when the intra prediction mode of the target block is a vertical mode, upper reference samples may be used for intra prediction. When the intra prediction mode is a vertical mode, a value of a pixel of a prediction block may be a value of a reference sample positioned vertically above the position of the pixel. Accordingly, the top reference samples that are top adjacent to the target block may be used for intra prediction. Also, values of pixels in one row of the prediction block may be the same as values of upper reference samples.

For example, when the intra prediction mode of the target block is a horizontal mode, left reference samples may be used for intra prediction. When the intra prediction mode is a horizontal mode, a value of a pixel of a prediction block may be a value of a reference sample located horizontally to the left of the pixel. Accordingly, left reference samples adjacent to the left of the target block may be used for intra prediction. Also, values of pixels of one column of the prediction block may be the same as values of left reference samples.

For example, when the mode value of the intra prediction mode of the target block is 34, at least some of the left reference samples, the upper left corner reference sample, and the upper reference samples may be used for intra prediction. When the mode value of the intra prediction mode is 34, the value of the pixel of the prediction block may be the value of the reference sample located diagonally at the top left of the pixel.

Also, when an intra prediction mode having a mode value of one of 52 to 66 is used, at least some of the upper right reference samples may be used for intra prediction.

Also, when an intra prediction mode having a mode value of one of 2 to 17 is used, at least some of lower left reference samples may be used for intra prediction.

Also, when an intra prediction mode having a mode value of one of 19 to 49 is used, the upper left corner reference sample may be used for intra prediction.

The number of reference samples used to determine the pixel value of one pixel of the prediction block may be one or two or more.

As described above, a pixel value of a pixel of a prediction block may be determined according to the position of the pixel and the position of the reference sample indicated by the direction of the intra prediction mode. When the position of the reference sample indicated by the position of the pixel and the direction of the intra prediction mode is an integer position, a value of one reference sample indicated by the integer position may be used to determine a pixel value of a pixel of the prediction block.

When the position of the reference sample indicated by the position of the pixel and the direction of the intra prediction mode is not an integer position, an interpolated reference sample may be generated based on two reference samples closest to the position of the reference sample. there is. The value of the interpolated reference sample may be used to determine a pixel value of a pixel of the predictive block. In other words, when the location of a pixel of the prediction block and the location of a reference sample indicated by the direction of the intra prediction mode indicate a gap between two reference samples, an interpolated value will be generated based on the values of the two samples. can

A prediction block generated by prediction may not be the same as the original target block. In other words, a prediction error, which is a difference between a target block and a prediction block, may exist, and a prediction error may also exist between pixels of the target block and pixels of the prediction block.

Hereinafter, the terms “difference”, “error” and “residual” may be used interchangeably and interchangeably.

For example, in the case of directional intra prediction, a larger prediction error may occur as the distance between a pixel of a prediction block and a reference sample increases. A discontinuity may occur between a prediction block generated by such a prediction error and a neighboring block.

Filtering of prediction blocks may be used to reduce prediction errors. Filtering may be adaptively applying a filter to a region considered to have a large prediction error among prediction blocks. For example, a region considered to have a large prediction error may be a boundary of a prediction block. Also, depending on the intra-prediction mode, a region considered to have a large prediction error among prediction blocks may be different, and filter characteristics may be different.

As shown in FIG. 8 , at least one of reference lines 0 to 3 may be used for intra prediction of a target block.

Each reference line in FIG. 8 may represent a reference sample line including one or more reference samples. The smaller the reference line number, the closer the reference sample line may be to the target block.

Samples of segment A and segment F may be obtained through padding using the nearest samples of segment B and segment E, respectively, instead of being obtained from reconstructed neighboring blocks.

Index information indicating a reference sample line to be used for intra prediction of a target block may be signaled. The index information may indicate a reference sample line used for intra prediction of a target block among a plurality of reference sample lines. For example, index information may have a value of 0 to 3.

When the upper boundary of the target block is the boundary of the CTU, only reference sample line 0 may be available. Therefore, in this case, index information may not be signaled. When a reference sample line other than reference sample line 0 is used, filtering of a prediction block described later may not be performed.

In the case of inter-color intra prediction, a prediction block for a target block of a second color component may be generated based on a corresponding reconstructed block of a first color component.

For example, the first color component may be a luma component, and the second color component may be a chroma component.

For intra prediction between color components, parameters of a linear model between a first color component and a second color component may be derived based on a template.

The template may include the top reference sample and/or the left reference sample of the target block, and the top reference sample and/or left reference sample of the reconstructed block of the first color component corresponding to these reference samples. there is.

For example, the parameters of the linear model are: 1) the value of the sample of the first color component having the largest value among the samples in the template, 2) the value of the sample of the second color component corresponding to the sample of this first color component, 3) the value of the sample of the first color component having the minimum value among the samples in the template and 4) the value of the sample of the second color component corresponding to the sample of the first color component.

When parameters of the linear model are derived, a prediction block for the target block may be generated by applying the corresponding reconstructed block to the linear model.

Depending on the image format, sub-sampling may be performed on the neighboring samples of the reconstructed block of the first color component and the corresponding reconstructed block. For example, when one sample of the second color component corresponds to four samples of the first color component, one corresponding sample may be calculated by subsampling the four samples of the first color component. there is. When subsampling is performed, derivation of parameters of the linear model and intra prediction between color components may be performed based on the subsampled corresponding samples.

Whether intra prediction between color components is performed and/or a template range may be signaled as an intra prediction mode.

A target block may be divided into 2 or 4 sub-blocks in a horizontal direction and/or a vertical direction.

Divided sub-blocks may be sequentially reconstructed. That is, as intra prediction is performed on a sub-block, a sub-prediction block for the sub-block may be generated. In addition, as inverse quantization and/or inverse transformation are performed on the sub-block, a sub-residual block for the sub-block may be generated. A reconstructed sub-block may be generated by adding the sub-prediction block to the sub-residual block. The reconstructed subblock may be used as a reference sample for intra prediction of a subblock of a later order.

A sub-block may be a block including a specified number (eg, 16) or more samples. Thus, for example, when the target block is an 8x4 block or a 4x8 block, the target block may be divided into two sub-blocks. Also, if the target block is a 4x4 block, the target block cannot be divided into sub blocks. If the target block has other sizes, the target block may be divided into 4 sub-blocks.

Information about whether intra prediction based on these sub-blocks is performed and/or a division direction (horizontal direction or vertical direction) may be signaled.

Such sub-block-based intra prediction may be limited to being performed only when using the reference sample line 0. When sub-block-based intra prediction is performed, filtering of a prediction block described later may not be performed.

A final prediction block may be generated by performing filtering on a prediction block generated by intra prediction.

Filtering may be performed by applying specific weights to the filtering target sample, the left reference sample, the top reference sample, and/or the top left reference sample.

A weight and/or a reference sample (or a range of reference samples or a location of the reference sample) used for filtering may be determined based on at least one of a block size, an intra prediction mode, and a position of a sample to be filtered within a prediction block. there is.

For example, filtering may be performed only for a specified intra prediction mode (eg, DC mode, planar mode, vertical mode, horizontal mode, diagonal mode, and/or adjacent diagonal mode).

The adjacent diagonal mode may be a mode having a number obtained by adding k to the number of the diagonal mode, or a mode having a number obtained by subtracting k from the number of the diagonal mode. In other words, the number of adjacent diagonal modes may be the sum of the number of diagonal modes and k, and may be the difference between the number of diagonal modes and k. For example, k may be a positive integer of 8 or less.

The intra prediction mode of the target block may be derived using intra prediction modes of neighboring blocks existing around the target block, and the derived intra prediction mode may be entropy-encoded and/or entropy-decoded.

For example, if the intra-prediction mode of the target block and the intra-prediction mode of the neighboring block are the same, information that the intra-prediction mode of the target block and the intra-prediction mode of the neighboring block are the same may be signaled using specified flag information. .

Also, for example, indicator information about a neighboring block having the same intra prediction mode as that of the target block among intra prediction modes of a plurality of neighboring blocks may be signaled.

For example, if the intra prediction mode of the target block and the intra prediction mode of the neighboring block are different, entropy encoding and/or entropy decoding based on the intra prediction mode of the neighboring block is performed to obtain information about the intra prediction mode of the target block. Entropy encoding and/or entropy decoding may be performed.

9 is a diagram for explaining an embodiment of an inter prediction process.

The rectangle shown in FIG. 9 may represent an image (or picture). Also, arrows in FIG. 9 may indicate prediction directions. An arrow pointing from a first picture to a second picture may indicate that the second picture refers to the first picture. That is, an image may be encoded and/or decoded according to a prediction direction.

Each image may be classified into an intra picture (I picture), a uni-prediction picture (P picture), and a bi-prediction picture (B picture) according to an encoding type. Each picture may be coded and/or decoded according to the coding type of each picture.

When a target image to be encoded is an I-picture, the target image may be encoded using data within the image itself without inter prediction referring to another image. For example, an I picture can be coded only with intra prediction.

When the target image is a P picture, the target image may be coded through inter prediction using only reference pictures existing in one direction. Here, the unidirectional direction may be a forward direction or a reverse direction.

When the target video is a B picture, the target video may be coded through inter prediction using reference pictures existing in both directions or inter prediction using reference pictures existing in one of forward and backward directions. Here, both directions may be forward and reverse directions.

P-pictures and B-pictures that are coded and/or decoded using reference pictures may be regarded as pictures in which inter prediction is used.

Below, inter prediction in inter mode according to an embodiment is described in detail.

Inter prediction or motion compensation may be performed using a reference image and motion information.

In the inter mode, the encoding apparatus 100 may perform inter prediction and/or motion compensation on a target block. The decoding apparatus 200 may perform inter prediction and/or motion compensation corresponding to inter prediction and/or motion compensation performed in the encoding apparatus 100 on a target block.

Motion information on the target block may be derived during inter prediction by each of the encoding apparatus 100 and the decoding apparatus 200 . Motion information may be derived using motion information of a reconstructed neighboring block, motion information of a collocated block, and/or motion information of a block adjacent to the collocated block.

For example, the encoding apparatus 100 or the decoding apparatus 200 performs prediction and/or motion compensation by using motion information of a spatial candidate and/or a temporal candidate as motion information of a target block. can be done A target block may mean a PU and/or a PU partition.

A spatial candidate may be a reconstructed block that is spatially adjacent to the target block.

A temporal candidate may be a reconstructed block corresponding to a target block in an already reconstructed collocated picture (col picture).

In inter prediction, the encoding apparatus 100 and the decoding apparatus 200 may improve encoding efficiency and decoding efficiency by using motion information of spatial candidates and/or temporal candidates. Motion information of spatial candidates may be referred to as spatial motion information. Motion information of temporal candidates may be referred to as temporal motion information.

Hereinafter, motion information of a spatial candidate may be motion information of a PU including the spatial candidate. Motion information of the temporal candidate may be motion information of a PU including the temporal candidate. Motion information of the candidate block may be motion information of a PU including the candidate block.

Inter prediction may be performed using a reference picture.

A reference picture may be at least one of a previous picture of the target picture or a subsequent picture of the target picture. A reference picture may refer to an image used for prediction of a target block.

In inter prediction, a region within a reference picture may be specified by using a reference picture index (or refIdx) indicating a reference picture and a motion vector to be described later. Here, a specified region within a reference picture may represent a reference block.

Inter prediction may select a reference picture and may select a reference block corresponding to a target block within the reference picture. In addition, inter prediction may generate a prediction block for a target block using the selected reference block.

Motion information may be derived during inter prediction by each of the encoding apparatus 100 and the decoding apparatus 200 .

A spatial candidate may be a block that 1) exists in the target picture, 2) has already been reconstructed through encoding and/or decoding, and 3) is located adjacent to or at a corner of the target block. Here, the block located at the corner of the target block may be a block vertically adjacent to a neighboring block horizontally adjacent to the target block or a block horizontally adjacent to a neighboring block vertically adjacent to the target block. "A block located at a corner of a target block" may have the same meaning as "a block adjacent to a corner of a target block". A "block located at a corner of a target block" may be included in a "block adjacent to the target block".

For example, spatial candidates include a reconstructed block located to the left of the target block, a reconstructed block located on top of the target block, a reconstructed block located at the lower left corner of the target block, and a reconstructed block located at the upper right corner of the target block. It can be a reconstructed block or a reconstructed block located in the upper left corner of the target block.

Each of the encoding apparatus 100 and the decoding apparatus 200 may identify a block existing at a position spatially corresponding to a target block within a col picture. A position of a target block in a target picture and a position of an identified block in a collocated picture may correspond to each other.

Each of the encoding apparatus 100 and the decoding apparatus 200 may determine a col block existing at a predefined relative position with respect to the identified block as a temporal candidate. The predefined relative position may be a position inside and/or outside the identified block.

For example, the collocated block may include a first collocated block and a second collocated block. When the coordinates of the identified block are (xP, yP) and the size of the identified block is (nPSW, nPSH), the first collocated block may be a block located at the coordinates (xP + nPSW, yP + nPSH). The second collocated block may be a block located at coordinates (xP + (nPSW >> 1), yP + (nPSH >> 1). The second call block may be selectively used when the first call block is unavailable.

The motion vector of the target block may be determined based on the motion vector of the collocated block. Each of the encoding apparatus 100 and the decoding apparatus 200 may scale a motion vector of a collocated block. A scaled motion vector of the collocated block may be used as a motion vector of the target block. Also, a motion vector of motion information of a temporal candidate stored in the list may be a scaled motion vector.

A ratio of the motion vector of the target block and the motion vector of the collocated block may be equal to the ratio of the first temporal distance to the second temporal distance. The first temporal distance may be the distance between the reference picture of the target block and the target picture. The second temporal distance may be a distance between a reference picture of a collocated block and a collocated picture.

A method of deriving motion information may change according to an inter prediction mode of a target block. For example, as inter prediction modes applied for inter prediction, an Advanced Motion Vector Predictor (AMVP) mode, a merge mode and a skip mode, a merge mode with motion vector difference, There may be sub-block merge mode, triangulation mode, inter-intra combined prediction mode, affine inter mode, and current picture reference mode. Merge mode may also be referred to as motion merge mode. Below, each of the modes is described in detail.

1) AMVP mode

When the AMVP mode is used, the encoding apparatus 100 may search for similar blocks in the neighborhood of the target block. The encoding apparatus 100 may obtain a prediction block by performing prediction on a target block using motion information of a similar block found. The encoding apparatus 100 may encode a residual block that is a difference between a target block and a prediction block.

1-1) Creating a predicted motion vector candidate list

When the AMVP mode is used as the prediction mode, each of the encoding apparatus 100 and the decoding apparatus 200 may generate a predictive motion vector candidate list using a motion vector of a spatial candidate, a motion vector of a temporal candidate, and a zero vector. there is. The predicted motion vector candidate list may include one or more predicted motion vector candidates. At least one of the motion vector of the spatial candidate, the motion vector of the temporal candidate, and the zero vector may be determined and used as the predicted motion vector candidate.

Hereinafter, the terms “predicted motion vector (candidate)” and “motion vector (candidate)” may be used interchangeably and may be used interchangeably.

Hereinafter, the terms “predicted motion vector candidate” and “AMVP candidate” may be used interchangeably and may be used interchangeably.

Hereinafter, the terms “prediction motion vector candidate list” and “AMVP candidate list” may be used in the same meaning and may be used interchangeably.

Spatial candidates may include reconstructed spatial neighboring blocks. In other words, the motion vector of the reconstructed neighboring block may be referred to as a spatial prediction motion vector candidate.

Temporal candidates may include a collocated block and a block adjacent to the collocated block. In other words, a motion vector of a collocated block or a motion vector of a block adjacent to the collocated block may be referred to as a temporal prediction motion vector candidate.

The zero vector may be a (0, 0) motion vector.

The predictive motion vector candidate may be a motion vector predictor for motion vector prediction. Also, in the encoding apparatus 100, the predicted motion vector candidate may be an initial motion vector search position.

1-2) Motion vector search using a predictive motion vector candidate list

The encoding apparatus 100 may determine a motion vector to be used for encoding a target block within a search range by using the predictive motion vector candidate list. Also, the encoding apparatus 100 may determine a predictive motion vector candidate to be used as a predictive motion vector of the target block from among predictive motion vector candidates in the predictive motion vector candidate list.

A motion vector to be used for encoding a target block may be a motion vector that can be encoded with minimal cost.

Also, the encoding apparatus 100 may determine whether to use the AMVP mode in encoding the target block.

1-3) Transmission of inter prediction information

The encoding apparatus 100 may generate a bitstream including inter prediction information required for inter prediction. The decoding apparatus 200 may perform inter prediction on a target block using inter prediction information of a bitstream.

The inter-prediction information includes 1) mode information indicating whether the AMVP mode is used, 2) predictive motion vector index, 3) motion vector difference (MVD), 4) reference direction, and 5) reference picture index. can do.

Hereinafter, the terms “predicted motion vector index” and “AMVP index” may be used interchangeably and may be used interchangeably.

Also, the inter prediction information may include a residual signal.

When the mode information indicates that the AMVP mode is used, the decoding apparatus 200 may obtain a predicted motion vector index, a motion vector difference, a reference direction, and a reference picture index from a bitstream through entropy decoding.

The predictive motion vector index may indicate a predictive motion vector candidate used for prediction of a target block among predictive motion vector candidates included in the predictive motion vector candidate list.

1-4) Inter prediction of AMVP mode using inter prediction information

The decoding apparatus 200 may derive a predicted motion vector candidate using the predicted motion vector candidate list, and may determine motion information of a target block based on the derived motion vector predicted candidate.

The decoding apparatus 200 may determine a motion vector candidate for a target block from among motion vector predictor candidates included in a predictor motion vector candidate list by using the predictor motion vector index. The decoding apparatus 200 may select a predicted motion vector candidate indicated by a predicted motion vector index as a predicted motion vector of a target block from among motion vector predicted candidates included in the predicted motion vector candidate list.

The encoding apparatus 100 may generate an entropy-encoded predicted motion vector index by applying entropy encoding to the predicted motion vector index, and may generate a bitstream including the entropy-encoded predicted motion vector index. The entropy-encoded predicted motion vector index may be signaled from the encoding apparatus 100 to the decoding apparatus 200 through a bitstream. The decoding apparatus 200 may extract an entropy-encoded predictive motion vector index from a bitstream and obtain the predictive motion vector index by applying entropy decoding to the entropy-encoded predictive motion vector index.

A motion vector actually used for inter prediction of a target block may not coincide with a predicted motion vector. MVD may be used to represent a difference between a motion vector actually used for inter prediction of a target block and a predicted motion vector. The encoding apparatus 100 may derive a predicted motion vector similar to a motion vector actually used for inter prediction of a target block in order to use an MVD having a size as small as possible.

MVD may be a difference between a motion vector of a target block and a predicted motion vector. The encoding apparatus 100 may calculate the MVD and generate an entropy-encoded MVD by applying entropy encoding to the MVD. The encoding apparatus 100 may generate a bitstream including an entropy-encoded MDV.

The MVD may be transmitted from the encoding device 100 to the decoding device 200 through a bitstream. The decoding apparatus 200 may extract the entropy-encoded MVD from the bitstream and obtain the MVD by applying entropy decoding to the entropy-encoded MVD.

The decoding apparatus 200 may derive the motion vector of the target block by adding the MVD and the predicted motion vector. In other words, the motion vector of the target block derived by the decoding apparatus 200 may be the sum of the MVD and the motion vector candidate.

Also, the encoding apparatus 100 may generate entropy-encoded MVD resolution information by applying entropy encoding to the calculated MVD resolution information, and may generate a bitstream including the entropy-encoded MVD resolution information. The decoding apparatus 200 may extract entropy-encoded MVD resolution information from a bitstream, and obtain MVD resolution information by applying entropy decoding to the entropy-encoded MVD resolution information. The decoding apparatus 200 may adjust the resolution of the MVD using the MVD resolution information.

Meanwhile, the encoding device 100 may calculate the MVD based on the affine model. The decoding apparatus 200 may derive an affine control motion vector of the target block through the sum of the MVD and the affine control motion vector candidate, and may derive a motion vector for a subblock using the affine control motion vector. there is.

A reference direction may indicate a reference picture list used for prediction of a target block. For example, the reference direction may indicate one of a reference picture list L0 and a reference picture list L1.

The reference direction indicates only a reference picture list used for prediction of the target block, and may not indicate that directions of reference pictures are limited to a forward direction or a backward direction. In other words, each of the reference picture list L0 and the reference picture list L1 may include forward and/or backward pictures.

That the reference direction is uni-direction may mean that one reference picture list is used. Bi-direction of the reference direction may mean that two reference picture lists are used. In other words, the reference direction can indicate one of: that only the reference picture list L0 is used, that only the reference picture list L1 is used, and two reference picture lists.

The reference picture index may indicate a reference picture used for prediction of a target block among reference pictures of a reference picture list. The encoding apparatus 100 may generate an entropy-coded reference picture index by applying entropy encoding to the reference picture index, and may generate a bitstream including the entropy-coded reference picture index. The entropy-encoded reference picture index may be signaled from the encoding device 100 to the decoding device 200 through a bitstream. The decoding apparatus 200 may extract an entropy-encoded reference picture index from a bitstream and obtain the reference picture index by applying entropy decoding to the entropy-encoded reference picture index.

When two reference picture lists are used for prediction of a target block. One reference picture index and one motion vector may be used for each reference picture list. Also, when two reference picture lists are used for prediction of a target block, two prediction blocks may be specified for the target block. For example, a (final) prediction block of the target block may be generated through an average or a weighted-sum of two prediction blocks of the target block.

The motion vector of the target block may be derived by the predicted motion vector index, MVD, reference direction, and reference picture index.

The decoding apparatus 200 may generate a prediction block for the target block based on the derived motion vector and the reference picture index. For example, the prediction block may be a reference block indicated by a derived motion vector in a reference picture indicated by a reference picture index.

By encoding the predicted motion vector index and the MVD without encoding the motion vector of the target block, the amount of bits transmitted from the encoding apparatus 100 to the decoding apparatus 200 can be reduced and encoding efficiency can be improved.

Motion information of a neighboring block reconstructed with respect to the target block may be used. In a specific inter-prediction mode, the encoding apparatus 100 may not separately encode the motion information of the target block itself. Motion information of the target block is not encoded, and other information capable of inducing motion information of the target block through motion information of the reconstructed neighboring block may be encoded instead. As other information is encoded instead, the amount of bits transmitted to the decoding apparatus 200 can be reduced and encoding efficiency can be improved.

For example, as an inter prediction mode in which motion information of the target block is not directly encoded, there may be a skip mode and/or a merge mode. In this case, the encoding apparatus 100 and the decoding apparatus 200 may use an identifier and/or an index indicating which unit's motion information among reconstructed neighboring units is used as the motion information of the target unit.

2) Merge Mode

As a method of deriving motion information of a target block, there is a merge. Merge may refer to merging of motions of a plurality of blocks. Merge may mean applying motion information of one block to another block as well. In other words, merge mode may refer to a mode in which motion information of a target block is derived from motion information of neighboring blocks.

When merge mode is used, the encoding apparatus 100 may perform prediction of motion information of a target block by using motion information of a spatial candidate and/or motion information of a temporal candidate. The spatial candidate may include a reconstructed spatial neighboring block spatially adjacent to the target block. A spatial neighboring block may include a left neighboring block and an upper neighboring block. Temporal candidates may include call blocks. The terms "spatial candidate" and "spatial merge candidate" may be used interchangeably and may be used interchangeably. The terms "temporal candidate" and "temporal merge candidate" may be used interchangeably and may be used interchangeably.

The encoding apparatus 100 may obtain a prediction block through prediction. The encoding apparatus 100 may encode a residual block that is a difference between a target block and a prediction block.

2-1) Creating a merge candidate list

When the merge mode is used, each of the encoding apparatus 100 and the decoding apparatus 200 may generate a merge candidate list using motion information of spatial candidates and/or motion information of temporal candidates. Motion information may include 1) a motion vector, 2) a reference picture index, and 3) a reference direction. The reference direction can be unidirectional or bidirectional. The reference direction may mean an inter prediction indicator.

The merge candidate list may include merge candidates. A merge candidate may be motion information. In other words, the merge candidate list may be a list in which motion information is stored.

Merge candidates may be motion information such as temporal candidates and/or spatial candidates. In other words, the merge candidate list may include motion information such as a temporal candidate and/or a spatial candidate.

Also, the merge candidate list may include a new merge candidate generated by a combination of merge candidates already existing in the merge candidate list. In other words, the merge candidate list may include new motion information generated by combining motion information already existing in the merge candidate list.

Also, the merge candidate list may include a history-based merge candidate. A history-based merge candidate may be motion information of a block encoded and/or decoded prior to a target block.

Also, the merge candidate list may include a merge candidate based on an average of two merge candidates.

Merge candidates may be specified modes for deriving inter prediction information. A merge candidate may be information indicating a specific mode for deriving inter prediction information. Inter prediction information of a target block may be derived according to a specified mode indicated by a merge candidate. In this case, the specified mode may include a process of deriving a series of inter prediction information. This specified mode may be an inter prediction information derivation mode or a motion information derivation mode.

Inter prediction information of a target block may be derived according to a mode indicated by a merge candidate selected by a merge index among merge candidates in the merge candidate list.

For example, the motion information derivation modes in the merge candidate list may be at least one of 1) a motion information derivation mode in units of sub-blocks and 2) an affine motion information derivation mode.

Also, the merge candidate list may include motion information of a zero vector. A zero vector may be referred to as a zero merge candidate.

In other words, the motion information in the merge candidate list is: 1) motion information of a spatial candidate, 2) motion information of a temporal candidate, 3) motion information generated by a combination of motion information already existing in the merge candidate list, 4) zero vector may be at least one of

Motion information may include 1) a motion vector, 2) a reference picture index, and 3) a reference direction. A reference direction may be referred to as an inter prediction indicator. The reference direction can be unidirectional or bidirectional. A unidirectional reference direction may represent L0 prediction or L1 prediction.

The merge candidate list may be generated before prediction by merge mode is performed.

The number of merge candidates in the merge candidate list may be predefined. The encoding device 100 and the decoding device 200 may add merge candidates to the merge candidate list according to a predefined method and a predefined order so that the merge candidate list includes a predefined number of merge candidates. The merge candidate list of the encoding device 100 and the merge candidate list of the decoding device 200 may be the same through a predefined method and a predefined ranking.

Merge may be applied in units of CUs or units of PUs. When merging is performed in units of CUs or PUs, the encoding apparatus 100 may transmit a bitstream including predefined information to the decoding apparatus 200. For example, the predefined information includes 1) information indicating whether merging is to be performed for each block partition, 2) which block to merge with among blocks that are spatial candidates and/or temporal candidates for the target block. It may contain information on whether

2-2) Motion vector search using merge candidate list

The encoding apparatus 100 may determine a merge candidate to be used for encoding a target block. For example, the encoding apparatus 100 may perform predictions on a target block using merge candidates of the merge candidate list and generate residual blocks for the merge candidates. The encoding apparatus 100 may use a merge candidate that requires the least cost in encoding the prediction and residual blocks for encoding the target block.

Also, the encoding apparatus 100 may determine whether to use merge mode in encoding a target block.

2-3) Transmission of inter prediction information

The encoding apparatus 100 may generate a bitstream including inter prediction information required for inter prediction. The encoding apparatus 100 may generate entropy-encoded inter-prediction information by performing entropy encoding on the inter-prediction information, and may transmit a bitstream including the entropy-encoded inter-prediction information to the decoding apparatus 200. Entropy-encoded inter prediction information may be signaled from the encoding apparatus 100 to the decoding apparatus 200 through a bitstream. The decoding apparatus 200 may extract entropy-encoded inter prediction information from a bitstream and obtain inter prediction information by performing entropy decoding on the entropy-encoded inter prediction information.

The decoding apparatus 200 may perform inter prediction on a target block using inter prediction information of a bitstream.

The inter prediction information may include 1) mode information indicating whether merge mode is used, 2) merge index, and 3) correction information.

Also, the inter prediction information may include a residual signal.

The decoding apparatus 200 may obtain the merge index from the bitstream only when the mode information indicates that the merge mode is used.

Mode information may be a merge flag. A unit of mode information may be a block. Information about a block may include mode information, and the mode information may indicate whether merge mode is applied to the block.

The merge index may indicate a merge candidate used for prediction of a target block among merge candidates included in the merge candidate list. Alternatively, the merge index may indicate which block among neighboring blocks that are spatially or temporally adjacent to the target block is merged with.

The encoding apparatus 100 may select a merge candidate having the highest encoding performance among merge candidates included in the merge candidate list, and set a merge index value to indicate the selected merge candidate.

The correction information may be information used for motion vector correction. The encoding device 100 may generate correction information. The decoding apparatus 200 may correct the motion vector of the merge candidate selected by the merge index based on the correction information.

The correction information may include at least one of information indicating whether correction is performed, correction direction information, and correction size information. A prediction mode for correcting a motion vector based on signaled correction information may be referred to as a merge mode with motion vector difference.

2-4) Inter prediction of merge mode using inter prediction information

The decoding apparatus 200 may perform prediction on a target block by using a merge candidate indicated by a merge index among merge candidates included in the merge candidate list.

The motion vector of the target block may be specified by the motion vector of the merge candidate indicated by the merge index, the reference picture index, and the reference direction.

3) Skip Mode

The skip mode may be a mode in which motion information of a spatial candidate or motion information of a temporal candidate is applied to a target block as it is. Also, the skip mode may be a mode not using a residual signal. That is to say, when skip mode is used, the reconstructed block may be the same as the predicted block.

A difference between merge mode and skip mode may be transmission or use of a residual signal. In other words, skip mode may be similar to merge mode except that no residual signal is transmitted or used.

When the skip mode is used, the encoding apparatus 100 transmits information indicating which block's motion information among spatial or temporal candidate blocks is used as the motion information of the target block to the decoding apparatus 200 through a bitstream. can transmit The encoding apparatus 100 may generate entropy-encoded information by performing entropy encoding on such information, and may signal the entropy-encoded information to the decoding apparatus 200 through a bitstream. The decoding apparatus 200 may extract entropy-encoded information from a bitstream and obtain information by performing entropy decoding on the entropy-encoded information.

Also, when the skip mode is used, the encoding device 100 may not transmit other syntax element information such as MVD to the decoding device 200. For example, when the skip mode is used, the encoding apparatus 100 may not signal a syntax element related to at least one of the MVD, the coded block flag, and the transform coefficient level to the decoding apparatus 200.

3-1) Creation of merge candidate list

Skip mode can also use a merge candidate list. In other words, the merge candidate list can be used in both merge mode and skip mode. In this regard, the merge candidate list may be referred to as a "skip candidate list" or a "merge/skip candidate list".

Alternatively, skip mode may use a separate candidate list different from merge mode. In this case, in the description below, the merge candidate list and the merge candidate may be replaced with the skip candidate list and the skip candidate, respectively.

The merge candidate list may be generated before prediction by skip mode is performed.

3-2) Motion vector search using merge candidate list

The encoding apparatus 100 may determine a merge candidate to be used for encoding a target block. For example, the encoding apparatus 100 may perform predictions on a target block using merge candidates of a merge candidate list. The encoding apparatus 100 may use a merge candidate requiring a minimum prediction cost for encoding a target block.

Also, the encoding apparatus 100 may determine whether to use a skip mode in encoding a target block.

3-3) Transmission of inter prediction information

The inter-prediction information may include 1) mode information indicating whether a skip mode is used and 2) a skip index.

The skip index may be the same as the aforementioned merge index.

When the skip mode is used, the target block may be coded without a residual signal. Inter prediction information may not include a residual signal. Alternatively, the bitstream may not include a residual signal.

The decoding apparatus 200 may obtain the skip index from the bitstream only when the mode information indicates that the skip mode is used. As mentioned above, the merge index and skip index may be the same. The decoding apparatus 200 may obtain the skip index from the bitstream only when the mode information indicates that merge mode or skip mode is used.

The skip index may indicate a merge candidate used for prediction of a target block among merge candidates included in the merge candidate list.

3-4) Skip Mode Inter Prediction Using Inter Prediction Information

The decoding apparatus 200 may perform prediction on a target block by using a merge candidate indicated by a skip index among merge candidates included in the merge candidate list.

A motion vector of a target block may be specified by a motion vector of a merge candidate indicated by a skip index, a reference picture index, and a reference direction.

4) Current picture reference mode

The current picture reference mode may mean a prediction mode using a pre-reconstructed region in a target picture to which a target block belongs.

A motion vector may be used to specify the pre-reconstructed area. Whether the target block is encoded in the current picture reference mode can be determined using the reference picture index of the target block.

A flag or index indicating whether the target block is a block encoded in the current picture reference mode may be signaled from the encoding device 100 to the decoding device 200. Alternatively, whether the target block is a block coded in the current picture reference mode may be inferred through a reference picture index of the target block.

When the target block is coded in the current picture reference mode, the target picture may exist at a fixed position or an arbitrary position in the reference picture list for the target block.

For example, the fixed position may be a position where the value of the reference picture index is 0 or the last position.

If the target picture exists at an arbitrary position in the reference picture list, a separate reference picture index indicating the arbitrary position may be signaled from the encoding apparatus 100 to the decoding apparatus 200.

5) Subblock merge mode

A sub-block merge mode may mean a mode for deriving motion information for a sub-block of a CU.

When the sub-block merge mode is applied, motion information of a call sub-block of a target sub-block in a reference image (ie, a sub-block based temporal merge candidate) and/or an affine control point motion vector A subblock merge candidate list may be generated using an affine control point motion vector merge candidate.

6) Triangle partition mode

In the triangular division mode, divided target blocks may be generated by dividing the target block in a diagonal direction. For each divided target block, motion information of each divided target block may be derived, and prediction samples for each divided target block may be derived using the derived motion information. A prediction sample of the target block may be derived through a weighted sum of prediction samples of the divided target blocks.

7) Inter-intra combined prediction mode

The inter-intra combined prediction mode may be a mode in which a prediction sample of a target block is derived by using a weighted sum of prediction samples generated by inter prediction and prediction samples generated by intra prediction.

In the aforementioned modes, the decoding apparatus 200 may perform self-correction on the derived motion information. For example, the decoding apparatus 200 may search a specific area based on a reference block indicated by the derived motion information to search for motion information having a minimum sum of absolute differences (SAD). and the searched motion information may be derived as corrected motion information.

In the aforementioned modes, the decoding apparatus 200 may perform compensation for prediction samples derived through inter prediction using an optical flow.

In the aforementioned AMVP mode, merge mode, and skip mode, motion information to be used for prediction of a target block among motion information in the list may be specified through an index of the list.

In order to improve encoding efficiency, the encoding apparatus 100 may signal only the index of an element that causes the least cost in inter prediction of a target block among elements in the list. The encoding device 100 may encode the index and signal the encoded index.

Accordingly, the aforementioned lists (ie, the predicted motion vector candidate list and the merge candidate list) may have to be derived in the same manner based on the same data in the encoding apparatus 100 and the decoding apparatus 200. Here, the same data may include a reconstructed picture and a reconstructed block. Also, in order to specify an element by index, the order of the elements within the list may need to be constant.

10 shows spatial candidates according to an example.

In Fig. 10, the positions of spatial candidates are shown.

A large block in the middle may represent a target block. Five small blocks may represent spatial candidates.

Coordinates of the target block may be (xP, yP), and the size of the target block may be (nPSW, nPSH).

The spatial candidate A ₀ may be a block adjacent to the lower left corner of the target block. A ₀ may be a block occupying pixels of coordinates (xP - 1, yP + nPSH).

Spatial candidate A ₁ may be a block adjacent to the left of the target block. A ₁ may be the lowest block among blocks adjacent to the left side of the target block. Alternatively, A ₁ may be a block adjacent to the top of A ₀ . A ₁ may be a block occupying pixels of coordinates (xP - 1, yP + nPSH - 1).

Spatial candidate B ₀ may be a block adjacent to the upper right corner of the target block. B ₀ may be a block occupying pixels of coordinates (xP + nPSW, yP - 1).

Spatial candidate B ₁ may be a block adjacent to the top of the target block. B ₁ may be the rightmost block among blocks adjacent to the top of the target block. Alternatively, B ₁ may be a block adjacent to the left side of B ₀ . B ₁ may be a block occupying pixels of coordinates (xP + nPSW - 1, yP - 1).

Spatial candidate B ₂ may be a block adjacent to the upper left corner of the target block. B ₂ may be a block occupying pixels of coordinates (xP - 1, yP - 1).

Determination of the availability of spatial and temporal candidates

In order to include the motion information of the spatial candidate or the motion information of the temporal candidate in the list, it must be determined whether the motion information of the spatial candidate or the motion information of the temporal candidate is available.

Hereinafter, a candidate block may include a spatial candidate and a temporal candidate.

For example, the above determination may be made by sequentially applying steps 1) to 4) below.

Step 1) If the PU including the candidate block is outside the boundary of the picture, the availability of the candidate block may be set to false. "Availability is set to false" may mean the same as "availability is set to unavailability".

Step 2) If the PU including the candidate block is outside the boundary of the slice, the availability of the candidate block may be set to false. If the target block and the candidate block are located in different slices, the availability of the candidate block may be set to false.

Step 3) If the PU including the candidate block is outside the boundary of the tile, the availability of the candidate block may be set to false. If the target block and the candidate block are located in different tiles, the availability of the candidate block may be set to false.

Step 4) If the prediction mode of the PU including the candidate block is an intra prediction mode, availability of the candidate block may be set to false. If the PU containing the candidate block does not use inter prediction, the availability of the candidate block may be set to false.

As shown in FIG. 11 , in adding motion information of spatial candidates to the merge list, the order of A ₁ , B ₁ , B ₀ , A ₀ and B ₂ may be used. That is, in the order of A ₁ , B ₁ , B ₀ , A ₀ and B ₂ , motion information of available spatial candidates may be added to the merge list.

Merge list derivation method in merge mode and skip mode

As described above, the maximum number of merge candidates in the merge list may be set. The set maximum number is indicated by N. The set number may be transmitted from the encoding device 100 to the decoding device 200. A slice header of a slice may include N. In other words, the maximum number of merge candidates of the merge list for the target block of the slice may be set by the slice header. For example, the value of N may be 5 by default.

Motion information (ie, merge candidates) may be added to the merge list in the order of steps 1) to 4) below.

Step 1) Available spatial candidates among spatial candidates may be added to the merge list. Motion information of available spatial candidates may be added to the merge list in the order shown in FIG. 11 . In this case, if motion information of an available spatial candidate overlaps with other motion information already existing in the merge list, the motion information may not be added to the merge list. Checking whether it overlaps with other motion information existing in the list can be abbreviated as "redundancy check".

A maximum of N pieces of motion information may be added.

Step 2) If the number of pieces of motion information in the merge list is smaller than N and a temporal candidate is available, the motion information of the temporal candidate may be added to the merge list. In this case, if motion information of an available temporal candidate overlaps with other motion information already existing in the merge list, the motion information may not be added to the merge list.

Step 3) If the number of pieces of motion information in the merge list is smaller than N and the type of the target slice is “B”, the combined motion information generated by combined bi-prediction is added to the merge list. can

The target slice may be a slice including the target block.

The combined motion information may be a combination of L0 motion information and L1 motion information. The L0 motion information may be motion information referring only to the reference picture list L0. The L1 motion information may be motion information referring only to the reference picture list L1.

Within the merge list, there may be one or more L0 motion information. Also, within the merge list, there may be one or more L1 motion information.

The combined motion information may be one or more. In generating the combined motion information, which L0 motion information and which L1 motion information among one or more pieces of L0 motion information and one or more pieces of L1 motion information are to be used may be predefined. One or more pieces of combined motion information may be generated in a predefined order by combined bidirectional prediction using pairs of different pieces of motion information in a merge list. One of the pairs of different motion information may be L0 motion information and the other may be L1 motion information.

For example, the combined motion information added with the highest priority may be a combination of L0 motion information having a merge index of 0 and L1 motion information having a merge index of 1. If the motion information with a merge index of 0 is not L0 motion information or the motion information with a merge index of 1 is not L1 motion information, the combined motion information may not be generated and added. Motion information added next may be a combination of L0 motion information having a merge index of 1 and L1 motion information having a merge index of 0. The following specific combinations may follow other combinations in the field of encoding/decoding of video.

In this case, if the combined motion information overlaps with other motion information already existing in the merge list, the combined motion information may not be added to the merge list.

Step 4) If the number of pieces of motion information in the merge list is smaller than N, zero vector motion information may be added to the merge list.

The zero vector motion information may be motion information in which a motion vector is a zero vector.

One or more zero vector motion information may be provided. Reference picture indexes of one or more zero vector motion information may be different from each other. For example, the value of the reference picture index of the first zero vector motion information may be 0. The value of the reference picture index of the second zero vector motion information may be 1.

The number of zero vector motion information may be equal to the number of reference pictures in the reference picture list.

The reference direction of the zero vector motion information may be bidirectional. Both motion vectors may be zero vectors. The number of zero vector motion information may be the smaller of the number of reference pictures in the reference picture list L0 and the number of reference pictures in the reference picture list L1. Alternatively, when the number of reference pictures in the reference picture list L0 and the number of reference pictures in the reference picture list L1 are different from each other, a unidirectional reference direction may be used for a reference picture index applicable to only one reference picture list.

The encoding apparatus 100 and/or the decoding apparatus 200 may sequentially add zero vector motion information to the merge list while changing the reference picture index.

If the zero vector motion information overlaps with other motion information already present in the merge list, the zero vector motion information may not be added to the merge list.

The order of steps 1) to 4) described above is merely exemplary, and the order between steps may be interchanged. Also, some of the steps may be omitted according to predefined conditions.

Derivation method of predictive motion vector candidate list in AMVP mode

The maximum number of motion vector predictor candidates in the predictor motion vector candidate list may be predefined. The predefined maximum number is denoted by N. For example, the predefined maximum number may be 2.

Motion information (ie, predicted motion vector candidates) may be added to the predicted motion vector candidate list in the order of steps 1) to 3) below.

Step 1) Among the spatial candidates, available spatial candidates may be added to the predicted motion vector candidate list. Spatial candidates may include a first spatial candidate and a second spatial candidate.

The first spatial candidate may be one of A ₀ , A ₁ , scaled A ₀ and scaled A ₁ . The second spatial candidate may be one of B ₀ , B ₁ , B ₂ , scaled B ₀ , scaled B ₁ , and scaled B ₂ .

Motion information of available spatial candidates may be added to the predicted motion vector candidate list in the order of the first spatial candidate and the second spatial candidate. In this case, when motion information of an available spatial candidate overlaps with other motion information already existing in the motion vector predictor candidate list, the motion information may not be added to the predictor motion vector candidate list. In other words, when the value of N is 2, if the motion information of the second spatial candidate is identical to the motion information of the first spatial candidate, the motion information of the second spatial candidate may not be added to the predicted motion vector candidate list.

A maximum of N pieces of motion information may be added.

Step 2) If the number of pieces of motion information in the predicted motion vector candidate list is smaller than N and a temporal candidate is available, the motion information of the temporal candidate may be added to the predicted motion vector candidate list. In this case, when motion information of an available temporal candidate overlaps with other motion information already existing in the motion vector predictor candidate list, the motion information may not be added to the predictor motion vector candidate list.

Step 3) If the number of pieces of motion information in the predicted motion vector candidate list is smaller than N, zero vector motion information may be added to the predicted motion vector candidate list.

One or more zero vector motion information may be provided. Reference picture indexes of one or more zero vector motion information may be different from each other.

The encoding apparatus 100 and/or the decoding apparatus 200 may sequentially add zero vector motion information to the predictive motion vector candidate list while changing the reference picture index.

When the zero vector motion information overlaps with other motion information already existing in the predicted motion vector candidate list, the zero vector motion information may not be added to the predicted motion vector candidate list.

The description of the zero vector motion information described above for the merge list can also be applied to the zero vector motion information. Redundant descriptions are omitted.

The order of steps 1) to 3) described above is merely exemplary, and the order between steps may be interchanged. Also, some of the steps may be omitted according to predefined conditions.

As shown in FIG. 12, a quantized level may be generated by performing a transform and/or quantization process on the residual signal.

A residual signal may be generated as a difference between an original block and a prediction block. Here, the prediction block may be a block generated by intra prediction or inter prediction.

The residual signal may be converted into the frequency domain through a transform process that is part of the quantization process.

Transformation kernels used for transformation may include various DCT kernels such as Discrete Cosine Transform (DCT) type 2 (DCT-II) and Discrete Sine Transform (DST) kernels. .

These transform kernels may perform a separable transform or a 2D (2D) non-separable transform on the residual signal. The separable transform may be a transform that performs a one-dimensional (1D) transform on the residual signal in each of a horizontal direction and a vertical direction.

DCT type and DST type adaptively used for 1D conversion may include DCT-V, DCT-VIII, DST-I and DST-VII in addition to DCT-II as shown in Table 3 and Table 4 below, respectively. there is.

[Table 3]

[Table 4]

As shown in Tables 3 and 4, a transform set may be used in deriving the DCT type or DST type to be used for transformation. Each transform set may include a plurality of transform candidates. Each transformation candidate may be a DCT type or a DST type.

Table 5 below shows an example of a transform set applied in the horizontal direction and a transform set applied in the vertical direction according to the intra prediction mode.

[Table 5]

In Table 5, vertical transform set numbers and horizontal transform set numbers applied to the horizontal direction of the residual signal are indicated according to the intra prediction mode of the target block.

As illustrated in Table 5, transform sets applied in the horizontal and vertical directions may be predefined according to the intra prediction mode of the target block. The encoding apparatus 100 may perform transform and inverse transform on the residual signal using a transform included in a transform set corresponding to the intra prediction mode of the target block. Also, the decoding apparatus 200 may perform an inverse transform on the residual signal using a transform included in a transform set corresponding to the intra prediction mode of the target block.

In these transforms and inverse transforms, the set of transforms applied to the residual signal may be determined as exemplified in Tables 3, 4, and 5, and may not be signaled. Transformation indication information may be signaled from the encoding device 100 to the decoding device 200. The transform indication information may be information indicating which transform candidate among a plurality of transform candidates included in a transform set applied to the residual signal is used.

For example, when the size of the target block is 64x64 or less, transform sets each having three transforms may be configured according to the intra prediction mode. An optimal transform method can be selected among all 9 multiple transform methods resulting from a combination of three transforms in the horizontal direction and three transforms in the vertical direction. Encoding efficiency can be improved by encoding and/or decoding the residual signal using such an optimal conversion method.

In this case, information on which transform among transforms belonging to the transform set is used for at least one of the vertical transform and the horizontal transform may be entropy encoded and/or decoded. A truncated unary binarization may be used to encode and/or decode this information.

As described above, a method using various transforms may be applied to a residual signal generated by intra prediction or inter prediction.

Transformation may include at least one of a primary transformation and a secondary transformation. A transform coefficient may be generated by performing a primary transform on the residual signal, and a secondary transform coefficient may be generated by performing a secondary transform on the transform coefficient.

A primary transformation may be named primary. Also, the primary transform may be referred to as an adaptive multiple transform (AMT). As described above, AMT may mean that different transforms are applied to each of the 1D directions (ie, the vertical direction and the horizontal direction).

The secondary transform may be a transform for improving the energy concentration of the transform coefficient generated by the primary transform. Like the first-order transformation, the second-order transformation may be a separable transformation or a non-separable transformation. The non-separable transform may be a non-separable secondary transform (NSST).

The primary transformation may be performed using at least one of a plurality of predefined transformation methods. For example, a plurality of predefined transform methods include a Discrete Cosine Transform (DCT), a Discrete Sine Transform (DST), and a Karhunen-Loeve Transform (KLT) based transform. can include

Also, the primary transform may be a transform having various transform types according to a kernel function defining DCT or DST.

For example, the transform type includes: 1) prediction mode of the target block (eg, one of intra prediction and inter prediction), 2) size of the target block, 3) shape of the target block, 4) intra prediction mode of the target block , 5) components of the target block (eg, one of a luma component and a chroma component) and 6) a partition type applied to the target block (eg, Quad Tree (QT), Binary Tree (BT)). ) and one of a ternary tree (TT)).

For example, the primary transform includes transforms such as DCT-2, DCT-5, DCT-7, DST-7, DST-1, DST-8 and DCT-8 according to the transform kernel presented in Table 6 below. can do. In Table 6, various transform types and transform kernel functions for Multiple Transform Selection (MTS) are illustrated.

MTS may mean that a combination of one or more DCT and/or DST transform kernels is selected for horizontal and/or vertical transform of the residual signal.

[Table 6]

In Table 6, i and j may be integer values greater than or equal to 0 and less than or equal to N-1.

A secondary transform may be performed on transform coefficients generated by performing the primary transform.

As in the first-order transform, a set of transforms can be defined in the second-order transform. Methods for deriving and/or determining a set of transforms, such as those described above, may be applied to first-order as well as second-order transforms.

A primary transform and a secondary transform can be determined for a specified object.

For example, the first transform and the second transform may be applied to one or more signal components of a luma component and a chroma component. Whether to apply the primary transform and/or the secondary transform may be determined according to at least one of coding parameters for a target block and/or a neighboring block. For example, whether to apply the first transform and/or the second transform may be determined by the size and/or shape of the target block.

In the encoding device 100 and the decoding device 200, transformation information indicating a transformation method used for a target may be derived by using specified information.

For example, the transformation information may include an index of a transformation to be used for primary transformation and/or secondary transformation. Alternatively, the transform information may indicate that the primary transform and/or the secondary transform are not used.

For example, when the target of the primary transform and the secondary transform is a target block, the transform method(s) applied to the primary transform and/or the secondary transform indicated by transform information may be applied to the target block and/or neighboring blocks. It may be determined according to at least one of the coding parameters for

Alternatively, transformation information indicating a transformation method for a specified target may be signaled from the encoding apparatus 100 to the decoding apparatus 200.

For example, whether or not a primary transform is used for one CU, an index indicating the primary transform, whether or not a secondary transform is used, and an index indicating the secondary transform may be derived as transformation information in the decoding apparatus 200 there is. Alternatively, conversion information indicating whether or not the primary transform is used for one CU, an index indicating the primary transform, whether or not a secondary transform is used, and an index indicating the secondary transform may be signaled.

A quantized transform coefficient (ie, a quantized level) may be generated by performing quantization on a residual signal or a result generated by performing the primary transform and/or the secondary transform.

13 illustrates diagonal scanning according to an example.

14 illustrates horizontal scanning according to an example.

15 illustrates vertical scanning according to an example.

The quantized transform coefficients may be scanned according to at least one of (up-right) diagonal scanning, vertical scanning, and horizontal scanning according to at least one of an intra prediction mode, a block size, and a block shape. A block may be a transform unit.

Each scanning can start at a specified start point and end at a specified end point.

For example, quantized transform coefficients may be changed into a one-dimensional vector form by scanning coefficients of a block using the diagonal scanning of FIG. 13 . Alternatively, the horizontal scanning of FIG. 14 or the vertical scanning of FIG. 15 may be used instead of diagonal scanning according to the block size and/or intra prediction mode.

Vertical scanning may be scanning two-dimensional block form coefficients in a column direction. Horizontal scanning may be scanning two-dimensional block form coefficients in a row direction.

In other words, which of diagonal scanning, vertical scanning, and horizontal scanning is to be used may be determined according to the size of the block and/or the inter-prediction mode.

As shown in FIGS. 13, 14, and 15, quantized transform coefficients may be scanned along a diagonal, horizontal, or vertical direction.

Quantized transform coefficients may be expressed in block form. A block may include a plurality of sub-blocks. Each sub-block may be defined according to a minimum block size or a minimum block shape.

In scanning, a scanning order according to a type or direction of scanning may be applied to subblocks first. Also, a scanning order according to a scanning direction may be applied to quantized transform coefficients in a sub-block.

For example, as shown in FIGS. 13, 14, and 15, when the size of the target block is 8x8, transform coefficients quantized by primary transform, secondary transform, and quantization of the residual signal of the target block are can be created Thereafter, one of three scanning orders may be applied to the four 4x4 subblocks, and quantized transform coefficients may be scanned according to the scanning order for each 4x4 subblock.

The encoding apparatus 100 may generate an entropy-encoded quantized transform coefficient by performing entropy encoding on the scanned quantized transform coefficients, and may generate a bitstream including the entropy-coded quantized transform coefficients. .

The decoding apparatus 200 may extract entropy-encoded quantized transform coefficients from a bitstream and generate quantized transform coefficients by performing entropy decoding on the entropy-coded quantized transform coefficients. Quantized transform coefficients may be arranged in a 2D block form through inverse scanning. At this time, as a reverse scanning method, at least one of (upper right) diagonal scan, vertical scan, and horizontal scan may be performed.

In the decoding apparatus 200, inverse quantization may be performed on quantized transform coefficients. Depending on whether the second-order inverse transform is performed, the second-order inverse transform may be performed on the result generated by performing the inverse quantization. Also, depending on whether the first-order inverse transform is performed, the first-order inverse transform may be performed on the result generated by performing the second-order inverse transform. A reconstructed residual signal may be generated by performing a first-order inverse transform on a result generated by performing a second-order inverse transform.

For a luma component reconstructed through intra prediction or inter prediction, inverse mapping of a dynamic range may be performed before in-loop filtering.

The dynamic range may be divided into 16 equal pieces, and a mapping function for each piece may be signaled. The mapping function may be signaled at the slice level or tile group level.

An inverse mapping function for performing inverse mapping may be derived based on the mapping function.

In-loop filtering, reference picture storage, and motion compensation may be performed in the inversely mapped region.

A prediction block generated through inter prediction may be converted into a mapped region by mapping using a mapping function, and the converted prediction block may be used to generate a reconstructed block. However, since intra prediction is performed in a mapped region, a prediction block generated by intra prediction can be used to generate a reconstructed block without mapping and/or inverse mapping.

For example, when the target block is a residual block of a chroma component, the residual block may be converted into an inversely mapped region by scaling the chroma component of the mapped region.

Whether scaling is available may be signaled at the slice level or tile group level.

For example, scaling can be applied only when mapping for luma components is available, and partitioning of luma components and partitioning of chroma components follow the same tree structure.

Scaling may be performed based on an average of values of samples of a luma prediction block corresponding to the chroma prediction block. In this case, when the target block uses inter prediction, the luma prediction block may mean a mapped luma prediction block.

A value required for scaling may be derived by referring to a look-up table using an index of a piece to which an average of values of samples of the luma prediction block belongs.

By performing scaling on the residual block using the finally derived value, the residual block may be converted into an inversely mapped region. Then, for the chroma component block, reconstruction, intra prediction, inter prediction, in-loop filtering, and reference picture storage may be performed in the inversely mapped region.

For example, information indicating whether mapping and/or inverse mapping of the luma component and chroma component is available may be signaled through a sequence parameter set.

A prediction block of the target block may be generated based on the block vector. A block vector may indicate displacement between a target block and a reference block. A reference block may be a block within a target image.

In this way, a prediction mode in which a prediction block is generated by referring to a target image may be referred to as an intra block copy (IBC) mode.

IBC mode can be applied to CUs of a specified size. For example, IBC mode can be applied to MxN CUs. Here, M and N may be 64 or less.

The IBC mode may include a skip mode, a merge mode, and an AMVP mode. In the case of skip mode or merge mode, a merge candidate list may be constructed, and a merge index may be signaled so that one merge candidate among merge candidates of the merge candidate list may be specified. A block vector of a specified merge candidate may be used as a block vector of a target block.

In case of AMVP mode, a differential block vector may be signaled. Also, the predicted block vector may be derived from the left neighboring block and the top neighboring block of the target block. Also, an index of which neighboring block is to be used may be signaled.

The prediction block of the IBC mode may be included in the target CTU or the left CTU, and may be limited to a block within a pre-constructed region. For example, the value of the block vector may be limited so that the prediction block of the target block is located within a specified region. The specified area may be an area of three 64x64 blocks that are encoded and/or decoded prior to the 64x64 block including the target block. As the value of the block vector is limited in this way, memory consumption and device complexity according to the implementation of the IBC mode can be reduced.

16 is a structural diagram of an encoding device according to an embodiment.

The encoding device 1600 may correspond to the aforementioned encoding device 100.

The encoding device 1600 includes a processing unit 1610, a memory 1630, a user interface (UI) input device 1650, a UI output device 1660, and storage that communicate with each other through a bus 1690. (1640). Also, the encoding device 1600 may further include a communication unit 1620 connected to the network 1699.

The processor 1610 may be a semiconductor device that executes processing instructions stored in a central processing unit (CPU), memory 1630, or storage 1640. The processing unit 1610 may be at least one hardware processor.

The processing unit 1610 may generate and process signals, data, or information input to the encoding device 1600, output from the encoding device 1600, or used inside the encoding device 1600, signals, Inspection, comparison, and judgment related to data or information can be performed. In other words, in an embodiment, data or information generation and processing, and data or information related inspection, comparison, and judgment may be performed by the processing unit 1610 .

The processing unit 1610 includes an inter prediction unit 110, an intra prediction unit 120, a switch 115, a subtractor 125, a transform unit 130, a quantization unit 140, an entropy encoding unit 150, and an inverse quantization unit. It may include a unit 160, an inverse transform unit 170, an adder 175, a filter unit 180, and a reference picture buffer 190.

An inter prediction unit 110, an intra prediction unit 120, a switch 115, a subtractor 125, a transform unit 130, a quantization unit 140, an entropy encoding unit 150, an inverse quantization unit 160, At least some of the inverse transform unit 170, the adder 175, the filter unit 180, and the reference picture buffer 190 may be program modules and may communicate with an external device or system. Program modules may be included in the encoding device 1600 in the form of an operating system, application program modules, and other program modules.

Program modules may be physically stored on various known storage devices. Also, at least some of these program modules may be stored in a remote storage device capable of communicating with the encoding device 1600 .

Program modules include routines, subroutines, programs, objects, components, and data that perform functions or operations according to an embodiment or implement abstract data types according to an embodiment. A data structure, etc. may be encompassed, but is not limited thereto.

Program modules may include instructions or codes executed by at least one processor of the encoding device 1600 .

The processing unit 1610 includes an inter prediction unit 110, an intra prediction unit 120, a switch 115, a subtractor 125, a transform unit 130, a quantization unit 140, an entropy encoding unit 150, and an inverse quantization unit. Instructions or codes of the unit 160, the inverse transform unit 170, the adder 175, the filter unit 180, and the reference picture buffer 190 may be executed.

Storage may represent memory 1630 and/or storage 1640 . Memory 1630 and storage 1640 may be various forms of volatile or non-volatile storage media. For example, the memory 1630 may include at least one of a ROM 1631 and a RAM 1632 .

The storage unit may store data or information used for the operation of the encoding device 1600 . In an embodiment, data or information of the encoding device 1600 may be stored in a storage unit.

For example, the storage unit may store pictures, blocks, lists, motion information, inter prediction information, and bitstreams.

The encoding device 1600 may be implemented in a computer system including a recording medium that can be read by a computer.

The recording medium may store at least one module required for the encoding device 1600 to operate. The memory 1630 may store at least one module, and the at least one module may be configured to be executed by the processing unit 1610 .

A function related to communication of data or information of the encoding device 1600 may be performed through the communication unit 1620 .

For example, the communication unit 1620 may transmit a bitstream to a decoding apparatus 1700 to be described later.

17 is a structural diagram of a decryption device according to an embodiment.

The decoding device 1700 may correspond to the decoding device 200 described above.

The decryption device 1700 includes a processing unit 1710, a memory 1730, a user interface (UI) input device 1750, a UI output device 1760, and storage that communicate with each other through a bus 1790. (1740). Also, the decoding apparatus 1700 may further include a communication unit 1720 connected to the network 1799.

The processor 1710 may be a semiconductor device that executes processing instructions stored in a central processing unit (CPU), memory 1730, or storage 1740. The processing unit 1710 may be at least one hardware processor.

The processing unit 1710 may generate and process signals, data, or information input to the decoding device 1700, output from the decoding device 1700, or used inside the decoding device 1700, and signals, Inspection, comparison, and judgment related to data or information can be performed. In other words, in an embodiment, data or information generation and processing, and data or information related inspection, comparison, and judgment may be performed by the processing unit 1710 .

The processing unit 1710 includes an entropy decoding unit 210, an inverse quantization unit 220, an inverse transform unit 230, an intra prediction unit 240, an inter prediction unit 250, a switch 245, an adder 255, a filter A unit 260 and a reference picture buffer 270 may be included.

An entropy decoding unit 210, an inverse quantization unit 220, an inverse transform unit 230, an intra prediction unit 240, an inter prediction unit 250, a switch 245, an adder 255, a filter unit 260, and At least some of the reference picture buffers 270 may be program modules, and may communicate with an external device or system. Program modules may be included in the decryption device 1700 in the form of an operating system, application program modules, and other program modules.

Program modules may be physically stored on various known storage devices. Also, at least some of these program modules may be stored in a remote storage device capable of communicating with the decryption device 1700 .

The program modules may include instructions or codes executed by at least one processor of the decoding device 1700 .

The processing unit 1710 includes an entropy decoding unit 210, an inverse quantization unit 220, an inverse transform unit 230, an intra prediction unit 240, an inter prediction unit 250, a switch 245, an adder 255, a filter Instructions or codes of unit 260 and reference picture buffer 270 may be executed.

Storage may represent memory 1730 and/or storage 1740 . Memory 1730 and storage 1740 may be various forms of volatile or non-volatile storage media. For example, the memory 1730 may include at least one of a ROM 1731 and a RAM 1732 .

The storage unit may store data or information used for the operation of the decoding device 1700 . In an embodiment, data or information of the decoding apparatus 1700 may be stored in the storage unit.

The decryption device 1700 may be implemented in a computer system including a recording medium that can be read by a computer.

The recording medium may store at least one module required for the decoding apparatus 1700 to operate. The memory 1730 may store at least one module, and the at least one module may be configured to be executed by the processing unit 1710 .

A function related to communication of data or information of the decoding device 1700 may be performed through the communication unit 1720 .

For example, the communication unit 1720 may receive a bitstream from the encoding device 1600.

Hereinafter, the processing unit may represent the processing unit 1610 of the encoding device 1600 and/or the processing unit 1710 of the decoding device 1700. For example, for functions related to prediction, the processing unit may represent switch 115 and/or switch 245. In the function related to inter prediction, the processing unit may represent the inter prediction unit 110, the subtractor 125, and the adder 175, and may represent the inter prediction unit 250 and the adder 255. In the function related to intra prediction, the processing unit may represent the intra prediction unit 120, the subtractor 125, and the adder 175, and may represent the intra prediction unit 240 and the adder 255. In the function related to transform, the processing unit may represent the transform unit 130 and the inverse transform unit 170, and may represent the inverse transform unit 230. In the function related to quantization, the processing unit may represent the quantization unit 140 and the inverse quantization unit 160, and may represent the inverse quantization unit 220. In the function related to entropy encoding and/or decoding, the processing unit may represent the entropy encoding unit 150 and/or the entropy decoding unit 210. In terms of filtering-related functions, the processing unit may represent the filter unit 180 and/or the filter unit 260. For functions related to reference pictures, the processing unit may indicate the reference picture buffer 190 and/or the reference picture buffer 270 .

In intra prediction of a target block, by using directional padding using already decoded reference samples in the vicinity of the target block or by using a combination of reference samples based on predicted positions, the target block Intra prediction may be performed for .

According to the principle of intra prediction, reliability of intra prediction may be lowered as the distance between the reference sample and the prediction location increases.

In addition, since prediction samples are generated according to a predefined constant and simple pattern in intra prediction, it may be difficult to predict various patterns within an image through intra prediction.

MPM may be used to signal information on an intra prediction mode selected by the encoding apparatus 100 from among various intra prediction modes applicable to a target block.

In the encoding/decoding of the intra prediction mode of the target block, the MPM constructs the MPM list so that the intra prediction mode having a high probability of occurrence is included in the MPM list, and a small number of binary It may be a method of allocating bits. When a specific intra prediction mode is used for a target block, binary bits allocated for the specific intra prediction mode may be signaled.

However, since the MPM list is constructed using only intra prediction modes of neighboring blocks of the target block, intra prediction modes that are not suitable for various patterns of the target block may be added to the MPM list. Bitstream overhead may be generated due to this limitation.

Also, a residual signal may be generated by intra prediction, and signaling of information indicating a transform mode of the residual signal may be required. Additional bitstream overhead may occur due to signaling of information indicating a conversion mode.

In embodiments, by inducing a machine learning-based intra prediction mode, intra prediction mode transmission and MPM list configuration can be efficiently performed.

In embodiments, by using a machine learning-based intra prediction model, prediction of various patterns of an image that cannot be performed by normal intra prediction may be performed. Also, in the embodiments, processing may be performed on a reference sample and a prediction sample for performing an intra prediction model.

In embodiments, weights for the various intra prediction modes may be determined by using machine learning, and a weighted sum of prediction samples may be derived. By this weighted sum, intra prediction that can provide additional various prediction patterns compared to basic intra prediction modes can be performed.

In embodiments, a transform mode for a residual signal generated by intra prediction may be predicted using a machine learning-based transform mode derivation model. Through this prediction, signaling for the conversion mode can be efficiently performed.

In embodiments, efficient signaling for intra prediction mode and transform mode may be performed. By such signaling, bitstream overhead can be reduced and encoding performance can be improved.

By performing efficient intra prediction with various patterns, an effect of improving the quality of a decoded image and reducing bitstream overhead can be expected.

The prediction method of the target block and the bitstream generation method of the embodiment may be performed by the encoding device 1600 . An embodiment may be a part of a method of encoding a target block or a method of video encoding.

Prediction may be one of the prediction methods described in the embodiments. For example, prediction may be inter prediction or intra prediction.

In step 1810, the processing unit 1610 may determine prediction information to be used for encoding of the target block.

Prediction information may include information used for prediction described in the embodiments.

For example, the prediction information may include inter prediction information. For example, the prediction information may include intra prediction information.

In step 1820, encoding of the coding information may be performed to generate coded coding information.

Coding information may refer to signaling/encoding/decoding information described in the embodiments. In other words, the coding information may be information used for the decoding apparatus 1700 to perform prediction corresponding to the prediction performed by the encoding apparatus 1600 .

For example, the coding information may include coding parameters mentioned in the embodiments.

At step 1830, processing unit 1610 may generate a bitstream.

A bitstream may include information about a target block. Also, the bitstream may include the information described above in the embodiments.

For example, a bitstream may include coded coding information or coding information.

For example, the bitstream may include coding parameters related to a target block and/or attributes of the target block.

Information included in the bitstream may be generated at step 1820, or at least partially at

steps

1810 and 1820.

The processing unit 1610 may store the generated bitstream in the storage 1640 . Alternatively, the communication unit 1620 may transmit the bitstream to the decoding device 1700.

A bitstream may include coded information about a target block. The processing unit 1610 may generate encoded information on the target block by performing entropy encoding on the information on the target block.

In step 1840, the processing unit 1610 may perform prediction on the target block using information on the target block and prediction information.

The processing unit 1610 may use coding information in predicting a target block. Alternatively, coding information may be generated to correspond to information used in prediction of the target block.

A prediction block may be generated by prediction of the target block. A residual block that is the difference between the target block and the predicted block may be generated. Information on the target block may be generated by applying transform and quantization to the residual block.

Information on the target block may include transform and quantized coefficients of the target block. A reconstructed residual block may be generated by applying inverse quantization and inverse transformation to transform and quantized coefficients of the target block. A reconstructed block that is the sum of the prediction block and the reconstructed residual block may be created.

A method of predicting a target block using the bitstream of the embodiment may be performed by the decoding apparatus 1700. The embodiment may be part of a decoding method of a target block or a video decoding method.

Prediction may be one of the prediction methods described above in embodiments. For example, prediction may be inter prediction or intra prediction.

In step 1910, the communication unit 1720 may obtain a bitstream. The communication unit 1720 may receive a bitstream from the encoding device 1600. The processing unit 1710 may store the obtained bitstream in the storage 1740 .

The processing unit 1710 may read a bitstream from the storage unit 1740 .

A bitstream may include information about a target block.

Information on the target block may include transform and quantized coefficients of the target block.

Also, the bitstream may include the information described above in the embodiments.

A computer-readable recording medium may include a bitstream, and prediction and decoding of a target block may be performed using information on the target block included in the bitstream.

The computer readable recording medium may be a non-transitory computer readable recording medium.

A bitstream may include coded information about a target block. The processing unit 1710 may generate information on the target block by performing entropy decoding on the encoded information on the target block.

In step 1920, the processing unit 1710 may obtain coding information from the bitstream.

The processing unit 1710 may generate coding information by performing decoding on the encoded coding information of the bitstream.

In step 1930, the processing unit 1710 may determine prediction information to be used for decoding the target block.

The prediction information in the decoding device 1700 may be the same as the prediction information in the encoding device 1600. In other words, processing unit 1710 may generate prediction information identical to the prediction information used in step 1840 to make the same prediction as the prediction performed in step 1840 .

Processing unit 1710 may determine prediction information using the methods used in the embodiments.

The processing unit 1710 may determine prediction information of the target block based on information related to a prediction method obtained from the bitstream.

Prediction information may include inter prediction information. Prediction information may include intra prediction information.

In operation 1940, the processing unit 1710 may perform prediction on the target block using information on the target block and prediction information.

The processing unit 1710 may use coding information in predicting the target block. A prediction block may be generated by predicting the target block.

20 is a flowchart of a prediction method according to an embodiment.

Step 1840 may include at least some of

steps

2010, 2020, 2030 and 2040. Step 1940 may include at least some of

steps

2010, 2020, 2030 and 2040.

In embodiments, the processing unit may refer to the processing unit 1610 of the encoding device 1600 and/or the processing unit 1710 of the decoding device 1700.

At step 2010, the processing unit may derive an intra prediction mode of the target block using the coding information.

The processing unit may derive the intra prediction mode of the target block using one or more of 1) an intra prediction mode indicator, 2) MPM, and 3) a machine learning-based intra prediction mode derivation method. A method of deriving an intra prediction mode based on machine learning may refer to a method of deriving an intra prediction mode based on machine learning.

In embodiments, “machine learning” may mean “neural network”. In embodiments, “machine learning” may be replaced with “neural network”.

The intra prediction mode indicator may be information indicating an intra prediction mode of a target block. The intra prediction mode indicator may be a coding parameter.

At step 2020, the processing unit may construct reference samples for intra prediction.

The processor may configure a reference sample for intra prediction using one or more of 1) reference sample selection, 2) reference sample filtering, and 3) reference sample scaling.

In step 2030, the processing unit may perform intra prediction on the target block.

The processing unit may perform intra prediction on the target block using the derived intra prediction mode and the reference sample.

The processing unit uses one or more methods of 1) non-directional intra prediction, 2) directional intra prediction, 3) machine learning-based prediction, 4) prediction sample blending, 5) prediction sample refinement, and 6) prediction sample scaling. Intra prediction may be performed according to the derived intra prediction mode for the block. Here, a reference sample configured for intra prediction of the target block may be used.

At step 2040, the processing unit may derive the transform mode of the target block.

Step 2040 may be performed prior to step 2010 . Alternatively, step 2040 may be performed between steps 2010 and 2020 or between

steps

2020 and 2030 .

대상 블록의 인트라 예측 모드의 유도Derivation of intra prediction mode of target block

Derivation of the intra prediction mode of the target block in step 2010 is described.

In performing encoding/decoding using intra prediction on a target block, an intra prediction mode of intra prediction may be derived.

In deriving the intra prediction mode of the target block, one or more methods of 1) intra prediction mode indicator, 2) MPM, and 3) machine learning-based intra prediction mode derivation may be used.

In embodiments, a target block may mean one or more of a coding block, a prediction block, and a transform block. Alternatively, the target block may mean a sub-block of the aforementioned block.

In embodiments, an intra prediction mode may mean an intra prediction direction such as DC mode, planar mode, and angular mode.

In embodiments, intra prediction mode may refer to non-directional intra prediction such as DC mode and planar mode. The intra prediction mode may mean directional intra prediction like each mode.

In embodiments, each mode may refer to one or more respective modes.

In embodiments, the intra prediction mode is Position Dependent Intra Prediction Combination (PDPC) mode, Cross-Component Linear Model (CCLM) mode, Intra Sub-partitions (Intra Sub-partitions). Partitions (ISP) mode, Multiple Reference Line (MRL) mode, Matrix weighted Intra Prediction (MIP) mode, Intra Block Copy (IBC) mode, and Machine Learning: ML) mode, etc. may mean prediction modes including intra prediction for the target block.

The intra prediction mode may be PDPC mode, CCLM mode, ISP mode, MRL mode, MIP mode, IBC mode or ML mode.

Here, the ML mode may include one or more of machine learning-based prediction, prediction sample blending, prediction sample improvement, and prediction sample scaling.

In embodiments, the size of the target block may be WxH.

W may be a positive integer. The value of W may be 2 ⁿ . Here, n may be an integer of 1 or greater.

H may be a positive integer. The value of H may be 2 ^m . Here, m may be an integer of 1 or greater.

인트라 예측 모드 지시자를 이용하는 대상 블록의 인트라 예측 모드의 유도Derivation of the intra prediction mode of the target block using the intra prediction mode indicator

In step 2010, the processing unit may derive an intra prediction mode of the target block using the intra prediction mode indicator.

The intra prediction mode indicator may indicate whether a specific intra prediction mode is performed.

An intra prediction mode indicator for the target block may be encoded/decoded/signaled.

The intra prediction mode indicator may be an indicator for the aforementioned intra prediction mode.

In embodiments, multiple intra prediction modes may be used. Intra prediction mode indicators for a plurality of intra prediction modes may be respectively used.

A specific intra prediction mode indicator may be used as an indicator for a specific intra prediction mode.

For example, the intra prediction mode indicator may include an intra prediction direction indicator.

For example, the intra prediction mode indicator may include a PDPC indicator, CCLM indicator, ISP indicator, MRL indicator, MIP indicator, IBC indicator, and ML mode indicator.

A value of 1 for the specific intra prediction mode indicator may indicate that a specific intra prediction mode is used in prediction of the target block.

A value of 0 for the specific intra prediction mode indicator may indicate that the specific intra prediction mode is not used in prediction of the target block.

For example, if the value of the ML mode indicator is '1', the ML mode may be derived as an intra prediction mode of the target block.

A specific intra prediction mode indicator may indicate N specific intra prediction modes. N may be an integer greater than or equal to 1.

For example, if the value of the ML mode indicator is '1', at least one of the N ML modes may be used as an intra prediction mode for the target block. N may be an integer greater than or equal to 1. Here, the additional intra prediction mode indicator may indicate intra prediction modes applied to the target block among N intra prediction modes specified by the specific intra prediction mode indicator.

For example, additional intra prediction modes may be intra prediction directions. The additional intra prediction mode indicator may indicate an intra prediction direction used for intra prediction of a target block among available intra prediction directions.

For example, additional intra prediction modes may be detailed methods in intra prediction mode. Additional intra prediction modes may indicate values of coding parameters used in the intra prediction mode.

The processing unit may determine an intra-prediction mode for a target block and/or an intra-prediction mode coding parameter using a specific intra-prediction mode indicator and an additional intra-prediction mode indicator. The processing unit may perform prediction on the target block using a specific intra prediction mode indicator and an additional intra prediction mode indicator.

MPM을 이용하는 대상 블록의 인트라 예측 모드의 유도Derivation of intra prediction mode of target block using MPM

At step 2010, the processing unit may use the MPM in deriving the intra prediction mode of the target block.

An MPM list with N candidate modes may be configured to use MPM. N may be a positive integer. A candidate mode may be MPM.

Candidate modes of the MPM list may be intra prediction modes. Alternatively, the candidate modes of the MPM list may be intra prediction directions.

Candidate modes included in the MPM list may be configured based on intra prediction modes of neighboring blocks adjacent to the target block. Here, the neighboring block may include a neighboring block temporally adjacent to the target block and a neighboring block spatially adjacent to the target block.

The number of candidate modes included in the MPM list may be determined based on one or more coding parameters. For example, the number of candidate modes may be determined by one or more of a slice type, a color component, a size of a target block, and a shape of the target block.

The MPM list may be plural. In other words, a plurality of MPM lists may be used to perform intra prediction on a target block.

In one embodiment, intra prediction modes in a plurality of MPM lists may not overlap each other. In other words, an intra prediction mode included in one MPM list among a plurality of MPM lists may not be included in other MPM list(s).

For example, for each intra prediction mode, it may be determined which MPM list among a plurality of MPM lists the intra prediction mode is to be included in. Alternatively, a plurality of MPM lists may be configured such that an intra prediction mode already included in one MPM list is not included in another MPM list.

The number of candidate modes of the plurality of MPM lists may be different from each other.

For example, two MPM lists may be configured for a target block. The two MPM lists may be a first MPM list and a second MPM list. The MPM list may include a first MPM list and a second MPM list.

For example, the first MPM list may include only intra prediction modes using machine learning as candidate modes. The intra prediction mode using machine learning may indicate a method of deriving an intra prediction mode based on machine learning.

For example, the second MPM list may include only intra prediction modes that do not use machine learning as candidate modes.

In constructing the MPM list, an intra prediction mode of a neighboring block of a target block may be replaced with another intra prediction mode. Other permuted intra prediction modes can be added to the MPM list as candidate modes.

For example, when an intra prediction mode of a neighboring block is an ML mode, the ML mode may be replaced with a DC mode, and the DC mode may be included in the MPM list.

For example, when an intra prediction mode of a neighboring block is an ML mode, the ML mode may be replaced with a planner mode, and the planner mode may be included in the MPM list.

For example, when the intra prediction mode of a neighboring block is the DC mode, the DC mode may be replaced with the ML mode, and the ML mode may be included in the MPM list.

For example, when an intra prediction mode of a neighboring block is a planner mode, the planner mode may be replaced with an ML mode, and the ML mode may be included in the MPM list.

For example, when an intra prediction mode of a neighboring block is an angular mode, each mode may be replaced with an ML mode, and the ML mode may be included in the MPM list.

Here, a look-up table (LUT) defining which intra mode each intra prediction mode is replaced with may be used. The LUT may be information indicating a permutation relationship between intra prediction modes. The encoding device 1600 and the decoding device 1700 may construct an LUT.

기계 학습에 기반하는 인트라 예측 모드 유도 방법Intra prediction mode derivation method based on machine learning

In step 2010, the processing unit may derive an intra prediction mode of the target block using a machine learning-based intra prediction mode derivation method.

Machine learning-based intra prediction mode derivation methods include Single Layer Perceptron (SLP), Multi Layer Perceptron (MLP), Deep Neural Network (DNN), Convolutional Neural Network (CNN), and It may be performed using one or more methods such as a Recurrent Neural Network (RNN).

Here, the input of the machine learning-based intra prediction mode derivation method may include one or more coding parameters.

For example, the input of the machine learning-based intra prediction mode derivation method is a reference sample, an MPM list, a quantization parameter (QP), a slice type, a color component, a size of a target block, a shape of the target block, and a segmentation of the target block. It may be one or more of the block partitioning structures.

Here, the output of the machine learning-based intra prediction mode derivation method may be one or more probabilities for one or more intra prediction modes. In other words, the machine learning-based intra prediction mode derivation method may derive an intra prediction mode probability for each intra prediction mode among a plurality of intra prediction modes.

For example, the output of the machine learning-based intra prediction mode derivation method may include probabilities for DC mode, planar mode, and angular mode.

For example, the output of the machine learning-based intra prediction mode derivation method may include a probability for the MIP mode.

For example, the output of the machine learning-based intra prediction mode derivation method may include a probability for the ML mode.

Here, the intra prediction mode of the target block may be derived using probabilities for the aforementioned intra prediction modes.

For example, one or more intra prediction modes having a probability greater than a threshold among available intra prediction modes may be derived as the intra prediction mode of the target block.

For example, intra prediction modes having a probability greater than a threshold may be derived as intra prediction modes of the target block.

For example, N intra prediction modes having the highest probabilities among available intra prediction modes may be derived as intra prediction modes of the target block. N may be a positive integer.

For example, N intra prediction modes selected in descending order of probabilities may be derived as intra prediction modes of the target block. N may be a positive integer.

In embodiments, the selected intra prediction modes may be termed candidate intra prediction modes or intermediate intra prediction modes. Alternatively, the derived intra prediction modes may be named candidate intra prediction modes or intermediate intra prediction modes.

Here, when there are a plurality of derived intra prediction modes, an additional intra prediction mode indicator may be signaled. The additional intra prediction mode indicator may indicate an intra prediction mode of a target block among derived intra prediction modes. An intra prediction mode indicated by an additional intra prediction mode indicator among selected intra prediction modes satisfying the above conditions or derived according to satisfying the above conditions is derived as an intra prediction mode of the target block and/or can be determined

An MPM list of a target block may be constructed using a probability for an intra prediction mode.

For example, an MPM list may be constructed using an intra prediction mode having a higher probability than a threshold.

For example, an MPM list may be constructed using the N intra prediction modes with the highest probabilities.

For example, an MPM list may be constructed using N intra prediction modes selected in descending order of probabilities.

복수의 참조 샘플 라인들Multiple reference sample lines

The reference sample line may correspond to the reference sample line described above with reference to FIG. 8 .

A reference sample line may include one or more reconstructed reference samples.

The plurality of reference sample lines may include reference sample line 0, reference sample line 1, reference sample line 2, and reference sample line 3.

The size of the target block may be W×H, and the upper left coordinates of the target block may be regarded as (0, 0).

Reference sample line 0 may be adjacent to the target block. Reference sample line 1 may be adjacent to reference sample line 0. Reference sample line 2 may be adjacent to reference sample line 1. Reference sample line 3 may be adjacent to reference sample line 2.

Reference sample line 0 may include an upper left reference sample, upper reference samples, and left reference samples. Coordinates of the upper left reference sample may be (-1, -1). The top reference samples may be a horizontal reference sample line adjacent to the right of the top left reference sample. The y coordinates of the top reference samples may be -1. The x-coordinates of the top reference samples may be greater than or equal to zero. The left reference samples may be vertical reference sample lines adjacent to the lower end of the upper left reference sample. The x coordinates of the left reference samples may be -1. The y coordinates of the top reference samples may be greater than or equal to 0.

In other words, the reference sample line 0 may be a set of reference samples having an x coordinate greater than or equal to -1, a y coordinate greater than or equal to -1, and an x coordinate or y coordinate equal to or greater than -1.

Reference sample line 1 may include an upper left reference sample, upper reference samples, and left reference samples. Coordinates of the upper left reference sample may be (-2, -2). The top reference samples may be a horizontal reference sample line adjacent to the right of the top left reference sample. The y coordinates of the top reference samples may be -2. The x-coordinates of the top reference samples may be greater than or equal to -1. The left reference samples may be vertical reference sample lines adjacent to the lower end of the upper left reference sample. The x coordinates of the left reference samples may be -2. The y coordinates of the top reference samples may be greater than or equal to -1.

In other words, the reference sample line 1 may be a set of reference samples having an x coordinate greater than -2, a y coordinate greater than -2, and either an x coordinate or a y coordinate equal to -2.

Reference sample line 2 may include an upper left reference sample, upper reference samples, and left reference samples. Coordinates of the upper left reference sample may be (-3, -3). The top reference samples may be a horizontal reference sample line adjacent to the right of the top left reference sample. The y coordinates of the top reference samples may be -3. The x-coordinates of the top reference samples may be greater than or equal to -2. The left reference samples may be vertical reference sample lines adjacent to the lower end of the upper left reference sample. The x coordinates of the left reference samples may be -3. The y coordinates of the top reference samples may be greater than or equal to -2.

In other words, the reference sample line 2 may be a set of reference samples having an x coordinate greater than or equal to -3, a y coordinate greater than or equal to -3, and an x coordinate or y coordinate equal to or greater than -3.

Reference sample line 3 may include an upper left reference sample, upper reference samples, and left reference samples. Coordinates of the upper left reference sample may be (-4, -4). The top reference samples may be a horizontal reference sample line adjacent to the right of the top left reference sample. The y coordinates of the top reference samples may be -4. The x coordinates of the top reference samples may be greater than or equal to -3. The left reference samples may be vertical reference sample lines adjacent to the lower end of the upper left reference sample. The x coordinates of the left reference samples may be -4. The y coordinates of the top reference samples may be greater than or equal to -3.

In other words, the reference sample line 3 may be a set of reference samples having an x coordinate greater than -4, a y coordinate greater than -4, and an x coordinate or y coordinate equal to -4.

참조 샘플 영역의 구성Configuration of the reference sample area

The x-coordinate of the reference sample in the reference sample area may be greater than or equal to -a and less than or equal to W-1+c. The y coordinate of the reference sample may be greater than or equal to -d and less than or equal to H-1+b. At least one of the x coordinate and the y coordinate of the reference sample may be equal to or less than -1.

In other words, the reference sample region may include 1) a first rectangle diagonally adjacent to the upper left corner of the target block, 2) a second rectangle adjacent to the right side of the first rectangle, and 3) a third rectangle adjacent to the lower end of the first rectangle. there is. The size of the first rectangle may be axd. The size of the second rectangle may be (W+c)×d. The size of the third rectangle may be a×(H+b).

The reference sample area may include an upper reference sample area and a left reference sample area.

The x-coordinate of the reference sample in the upper reference sample area may be greater than or equal to -a and less than or equal to W-1+c. The y coordinate of the reference sample may be greater than or equal to -d and less than or equal to -1.

The x-coordinate of the reference sample of the left reference sample area may be greater than or equal to -a and less than or equal to -1. The y coordinate of the reference sample may be greater than or equal to 0 and less than or equal to H-1+b.

In other words, the upper reference sample region may include 1) a first rectangle diagonally adjacent to the upper left corner of the target block and 2) a second rectangle adjacent to the right side of the first rectangle. The left reference sample area may be a third rectangle adjacent to a lower end of the first rectangle. The size of the first rectangle may be axd. The size of the second rectangle may be (W+c)×d. The size of the third rectangle may be a×(H+b).

The reference sample area may include a left reference sample area and an upper reference sample area.

The x-coordinate of the reference sample of the left reference sample area may be greater than or equal to -a and less than or equal to -1. The y coordinate of the reference sample may be greater than or equal to -d and less than or equal to H-1+b.

The x-coordinate of the reference sample in the upper reference sample area may be greater than or equal to 0 and less than or equal to W-1+c. The y coordinate of the reference sample may be greater than or equal to -d and less than or equal to -1.

In other words, the left reference sample area may include 1) a first rectangle diagonally adjacent to the upper left corner of the target block and 2) a second rectangle adjacent to the lower end of the first rectangle. The top reference sample area may be a third rectangle adjacent to the right side of the first rectangle. The size of the first rectangle may be a×d. The size of the second rectangle may be a×(H+b). The size of the third rectangle may be (W+c)×d.

The reference sample area may include an upper left reference sample area, an upper reference sample area, and a left reference sample area.

The x-coordinate of the reference sample in the upper left reference sample area may be greater than or equal to -a and less than or equal to -1. The y coordinate of the reference sample may be greater than or equal to -d and less than or equal to -1.

In other words, the upper left reference sample area may be a first rectangle diagonally adjacent to the upper left corner of the target block. The upper reference area may be a second rectangle adjacent to the right side of the first rectangle. The left reference sample area may be a third rectangle adjacent to a lower end of the first rectangle. The size of the first rectangle may be a×d. The size of the second rectangle may be (W+c)×d. The size of the third rectangle may be a×(H+b).

In the embodiments described above with reference to FIGS. 22 and 25 , a may be an integer greater than or equal to 0. d may be an integer greater than or equal to 0. b may be an integer greater than or equal to 0. b may be an integer greater than or equal to H. c may be an integer greater than or equal to 0. c may be an integer greater than or equal to W.

In the embodiments described above with reference to FIGS. 22 and 25 , a, b, c, and d may be size information indicating sizes of a plurality of reference sample regions.

Size information may be encoded/decoded/signaled for a specific unit. For example, a specific unit includes a Sequence Parameter Set (SPS), an Adaptive Parameter Set (APS), a Picture Parameter Set (PPS), a picture header, a slice header, It may be one or more of a Coding Tree Unit (CTU) and a Coding Unit (CU).

In the embodiments described above with reference to FIGS. 22 and 25 , each of a, b, c, and d may be determined based on one or more coding parameters. For example, each of a, b, c, and d may be determined based on one or more of an intra prediction mode of a target block, a size of the target block, and a shape of the target block.

In the embodiments described above with reference to FIGS. 22 and 25 , the reference sample area may include one or more sub-reference sample areas. The one or more sub reference sample regions may be one or more of a left upper reference sample region, an upper reference sample region, and a left reference sample region. In other words, each of the upper left reference sample area, the upper reference sample area, and the left reference sample area may be regarded as a sub reference sample area.

참조 샘플의 구성Composition of the reference sample

At step 2020, the processing unit may construct reference samples for intra prediction of the target block based on the derived intra prediction mode for the target block.

The reference sample may be constructed using one or more reconstructed samples or a combination of samples around the target block. Hereinafter, a reference sample may mean a reconstructed reference sample.

In embodiments, samples around the target block may refer to samples within a reference sample line and may refer to samples within a reference region.

In embodiments, samples adjacent to the target block may refer to samples adjacent to the target block.

In embodiments, the samples adjacent to the target block may mean samples whose distance from the target block is less than or equal to a threshold value. The distance from the target block may mean a larger value among 1) horizontal distance, 2) vertical distance, or 3) horizontal distance and vertical distance.

In constructing the reference sample for the target block, one or more methods of reference sample selection, reference sample filtering, and reference sample scaling may be used.

In constructing a reference sample for the target block, the reference sample may be selected using one or more reconstructed reference sample lines adjacent to the target block.

For example, one reference sample line among the plurality of reference sample lines of FIG. 21 may be selected. Reference sample(s) of one reference sample line among the plurality of reference sample lines of FIG. 21 may be selected.

For example, two or more reference sample lines among the plurality of reference sample lines of FIG. 21 may be selected. Reference samples of two or more reference sample lines among the plurality of reference sample lines of FIG. 21 may be selected.

In constructing a reference sample for the target block, the reference sample may be selected using one or more reconstructed reference sample regions adjacent to the target block.

For example, as shown in FIG. 22, one area may be configured as a reference sample area.

For example, as shown in FIGS. 23 and 24 , two regions may be configured as reference sample regions. The two regions may include an upper reference sample region and an upper left reference sample region.

For example, as shown in FIG. 25, three regions may be configured as reference sample regions. The three areas may include an upper left reference sample area, an upper reference sample area, and an upper left reference sample area.

For example, a reference sample of one reference sample area among a plurality of reference sample areas may be selected.

For example, reference samples of two or more reference sample regions among a plurality of reference sample regions may be selected.

참조 샘플을 구성하기 위한 정보의 시그널링Signaling of information to construct a reference sample

In order to configure a reference sample for a target block, reference sample configuration information may be signaled.

The reference sample configuration information may indicate a method for configuring a reference sample of a target block among the methods described above in relation to configuration of a reference sample. A method for configuring the reference sample of the target block may be selected according to the reference sample configuration information.

For example, the reference sample configuration information may include reference sample line number information. Reference sample line number information may indicate whether two or more reference sample lines are selected to determine reference samples for a target block.

For example, the reference sample configuration information may include reference sample line index information. Reference sample line index information may indicate an index indicating a reference sample line selected to determine reference samples for a target block. When two or more reference sample lines are selected to determine reference samples for a target block, two or more pieces of reference sample configuration information may be provided.

For example, the reference sample configuration information may include reference sample region number information. Reference sample region number information may indicate whether two or more reference sample regions are selected to determine reference samples for a target block.

For example, the reference sample configuration information may include reference sample area information.

The reference sample area information may indicate which reference sample area is used among the reference sample areas described with reference to FIGS. 22 to 25 .

For example, the reference sample area information may indicate which sub reference sample areas are used among the upper left reference sample area, the upper reference sample area, and the left reference sample area described with reference to FIGS. 22 to 25 .

For example, the reference sample configuration information may include reference sample region selection information. The reference sample region selection information may indicate a sub-reference sample region selected to determine a reference sample for a target block.

For example, the reference sample region selection information indicates a sub-reference sample region selected to determine a reference sample for a target block among sub-reference sample regions in the reference sample region described with reference to FIGS. 23, 24, or 25 can do.

참조 샘플의 구성 및 재구축된 샘플의 가용성에 대한 판단Determination of composition of reference sample and availability of reconstructed sample

In constructing the reference sample for the target block, the reference sample for the target block may be selected based on one or more coding parameters. For example, a reference sample for a target block may be selected based on at least one of a size of the target block, a shape of the target block, and an intra prediction mode of the target block.

In constructing the reference sample, it may be determined whether a reconstructed target reference sample around the target block is available. In other words, the availability of target samples around the target block can be determined.

Hereinafter, a target sample may refer to a sample for which availability is determined. The target reference sample line may refer to a reference sample line including the target sample.

When the target sample is located within a specific region including the target block, it may be determined that the target sample is unavailable. A target sample determined to be available may be used as a reference sample. Available samples may refer to samples determined to be available.

When the target sample is located outside a specific region including the target block, it may be determined that the target sample is unavailable. An unavailable sample may refer to a sample determined to be unavailable.

In embodiments, a specific region may be one or more of a target picture, a subpicture, a slice, a tile, a Virtual Pipeline Data Unit (VPDU), and a Coding Tree Unit (CTU).

When the target sample is determined to be an unavailable sample, other available reconstructed samples around the unavailable target sample may be used as a reference sample in place of the non-available target sample. Alternatively, sample values of other available reconstructed samples may be assigned to values of non-available target samples. Alternatively, the sample values of other available reconstructed samples may be considered as the values of non-available target samples.

For example, as shown in FIG. 26 , when the target sample is unavailable, a first replacement sample in the target reference sample line may replace the target sample as the reference sample.

Here, the first replacement sample may be an available sample adjacent to the target sample in a specific direction.

For example, the specific direction may be a direction further away from the target block. Alternatively, the specific direction may be a direction closer to the target block.

For example, the specific direction may be clockwise. Alternatively, the specific direction may be counterclockwise.

When a sample adjacent in a specific direction does not exist or when a sample adjacent in a specific direction is unavailable, the first replacement sample may be an available sample adjacent to the target sample in a direction opposite to the specific direction.

For example, as shown in FIG. 27 , when the target sample is unavailable, a second replacement sample in an adjacent reference sample line may replace the target sample as the reference sample. The adjacent reference sample line may be a sample line adjacent to the target reference sample line. The second replacement sample may be adjacent to the subject sample. In other words, the second replacement sample may be a first adjacent reference sample adjacent to the sample of interest, within the adjacent reference sample line.

For example, the target reference sample line may be between an adjacent reference sample line and the target block. In other words, an adjacent reference sample line may be a reference sample line further away from the target block than the target reference sample line. Alternatively, the adjacent reference sample line may be between the target reference sample line and the target block. In other words, the adjacent reference sample line may be a reference sample line closer to the target block than the target reference sample line.

If a sample adjacent to the target block and within an adjacent reference sample line is unavailable, the second replacement sample may be a sample within another adjacent reference sample line. The second replacement sample may be adjacent to the subject sample. In other words, the second replacement sample may be a second adjacent reference sample adjacent to the target sample, in another adjacent reference sample line.

Here, the other adjacent reference sample line may be another reference sample line adjacent to the target reference sample line other than the aforementioned adjacent reference sample line. That is to say, the target reference sample line may be between adjacent reference sample lines and other adjacent reference samples.

Here, the target sample may be between the first adjacent reference sample and the second adjacent reference sample.

In one embodiment, the second replacement sample may include a first adjacent reference sample in an adjacent reference sample line and a second adjacent reference sample in another adjacent reference sample line. The second replacement sample may be a weighted sum of a first adjacent reference sample in an adjacent reference sample line and a second adjacent reference sample in another adjacent reference sample line.

In one embodiment, a weighted sum of the first replacement sample and the second replacement sample may replace the non-available target sample as a reference sample.

참조 샘플에 대한 필터링Filtering on reference samples

In step 2020, a reference sample may be selected for prediction of the target block. Filtering may be applied to the selected reference sample.

In one embodiment, reference samples of a plurality of reference sample lines may be used as is (without filtering).

In one embodiment, filtering between samples within the same reference sample line may be used.

For example, samples may include a reference sample. Filtering using samples may be applied to the reference pixel.

For example, a reference sample may be constructed using samples to which filtering is applied.

In one embodiment, filtering between samples in different reference sample lines may be used.

In embodiments, filtering may be selectively applied. It may be determined whether filtering is to be applied to the samples based on one or more coding parameters. For example, whether to apply filtering to samples may be determined based on one or more of an intra prediction mode of the target block, a size of the target block, and a shape of the target block.

When filtering is applied, a filter type of a filter to be used for filtering may be selected from among a plurality of available filter types. A filter type of a filter to be used for filtering may be selected based on one or more coding parameters. For example, a filter type of a filter to be used for filtering may be selected based on one or more of the intra prediction mode of the target block, the size of the target block, and the shape of the target block.

For example, one or more of whether filtering is applied and a filter type may be selected based on whether the ML mode is performed for intra prediction of the target block. In other words, the filter type of the filter may be changed based on whether the ML mode is performed for intra prediction of the target block.

참조 샘플에 대한 스케일링Scaling for reference sample

In step 2020, scaling may be performed on one or more reference samples selected from the reference sample line or reference sample region.

A scaled reference sample may be created by scaling. The size and/or shape of the scaled reference sample may be determined based on one or more coding parameters. For example, the size and/or shape of the reference sample scaled by scaling may be determined based on one or more of an intra prediction mode of a block, a size of a block, a shape of a block, a slice type, and a color component. Here, the block may be a target block, a reference block, or a reference sample region.

Whether scaling is applied to one or more reference samples may be determined based on one or more coding parameters. For example, whether scaling is applied to one or more reference samples may be determined based on one or more of an intra prediction mode of a block, a block size, and a block shape. Here, the block may be a target block, a reference block, or a reference sample region.

In the process of performing intra prediction, scaling of reconstructed samples around the target block may be performed.

참조 샘플에 대한 다운-스케일링Down-scaling to reference sample

As shown in FIG. 28, down-scaling of reference samples may be performed using sub-sampling.

One reference sample may be selected from among a plurality of adjacent reference samples in a reference sample line by sub-sampling.

The sampling interval may be the number of a plurality of reference samples used to select one reference sample.

As shown in FIG. 29 , second down-scaling of the reference samples may be performed using sub-sampling.

One reference sample may be derived based on a plurality of adjacent reference samples in a reference sample line by sub-sampling. One reference sample may be derived by a calculation equation using a plurality of adjacent reference samples within a reference sample line. For example, the calculation formula may be a weighted sum of a plurality of reference samples.

The sampling interval may be the number of a plurality of reference samples used to derive one reference sample.

In embodiments, a sampling interval of sub-sampling may be determined based on one or more coding parameters. For example, the sampling interval may be determined based on one or more of an intra prediction mode of a block, a block size, a block shape, a slice type, and a color component. Here, the block may be a target block, a reference block, or a reference sample region.

참조 샘플에 대한 업-스케일링Up-scaling to reference sample

As shown in FIG. 30, up-scaling using an image interpolation filter may be performed on reference samples.

A plurality of up-scaled reference samples may be generated by an image interpolation filter using one or more reference samples as input. The number of generated reference samples may be greater than the number of one or more reference samples used as input.

The image interpolation filter may be a Discrete Cosine Transform (DCT) based interpolation filter, a Discrete Sine Transform (DST) based interpolation filter, a bi-linear interpolation filter, or a cubic filter. there is.

The size and/or shape of the up-scaled reference samples generated by up-scaling may be the same as the size and/or shape of the target block.

The size and/or shape of up-scaled reference samples generated by up-scaling may be the same as the size and/or shape of reference samples required to perform intra prediction on the target block.

In up-scaling, an overlapped region may occur between up-scaled reference samples generated through up-scaling of one or more reference samples.

A reference region may mean a reference sample or a plurality of reference samples. For example, the upper reference region may refer to a plurality of reference samples located at the upper end of the target block.

By applying up-scaling to the reference region, a reference region to which up-scaling is applied may be generated. A reference region to which up-scaling is applied may mean a plurality of reference regions.

For example, an up-scaled reference region may be created by applying up-scaling to an upper reference region. Each row of up-scaled reference regions may be identical to the top reference region.

참조 샘플과 관련된 처리Processing related to reference samples

Scaling of the reference sample of the target block may be performed based on one or more coding parameters. Here, the scaling may be one of the scalings described above in the embodiments. For example, one of the scalings described above in embodiments may be selected based on one or more coding parameters.

For example, scaling of a reference sample of a target block may be selected and/or configured based on one or more of a block size, a block shape, an intra prediction mode of the block, a slice type, and a color component. Here, the block may be a target block, a reference block, or a reference sample region.

Configuration of scaling may include selecting one of the scalings described above in embodiments.

When reference samples are configured according to embodiments, normalization may be performed on the configured reference samples.

For example, a maximum bit value among bit values of reference samples may be derived. The bit value of each reference sample of the reference samples may be divided by the maximum bit value derived. The bit value of the reference sample can be reconstructed by dividing the bit value of the reference sample by the maximum bit value.

For example, an average bit value of reference samples may be derived. An average bit value derived from the bit value of each reference sample of the reference samples may be subtracted. The bit value of the reference sample can be reconstructed by subtracting the average bit value from the bit value of the reference sample.

인트라 예측의 수행Perform intra prediction

In step 2030, the processing unit performs 1) non-directional intra prediction, 2) directional intra prediction, 3) machine learning-based intra prediction, and 4) prediction samples based on the intra prediction mode and the reference sample derived from the above-described embodiments. Intra prediction of the target block may be performed using one or more of blending, 5) prediction sample improvement, and 6) prediction sample scaling.

In performing intra prediction on a target block, non-directional intra prediction may be used. In embodiments, non-directional intra prediction may be one or more of DC mode, planner mode, PDPC mode, CCLM mode, IBC mode and MIP mode.

The DC mode may perform intra prediction on the target block using an average value of one or more reference samples among configured reference samples. In this case, filtering may be applied to one or more target samples located at the boundary of the target block.

In embodiments, the target sample may mean a sample that is a target of intra prediction within the target block.

The prediction method in DC mode may be performed differently based on one or more coding parameters. For example, the prediction method in the DC mode may be performed differently based on the size and/or shape of the target block.

The planner mode may perform intra prediction on the target block using a weighted sum of one or more reference samples determined according to the position of the target sample in the target block. Here, for each reference sample of one or more reference samples, a weight of the reference sample in the weighted sum may be determined based on a distance between the reference sample and the target sample.

The PDPC mode may perform prediction by updating a prediction sample using the position of the prediction sample and a reference sample. Here, the prediction sample may be a sample generated by one or more of prediction using non-directional mode, prediction using directional mode, prediction using ML mode, prediction sample blending, prediction sample improvement, and prediction sample scaling.

The CCLM mode may perform prediction on a target block using a model for a relationship between color components of reference samples.

The IBC mode can perform prediction on a target block by copying a part of an already reconstructed image in the target image to the target block.

The MIP mode may perform prediction on a target block using a matrix multiplication of a reference sample and a matrix. The matrix may be a predefined matrix.

In performing intra prediction on a target block, directional intra prediction may be used.

A directional prediction mode of directional intra prediction may be performed using one or more directional modes among a horizontal mode, a vertical mode, and an angular mode having a specific angle.

For example, the horizontal mode may perform prediction on a target block using one or more reference samples existing in a horizontal line to which the target sample belongs.

For example, the vertical mode may perform prediction on a target block using one or more reference samples existing in a vertical line to which the target sample belongs.

For example, each mode may perform prediction on a target block using one or more reference samples existing within a specific line to which the target sample belongs or existing around the specific line. Here, the specific line may be a line determined according to the angle of each mode.

In embodiments, the number of reference samples used for intra prediction of a target block may be N or more. N may be an integer of 2 or greater. For example, the number of reference samples may be one of 2, 3, 4, 5, and 6.

In embodiments, a filter used for filtering a target block may be an N-tap filter. N may be an integer of 2 or greater. For example, at least one of a 2-tap filter, a 3-tap filter, a 4-tap filter, a 5-tap filter, and a 6-tap filter may be used to filter the target block.

ML 모드를 사용하는 인트라 예측Intra prediction using ML mode

At step 2030, the processing unit may perform intra prediction on the target block using machine learning based prediction. Here, machine learning-based prediction may mean prediction using an ML mode.

A machine learning based intra prediction model for performing the ML mode can be constructed using one or more methods such as single-layer perceptron, multi-layer perceptron, deep neural network, convolutional neural network, and recurrent neural network.

In embodiments, "model" may mean "method", "structure" or "configuration". That is to say, "model", "method", "structure" and "configuration" may be substituted for each other in the embodiments.

Here, the input of the ML mode may include one or more coding parameters and may include one or more pieces of information described in the embodiments.

For example, an input of the ML mode may be one or more of a reference sample, an MPM list, a QP, a slice type, a color component, a size of a target block, a shape of the target block, and a division structure of the target block.

For example, the input of the ML mode may include N reference sample lines. N may be a positive integer.

In one embodiment, a reference sample line at a specific location may be used as an input to the ML mode.

In one embodiment, the reference sample line used as input to the ML mode may be selected in terms of Rate-Distortion Optimization (RDO).

In one embodiment, a reference sample line indicator for specifying a reference sample line used as an input of the ML mode may be determined and signaled. The reference sample line indicator may indicate a reference sample line used as an input of the ML mode among a plurality of reference sample lines.

For example, the input of the ML mode may be N reference sample regions. N may be a positive integer.

In one embodiment, a reference sample region at a specific location may be used as an input to the ML mode.

In one embodiment, the reference sample region to be used as an input for ML mode may be selected in terms of RDO.

In one embodiment, a reference sample region indicator for specifying a reference sample region used as an input of the ML mode may be determined and signaled. The reference sample area indicator may indicate a reference sample area used as an input of the ML mode among a plurality of reference sample areas.

For example, in the ML mode, when a plurality of reference sample lines are input, a sum of reference samples of the plurality of reference sample lines may be generated, and the sum of the reference samples may be input to a machine learning-based prediction model.

For example, in the ML mode, when a plurality of reference sample regions are input, a sum of reference samples of the plurality of reference sample regions may be generated, and the sum of the reference samples may be input to a machine learning-based prediction model.

For example, in the ML mode, when a plurality of reference sample lines are input, a concatenation of reference samples of the plurality of reference sample lines may be generated, and the concatenation of the reference samples may be input to a machine learning-based prediction model. can

For example, in the ML mode, when a plurality of reference sample regions are input, a chain of reference samples of the plurality of reference sample regions may be generated, and the chain of reference samples may be input to a machine learning-based prediction model.

For example, in the ML mode, when a plurality of reference sample lines are input, feature information for each reference sample line may be extracted inside the machine learning-based predictive model, and each reference of the reference sample line Feature information may be added to the sample.

For example, in the ML mode, when a plurality of reference sample areas are input, feature information for each reference sample area may be extracted from within a machine learning-based predictive model, and feature information is provided for each reference sample in the reference sample area. may be added.

The ML mode may consist of N machine learning-based intra prediction models. N may be a positive integer.

Each of the N machine learning-based intra prediction models may generate one or more prediction blocks.

For example, the ML mode may consist of a single machine learning-based intra prediction model. A machine learning-based intra prediction model can generate one prediction block.

For example, ML mode may consist of two or more machine learning-based intra prediction models. Structures of two or more machine learning-based intra prediction models may be the same. Here, by setting values of corresponding parameters of the two or more machine learning-based intra prediction models differently, different prediction blocks may be generated by the two or more machine learning-based intra prediction models.

For example, ML mode may consist of two or more machine learning-based intra prediction models. Structures of two or more machine learning-based intra prediction models may be different from each other.

The machine learning-based intra prediction model constituting the aforementioned ML mode may have various characteristics. Below, the characteristics of the machine learning-based intra prediction model constituting the ML mode are described. For example, the structure of a machine learning-based intra prediction model can be constructed using a combination of properties described in the embodiments.

ML 모드의 입력Input in ML mode

Embodiments of the input of the machine learning-based intra prediction model are described below.

For example, as shown in FIG. 31 , an input of a machine learning-based intra prediction model may be one reference sample line or one reference sample region. Alternatively, an input of the machine learning-based intra prediction model may be a reference sample of one reference sample line or one reference sample region.

A machine learning-based intra prediction model may include one or more layers. An input of the machine learning-based intra prediction model may be input to the top layer. A prediction block may be output from the lowest layer.

For example, as shown in FIG. 32, the input of the machine learning-based intra prediction model is 1) N reference sample lines, 2) N reference sample regions, or 3) N reference sample lines and N A combination of reference sample regions may be included. N may be a positive integer.

One reference sample may be used in each reference sample line of the N reference sample lines. Alternatively, one reference sample may be used in each reference sample region of the N reference sample regions.

A machine learning-based intra prediction model may include one or more layers. Inputs of the machine learning-based intra prediction model may be input to N top layers. A prediction block may be output from the lowest layer. N may be a positive integer.

As shown in FIG. 33 , an internal layer of the machine learning-based intra prediction model may be divided into M branch layers. M may be a positive integer.

Also, two or more branch layers among the divided M branch layers may be merged into one branch layer.

34 illustrates division into M branches in one layer according to an example.

M prediction blocks may be generated in one machine learning-based intra prediction model. M may be a positive integer.

As shown in FIG. 34 , division into M branches may be generated in one layer inside the machine learning-based intra prediction model. M may be a positive integer.

M prediction blocks may be generated from M branches, respectively.

As shown in FIG. 34 , divisions into two branches may be generated in each layer of (M−1) layers inside the machine learning-based intra prediction model. In other words, each layer of (M-1) layers among a plurality of internal layers of the machine learning-based intra prediction model may be divided into two branches.

Here, one of the two branches may be another (or next) layer, and the other branch may be a prediction block. Among (M-1) layers, layer m may be divided into a branch of layer m+1 and a branch of prediction block m. m may be an integer greater than or equal to 1 and less than or equal to M-1.

M prediction blocks may be generated from M branches, respectively. M-1 prediction blocks may be generated from a branch of prediction block 1 to a branch of prediction block M-1, and one prediction block may be generated from the lowest layer of the machine learning-based intra prediction model.

As shown in FIG. 35 , divisions into two or more branches may be generated in each layer of (M−1) layers inside the machine learning-based intra prediction model. In other words, each layer of (M-1) layers among a plurality of internal layers of the machine learning-based intra prediction model may be divided into two or more branches.

Here, two or more branches may be different (or next) layers or prediction blocks. Among (M-1) layers, layer m may be divided into a branch of layer m+1 and a branch of prediction block m. m may be an integer greater than or equal to 1 and less than or equal to M-1.

N prediction blocks may be generated from each of the N branches. N or less prediction blocks may be generated from a branch of prediction block 1 to a branch of prediction block N, and N prediction blocks may be generated from the lowest layer of the machine learning-based intra prediction model. N may be a positive integer greater than or equal to 1.

In embodiments, a layer may refer to a layer using one or more methods of a single-layer perceptron, a multi-layer perceptron, a deep neural network, a convolutional neural network, and a recurrent neural network, one or more layers, or a combination of a plurality of layers.

기계 학습 기반 인트라 예측 모델의 구성Construction of a machine learning-based intra prediction model

The machine learning-based intra prediction model and the number of machine learning-based intra prediction models constituting the ML mode may be configured based on coding parameters.

For example, the number of machine learning-based intra prediction models and machine learning-based intra prediction models is configured based on the size of the target block, the shape of the target block, the division structure of the target block, color components, slice types, QP and MPM lists, etc. It can be.

For example, if the size, width (W) and/or height (H) of the target block is smaller than a certain threshold, the ML mode may be configured to use a machine learning-based intra prediction model using a multi-layer perceptron. If the size, width and/or height of the target block is greater than a certain threshold, the ML mode can be configured to use a machine learning-based intra prediction model using a convolutional neural network.

For example, if the size, width (W) and/or height (H) of the target block is smaller than a certain threshold, the ML mode may be configured to use N machine learning-based intra prediction models. N may be a positive integer.

For example, it may be configured to use M machine learning intra prediction models when the size, width and/or height of the target block is greater than a specific threshold. M may be a positive integer. The value of N and the value of M may be different from each other.

The machine learning-based intra prediction mode information may indicate whether the machine learning-based intra prediction mode is performed.

The information on the number of intra prediction models may indicate the number of machine learning-based intra prediction models.

The intra prediction model structure information may indicate a machine learning based intra prediction model structure.

The intra prediction model parameter information may represent parameters used in a machine learning based intra prediction model. Intra prediction model parameter information may be plural.

One or more of machine learning deception intra prediction mode information, intra prediction model number information, intra prediction model structure information, and intra prediction model parameter information may be encoded/decoded/signaled for a specific unit.

A specific unit may be one or more of SPS, PPS, APS, PPS, picture header, slice header, CTU, and CU.

예측 샘플 블렌딩Predictive sample blending

In step 2030, prediction sample blending may be used in performing intra prediction on the target block. The prediction sample may be a plurality of prediction samples generated by one or more methods of non-directional prediction, directional prediction, machine learning-based prediction, prediction sample improvement, and prediction sample scaling. Also, a weight for each prediction sample of a plurality of prediction samples may be calculated using a weight induction model based on machine learning.

Here, the prediction samples used to perform prediction sample blending may be a plurality of prediction samples generated by prediction using one or more prediction modes of non-directional prediction, directional prediction, MIP mode, MRL mode, and ML mode. .

Here, the weight induction model for performing prediction sample blending may be constructed using one or more of a single-layer perceptron, a multi-layer perceptron, a deep neural network, a convolutional neural network, and a recurrent neural network.

Here, the input of the weight derivation model may include one or more coding parameters and one or more pieces of information described in the embodiments. For example, the input of the weight derivation model may include one or more of a reference sample, a prediction sample, an MPM list, a QP, a slice type, a color component, a size of a target block, a shape of the target block, and a partition structure of the target block. .

To perform predictive sample blending, the weight derivation model may output weights for predictive sample blending. Prediction sample blending may be performed using the weights obtained from the weight derivation model.

A weight induction model for performing predictive sample blending may output one weight for a prediction block to which predictive sample blending is applied. In other words, one weight per prediction block can be output.

Here, one output weight may be shared among prediction samples within a prediction block. In other words, one output weight may be applied to all prediction samples in a prediction block.

Normalization may be applied to the weights obtained from the weight derivation model. A plurality of weights may be adjusted such that the sum of the plurality of weights used for prediction sample blending becomes 1 by normalization.

A weight induction model for performing prediction sample blending may output one weight for each prediction sample of a plurality of prediction samples in a prediction block. In other words, one weight per prediction sample can be output.

Here, one output weight may be used by one prediction sample. In other words, one output weight may be applied to one prediction sample in a prediction block.

In addition, normalization may be applied to weights corresponding to the same position of the prediction block among weights obtained from the weight induction model. A plurality of weights may be adjusted such that the sum of the plurality of weights used for prediction sample blending becomes 1 by normalization.

예측 샘플 블렌딩의 수행Perform predictive sample blending

In embodiments, weights of prediction samples may be derived by a weight derivation model.

In one embodiment, prediction sample blending may be performed using only prediction samples having a weight greater than a certain threshold.

In embodiments, prediction sample blending may be performed using N prediction samples selected in descending order of weights of the prediction samples. Alternatively, prediction sample blending may be performed using prediction samples corresponding to the N largest weights. Here, weights may be generated by a weight induction model.

N may be prediction sample number information.

Information on the number of prediction information may be encoded/decoded/signaled for a specific unit. A specific unit may be one or more of SPS, PPS, APS, PPS, picture header, slice header, CTU, and CU.

The predictive sample blending indicator may indicate whether predictive sample blending is performed.

The predictive sample blending indicator may be encoded/decoded/signaled for a specific unit. A specific unit may be one or more of SPS, PPS, APS, PPS, picture header, slice header, CTU, and CU.

The predictive sample blending indicator may be encoded/decoded/signaled when the value of the preceding indicator is a specific value. The specific value can be 1. Here, a value of 1 may mean that a target indicated by a specific preceding indicator is executed. Alternatively, a value of 1 may mean that processing using a target indicated by a specific preceding indicator is applied.

The preceding indicator may be an indicator that is first encoded/decoded/signaled before encoding/decoding/signaling of the predictive sample blending indicator is performed.

For example, the preceding indicator may be an intra prediction mode use indicator. The intra prediction mode use indicator may indicate whether the intra prediction mode is performed.

For example, the preceding indicator may be an intra prediction mode indicator or a lower indicator included in the intra prediction mode indicator.

For example, the preceding indicator may be a MIP indicator. The MIP indicator may indicate whether MIP is performed.

For example, the preceding indicator may be a planner mode indicator. The planner mode indicator may indicate whether the planner mode is performed.

For example, the preceding indicator may be an MPM indicator. The MPM indicator may indicate whether MPM is performed.

For example, the preceding indicator may be an additional intra prediction mode indicator.

For example, the predictive sample blending indicator may be encoded/decoded/signaled when the MIP mode indicator indicates that the MIP mode is used.

For example, the predictive sample blending indicator may be encoded/decoded/signaled when the MPM indicator indicates that MPM is used.

For example, the predictive sample blending indicator may be encoded/decoded/signaled when indicating that the planar mode indicator planar mode is used.

For example, the prediction sample blending indicator may be encoded/decoded/signaled when the intra prediction mode use indicator indicates that the intra prediction mode is used.

예측 샘플 개선Prediction sample improvement

In step 2030, intra prediction of the target block may be performed by performing prediction sample improvement. Prediction sample improvement may refer to improvement on prediction samples.

Prediction samples may be generated by one or more methods of non-directional prediction, directional prediction, machine learning-based prediction, prediction sample blending, and prediction sample scaling.

In addition, a prediction sample improvement model based on machine learning may be used for prediction sample improvement.

The predictive sample improvement model may be constructed using one or more methods of a single-layer perceptron, a multi-layer perceptron, a deep neural network, a convolutional neural network, and a recurrent neural network.

An improved prediction sample may be generated by improving the prediction sample for the prediction sample. In one embodiment, the improved prediction sample may be the final prediction sample for the target block.

The input of the predictive sample improvement model may include one or more coding parameters and may include one or more pieces of information described in the embodiments.

For example, the input of the prediction sample improvement model is one of prediction sample, reference sample line, reference sample area, MPM list, QP, slice type, color component, size of the target block, shape of the target block, and partition structure of the target block. may be ideal

For example, the input of the prediction sample improvement model may include prediction samples generated by prediction using the ML mode and reference samples surrounding the target block.

For example, inputs to the predictive sample improvement model may include prediction samples generated by prediction using the IBC mode and reference samples surrounding the target block.

A predictive sample improvement model may be determined based on one or more coding parameters. For example, the prediction sample improvement model may be determined based on one or more of the size of the target block, the shape of the target block, the division structure of the target block, a color component, a slice type, and a QP and MPM list.

Whether predictive sample improvement is performed may be determined based on one or more coding parameters. For example, whether prediction sample improvement is performed may be determined based on one or more of a size of a target block, a shape of the target block, a partition structure of the target block, a color component, a slice type, and a QP and MPM list.

For example, when the size, width (W), and/or height (H) of the target block are smaller than a specific threshold, prediction sample improvement may be performed using a machine learning-based prediction sample model using a multi-layer perceptron.

For example, when the size, width (W), and/or height (H) of the target block are greater than a specific threshold, prediction sample improvement may be performed using a machine learning-based prediction sample model using a convolutional neural network.

For example, prediction sample improvement may be performed when the size, width (W), and/or height (H) of the target block are smaller than a specific threshold.

For example, prediction sample improvement may be performed when the size, width (W), and/or height (H) of the target block are greater than a specific threshold.

The prediction sample improvement indicator may indicate whether prediction sample improvement is performed.

The prediction sample improvement indicator may be encoded/decoded/signaled for a specific unit. A specific unit may be one or more of SPS, PPS, APS, PPS, picture header, slice header, CTU, and CU.

예측 샘플 스케일링Prediction Sample Scaling

In operation 2030, prediction sample scaling may be used in performing the intra prediction mode for the target block. Prediction sample scaling may mean scaling of prediction samples. Scaled prediction samples may be generated by scaling the prediction samples.

The size and/or shape of the scaled prediction sample may be determined based on one or more coding parameters. For example, the size and/or shape of the scaled prediction sample may be determined based on one or more of an intra prediction mode of a target block, a size of the target block, a shape of the target block, a slice type, and a color component.

Whether predictive sample scaling is performed may be determined based on one or more coding parameters. For example, whether prediction sample scaling is performed may be determined based on one or more of an intra prediction mode of a target block, a size of the target block, and a shape of the target block.

For example, down-scaling of the prediction sample may be performed by using sub-sampling of the final prediction sample of the target block.

A sampling interval for down-scaling may be determined based on one or more coding parameters. For example, the down-scaling sampling interval may be determined based on one or more of an intra prediction mode of a target block, a size of the target block, a shape of the target block, a slice type, and a color component.

For example, up-scaling may be performed by using an image interpolation filter for the final predicted sample of the target block.

The image interpolation filter may be a DCT-based interpolation filter, a DST-based interpolation filter, a bi-linear interpolation filter, or a cubic filter.

변환 모드 유도Conversion mode derivation

In operation 2040, a transform mode of a target mode may be derived while encoding/decoding using intra prediction is performed on the target block.

The transform mode may be a transform mode applied to residual samples of the original sample of the target block and the prediction sample of the prediction block.

The transform mode may include at least one of a first transform, a first inverse transform, a secondary transform, and a secondary inverse transform.

Derivation of the transformation mode may be performed using a machine learning-based transformation mode prediction model.

The transform mode prediction model may be constructed using one or more methods of single-layer perceptron, multi-layer perceptron, deep neural network, convolutional neural network, and recurrent neural network.

Here, the first transform and the first inverse transform may be performed using one or more of a plurality of predefined transform methods.

For example, the plurality of predefined transform methods may include DCT, DST, Karhunen-Loeve Transform (KLT), and the like.

Here, a secondary transform may be performed on transform coefficients generated by performing the primary transform. A first-order inverse-transform may be performed on transform coefficients generated by performing a second-order inverse-transform.

In one embodiment, the input of the transform mode prediction model may include one or more coding parameters. For example, the input to the transform mode prediction model may include one or more of prediction samples, reference samples, MPM lists, QPs, slice types, color components, size of the target block, shape of the target block, and partition structure of the target block. there is.

In an embodiment, an output of the transform mode prediction model may be a probability of at least one transform mode among predefined transform modes.

For example, the output of a transform mode prediction model based on machine learning can be DCT-2 transform mode, DCT-5 transform mode, DCT-8 transform mode, DST-1 transform mode, DST-4 transform mode, and DST-7 transform mode. Can include probabilities for modes. Here, the probability of the transformation mode in the vertical direction and the probability of the transformation mode in the horizontal direction may be derived respectively.

For example, an output of a transform mode prediction model based on machine learning may include a probability for a predefined secondary transform mode in a secondary transform.

For example, the output of the machine learning-based transform mode prediction model may include probabilities for predefined transform modes in the first transform and the second transform.

Here, the output of the transform mode prediction model may include a probability for a transform mode set including N transform modes. N may be a positive integer.

For example, the output of a transformation mode prediction model based on machine learning may include probabilities of transformation mode sets for horizontal and vertical directions.

For example, an output of a transform mode prediction model based on machine learning may be a probability of a transform mode set including N transform modes among secondary transform modes. N may be an integer greater than or equal to 1.

The intra prediction mode of the target block may be derived using the above-described probability of the transformation mode and/or probability of the transformation mode set.

For example, when the transform mode prediction model outputs probabilities of a plurality of transform modes, transform of the target block may be performed using a transform mode having the highest probability among the output probabilities.

For example, when the transform mode prediction model outputs probabilities of a plurality of transform mode sets, transform of the target block may be performed using a transform mode set having the highest probability among the output probabilities.

Transformation mode index information may indicate a transformation mode used for transformation of a target block among transformation modes of a transformation mode set. Transformation mode index information may be encoded/decoded/signaled.

A conversion mode transmission table for a target block may be configured using the conversion mode probability and/or the conversion mode set probability.

For example, when the transform mode prediction model outputs probabilities of transform modes, the output transform modes may be sorted in descending order of probabilities of transform modes. Sorting can create re-ordered transformation modes.

Transformation mode selection information may indicate a transformation mode used for a target block. For example, the transform mode selection information may indicate an order or index of a transform mode used for a target block among transform modes or re-ordered transform modes.

Transformation mode selection information may be encoded/decoded/signaled.

A transformation mode used for a target block may be derived through an RDO process.

For example, when the transform mode prediction model outputs probabilities of transform mode sets, the output transform mode sets may be sorted in descending order of probabilities of the mode sets. Sorting may result in re-ordered sets of transformation modes.

The transform mode set selection information may indicate a transform set mode used for a target block. For example, the transform mode set selection information may indicate the order or index of a transform mode set used for a target block among transform modes or re-ordered transform mode sets.

Transformation mode set selection information may be encoded/decoded/signaled.

Here, transform mode selection information indicating a transform mode applied to the target block among transform modes of the selected transform mode set may be additionally encoded/decoded/signaled.

블록의 크기에 따른 기계 학습을 사용하는 예측Prediction using machine learning based on block size

In embodiments, prediction using machine learning may refer to prediction using a neural network. In embodiments, “machine learning” and “neural network” may be substituted for each other.

Prediction for a block of a specific size may be excluded from the machine learning-based intra prediction mode of the embodiments. In other words, if the size of the target block is a specific size, the machine learning-based intra prediction mode may not be used and/or applied. For example, a specific size may be 32×32. As another example, if the size of the target block is greater than or equal to a specific size, a machine learning-based intra prediction mode may be used and/or applied. For example, a specific size may be 8x8.

Thus, a certain number of neural networks may be used for the machine learning-based intra prediction mode. For example, the number of neural networks may be 7.

Neural networks may be respectively used for prediction of target blocks of different sizes.

For example, S may be a set of sizes of the target block, and S may be {4×4, 8×4, 16×4, 32×4, 8×8, 16×8, 16×16} .

In embodiments, f _m,n (., θ _m,n ) may represent a neural network predicting m×n blocks. For example, f _16,16 (., θ _16,16 ) may indicate a neural network predicting 16×16 blocks.

Parameters of f _m,n (., θ _m,n ) may be gathered within θ _m,n . For example, parameters of f _16,16 (., θ _16,16 ) may be collected within θ _16,16 .

Prediction for blocks of a specific size described above may be performed using intra prediction modes for other sizes.

For example, prediction of a target block having a size of 32×32 may be performed through f _16,16 (., θ _16,16 ). Prediction of a target block having a size of 32×32 may be performed by steps 1) and 2) below. Thus, a particular neural network can be used for two or more different sizes of blocks.

Step 1) The context surrounding a given 32×32 block may be down-sampled in the horizontal and vertical directions before being input to f _16,16 (., θ _16,16 ). In other words, the down-sampled context can be input as f _16,16 (., θ _16,16 ). A context may be an extended set of decoded reference samples. A context may include decoded reference samples in one or more specific regions. The sampling interval of down-sampling may be 2.

Step 2) Prediction samples generated by prediction using machine learning may be interpolated by 2 in the horizontal direction and in the horizontal direction.

The above description of the 32×32 block, the description of step 1) and the description of step 2) can also be applied to the 64×64 block. However, for a 64×64 block, down-sampling factors in the horizontal and vertical directions may be 4.

For example, prediction of a target block having a size of 128×128 may be performed through f _16,16 (., θ _16,16 ) using the above-described embodiments. However, for a 128×128 block, down-sampling factors in the horizontal and vertical directions may be 8.

f _16,16 (., θ _16,16 ) may have characteristics of a fully-connected architecture. This feature may be the same as other neural networks in intra prediction mode based on machine learning. Pruning can be applied to θ _16,16 .

In embodiments, an interval and/or a factor during down-sampling may be a pre-defined positive integer for each block size. Pre-set positive integers different from each other according to block sizes may be used as intervals and/or factors during down-sampling, respectively. In addition, up-sampling and/or interpolation may be performed according to the interval and/or factor of the down-sampling. In other words, the interval and/or factor of up-sampling and/or interpolation may be determined based on the interval and/or factor of down-sampling.

A plurality of different types may be used for information used in machine learning based intra prediction mode.

For example, in embodiments, 16-bit integers may be used for model weights and intermediate results of a neural network in machine learning based intra prediction mode. A 32-bit integer may be used for internal computations of the neural network.

As another example, 16-bit integers may be used for both the model weights and intermediate results of the neural network and internal operations of the neural network.

인트라 예측 모드들로의 학습된 매핑(learned mapping)을 갖는 기계 학습 기반 인트라 예측 모드Machine learning based intra prediction mode with learned mapping to intra prediction modes

For a target block predicted by the learned mapping using the machine learning-based intra prediction mode, the machine learning-based intra prediction mode is conventional intra prediction of existing coding methods such as Versatile Video Coding (VVC) and the like. can be expressed by mode. An existing intra prediction mode of an existing coding method may be an intra prediction direction. For example, the target block may be a luma coding block.

An index for such an existing intra prediction mode may be involved in generating an MPM list of a neighboring block that is a neighbor of a target block. A neighboring block may be a luma coding block. In other words, the machine learning-based intra prediction mode for the target block may correspond to the existing intra prediction mode, and an MPM list of neighboring blocks of the target block may be generated using the corresponding intra prediction mode. Alternatively, an existing intra prediction mode may be determined by a machine learning-based intra prediction mode for the target block, and an MPM list of neighboring blocks of the target block may be generated using the determined intra prediction mode.

Here, neural networks used in the machine learning-based intra prediction mode may use 16-bit signed integer parameters and 16-bit signed intermediate representations.

In FIG. 39, Y may be a target block having a size of w×h. Y can be predicted by machine learning based intra prediction mode.

In FIG. 40, X may represent a context for a target block.

X may consist of 1) n _a rows of decoded samples at the top of Y and 2) n _l columns of decoded samples to the left of Y. Alternatively, X may be constructed using 1) n _a rows of decoded samples on top of Y and 2) n _l columns of decoded samples to the left of Y.

First, as shown in FIG. 40, preprocessing may be performed on X. Context preprocessed by performing preprocessing on X

can be created.

Next, f _h,w (., θ _h,w ) to return 3 outputs

can take

may be input as f _h,w (., θ _h,w ). the three outputs are

, repIdx and grpIdx.

Here, f _h,w (., θ _h,w ) may represent a dedicated neural network for blocks having a size of w×h. Parameters f _h ,w can be determined by θ _h ,w.

Here, the parameter precision of the parameters may be a 16-bit signed integer.

As shown in Figure 40, the first output

Postprocessing may be performed on

Prediction of Y by performing post-processing on

can be created.

may be a prediction sample, prediction block or prediction signal of Y.

The second output may gather the two indices grpIdx ₁ and grpIdx ₂ . Or, the second output may include the two indices grpIdx ₁ and grpIdx ₂ .

grpIdx _i may indicate an index characterizing a Low-Frequency Non-Separable Transform (LFNST) kernel index, and whether primary transform coefficients are transposed. can indicate

Here, the main transform coefficients may be generated by applying DCT-2 in the horizontal direction and applying DCT-2 in the vertical direction to the residual of machine learning intra prediction.

Here, lfnstIdx may be i. i can be either 1 or 2. lfnstIdx may be an index pointing to LFNST.

The third output may be the index repIdx. repIdx may be an index indicating an existing intra prediction mode of an existing coding method such as VVC.

For example, an existing intra prediction mode may be an intra prediction direction. repIdx may indicate one of planner mode, DC mode, or each intra prediction direction.

A prediction of Y may be generated from decoded reference samples surrounding Y by the intra prediction mode indicated by repIdx.

Prediction blocks generated by the intra prediction mode pointed to by repIdx are best

can represent In other words, by the intra prediction mode (or intra prediction direction) indicated by repIdx among a plurality of available intra prediction modes (or a plurality of intra prediction directions)

Prediction may be performed that produces a prediction block that best fits (or is closest to).

When generating the MPM list of the target block, a left block adjacent to the left of the target block may be predicted based on a machine learning-based intra prediction mode. The index for the machine learning-based intra prediction mode may be replaced with repIdx. The index for the machine learning-based intra prediction mode may indicate a machine learning-based intra prediction mode used for prediction of a left block among a plurality of available machine learning-based intra prediction modes. repIdx can be derived through prediction of the left block. repIdx can be added to the MPM list as MPM. Here, MPM may be a candidate index indicating a machine learning-based intra prediction mode.

When generating the MPM list of the target block, an upper block adjacent to the upper end of the target block may be predicted based on a machine learning-based intra prediction mode. The index for the machine learning-based intra prediction mode may be replaced with repIdx. The index for the machine learning-based intra prediction mode may indicate a machine learning-based intra prediction mode used for prediction of an upper block among a plurality of available machine learning-based intra prediction modes. repIdx can be derived through prediction of the upper block. repIdx can be added to the MPM list as MPM. Here, MPM may be a candidate index indicating a machine learning-based intra prediction mode.

In embodiments, each of the target block, neighboring block, top block, and left block may be a luma coding block.

In embodiments, the neural network used for the machine learning based intra prediction mode may include fully-connected layers.

Different layers may be selected and used according to block sizes. For example, four cascaded fully-connected layers can be used for a neural network that performs prediction on 16×16 blocks. Three concatenated fully-connected layers can be used for a neural network that performs prediction on blocks of sizes other than 16×16.

The above embodiments may be performed in the same method and/or a corresponding method in the encoding device 1600 and the decoding device 1700. Also, a combination of one or more of the above embodiments may be used in encoding and/or decoding an image.

The order in which the above embodiments are applied may be different in the encoding device 1600 and the decoding device 1700. Alternatively, the order in which the above embodiments are applied may be the same (at least partially) in the encoding device 1600 and the decoding device 1700.

The above embodiments may be performed for each of the luma signal and chroma signal. The above embodiments may perform the same with respect to the luma signal and the chroma signal.

The shape of a block to which the above embodiments are applied may have a square shape or a non-square shape.

Whether to apply and/or perform at least one of the above embodiments may be determined based on a condition for a block size. In other words, at least one of the above embodiments may be applied and/or performed when a condition for a block size is satisfied. The condition may include a minimum block size and a maximum block size. The block may be one of the blocks described above in the embodiments and the units described above in the embodiments. A block to which the minimum block size is applied and a block to which the maximum block size is applied may be different.

For example, when the size of a block is equal to or greater than the minimum size and/or when the size of a block is equal to or less than the maximum size, the above-described embodiment may be applied and/or performed. When the size of the block is greater than the minimum size and/or when the size of the block is less than or equal to the maximum size, the above-described embodiment may be applied and/or performed.

For example, the above-described embodiment may be applied only when the size of a block is a predefined block size. The predefined block size may be 2x2, 4x4, 8x8, 16x16, 32x32, 64x64 or 128x128. The predefined block size may be (2*SIZE _X )x(2*SIZE _Y ). SIZE _X can be one of 1 or more integers. SIZE _Y can be one of 1 or more integers.

For example, the above-described embodiment may be applied only when the size of a block is greater than or equal to the minimum block size. The above-described embodiment can be applied only when the size of a block is greater than the minimum block size. Block minimum size can be 2x2, 4x4, 8x8, 16x16, 32x32, 64x64 or 128x128. Alternatively, the minimum block size may be (2*SIZE _{MIN_X} )x(2*SIZE _{MIN_Y} ). SIZE _{MIN_X} can be one of 1 or more integers. SIZE _{MIN_Y} can be one of 1 or more integers.

For example, the above-described embodiment may be applied only when the size of a block is less than or equal to the maximum block size. The above-described embodiment can be applied only when the size of a block is smaller than the maximum block size. The maximum block size can be 2x2, 4x4, 8x8, 16x16, 32x32, 64x64 or 128x128. Alternatively, the maximum block size may be (2*SIZE _{MAX_X} )x(2*SIZE _{MAX_Y} ). SIZE _{MAX_X} can be one of 1 or more integers. SIZE _{MAX_Y} can be one of 1 or more integers.

For example, the above-described embodiment may be applied only when the size of a block is greater than or equal to the minimum block size and less than or equal to the maximum block size. The foregoing embodiment can be applied only when the block size is greater than the minimum block size and less than or equal to the maximum block size. The above-described embodiment can be applied only when the size of a block is greater than or equal to the minimum block size and smaller than the maximum block size. The above-described embodiment can be applied only when the block size is larger than the minimum block size and smaller than the maximum block size.

In the above-described embodiments, the size of a block may mean a horizontal size of a block or a vertical size of a block. The size of a block can mean both the horizontal size of the block and the vertical size of the block. Also, the size of a block may mean the area of the block. Each of the area, minimum block size, and maximum block size may be one of integers greater than or equal to 1. Also, the size of a block may mean a result (or value) of a known formula using the horizontal size and vertical size of a block or a result (or value) of a formula in an embodiment.

Also, in the above embodiments, the first embodiment may be applied to the first size, and the second embodiment may be applied to the second size.

The above embodiments may be applied according to a temporal layer. A separate identifier may be signaled to identify a temporal layer to which the above embodiments are applicable, and the above embodiments may be applied to a temporal layer specified by the identifier. The identifier herein may be defined as the lowest layer and/or the highest layer to which the above embodiment is applicable, or may be defined to indicate a specific layer to which the above embodiment is applied. In addition, a fixed temporal layer to which the above embodiment is applied may be defined.

For example, the above embodiments may be applied only when the temporal layer of the target image is the lowest layer. For example, the above embodiments may be applied only when the temporal layer identifier of the target image is 1 or more. For example, the above embodiments may be applied only when the temporal layer of the target image is the highest layer.

A slice type or tile group type to which the above embodiments are applied may be defined, and the above embodiments may be applied according to the corresponding slice type or tile group type.

In the above-described embodiments, in applying a specified process to a specified subject, a specified condition may be required, and if it is described that the specified process is processed under a specified decision, the specified coding parameter If it is described that whether or not a specified condition is satisfied based on a specific coding parameter or a specified decision is made based on a specified coding parameter, it can be interpreted that the specified coding parameter can be replaced with another coding parameter. That is to say, a coding parameter that affects a specified condition or a specified decision may be regarded as exemplary only, and it will be understood that a combination of one or more other coding parameters in addition to the specified coding parameter plays a role of the above specified coding parameter. can

In the foregoing embodiments, the methods are described on the basis of a flow chart as a series of steps or units, but the present invention is not limited to the order of steps, and some steps may occur in a different order or concurrently with other steps as described above. can In addition, those skilled in the art will understand that the steps shown in the flow chart are not exclusive, that other steps may be included, or that one or more steps of the flow chart may be deleted without affecting the scope of the present invention. You will understand.

The embodiments described above include examples of various aspects. While not all possible combinations for representing the various aspects may be described, those skilled in the art will recognize that other combinations than those explicitly described are possible. Accordingly, it is intended that the present invention cover all other substitutions, modifications and variations falling within the scope of the following claims.

Embodiments according to the present invention described above may be implemented in the form of program instructions that can be executed through various computer components and recorded on a computer-readable recording medium. The computer readable recording medium may include program instructions, data files, data structures, etc. alone or in combination. Program instructions recorded on the computer-readable recording medium may be specially designed and configured for the present invention, or may be known and usable to those skilled in the art of computer software.

A computer-readable recording medium may contain information used in embodiments according to the present invention. For example, a computer-readable recording medium may include a bitstream, and the bitstream may include information described in embodiments according to the present invention.

A bitstream may include computer-executable code and/or programs. Computer-executable codes and/or programs may include information described in the embodiments and may include syntax elements described in the embodiments. In other words, the information and syntax elements described in the actual examples may be regarded as computer-executable code in a bitstream, and may be regarded as at least part of a computer-executable code and/or program represented in a bitstream. The recording medium may include a non-transitory computer-readable medium.

Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks and magnetic tapes, optical recording media such as CD-ROMs and DVDs, and magneto-optical media such as floptical disks. media), and hardware devices specially configured to store and execute program instructions, such as ROM, RAM, flash memory, and the like. Examples of program instructions include high-level language codes that can be executed by a computer using an interpreter or the like as well as machine language codes such as those produced by a compiler. The above hardware devices may be configured to act as one or more software modules to perform processing according to the present invention and vice versa.

In the above, the present invention has been described by specific details such as specific components and limited embodiments and drawings, but these are provided only to help a more general understanding of the present invention, and the present invention is not limited to the above embodiments. No, and those skilled in the art to which the present invention pertains may seek various modifications and variations from these descriptions.

Therefore, the spirit of the present invention should not be limited to the above-described embodiments and should not be determined, and not only the claims to be described later, but also all modifications equivalent or equivalent to these claims fall within the scope of the spirit of the present invention. would be said to belong.

Claims

inducing an intra prediction mode of a target block;

constructing a reference sample for the intra prediction; and

Performing the intra prediction on the target block using the intra prediction mode and the reference sample

Image decoding method comprising a.
According to claim 1,

The intra prediction mode is a prediction mode including intra prediction.
According to claim 2,

The intra prediction modes include a Position Dependent Intra Prediction Combination (PDPC) mode, a Cross-Component Linear Model (CCLM) mode, and an Intra Sub-Partitions (ISP) mode. mode, Multiple Reference Line (MRL) mode, Matrix weighted Intra Prediction (MIP) mode, Intra Block Copy (IBC) mode, or Machine Learning (ML) mode. Video decoding method.
According to claim 3,

The ML mode includes one or more of machine learning-based prediction, prediction sample blending, prediction sample improvement, and prediction sample scaling.
According to claim 1,

The intra prediction mode is derived using a most probable mode (MPM) list.
According to claim 1,

The MPM list includes a first MPM list and a second MPM list.
According to claim 6,

The first MPM list includes an intra prediction mode using machine learning;

The second MPM list includes an intra prediction mode that does not use machine learning.
According to claim 7,

In constructing the MPM list, an intra prediction mode of a neighboring block of the target block is replaced with another intra prediction mode, and the replaced intra prediction mode is added to the MPM list.
According to claim 1,

The intra prediction mode is derived using a machine learning-based intra prediction mode derivation method.
According to claim 9,

The machine learning-based intra prediction mode derivation method includes a single layer perceptron (SLP), a multi-layer perceptron (MLP), a deep neural network (DNN), and a convolutional neural network (CNN). and a recurrent neural network (RNN).
According to claim 9,

An output of the machine learning-based intra prediction mode derivation method is one or more probabilities for one or more intra prediction modes.
According to claim 1,

The reference sample is an image decoding method configured using reference sample filtering.
According to claim 1,

The reference sample is configured using reference sample scaling.
According to claim 1,

The intra prediction is an image decoding method performed using machine learning-based prediction.
According to claim 1,

The intra prediction is performed using predictive sample blending.
According to claim 1,

The intra prediction is performed using prediction sample improvement.
According to claim 1,

The intra prediction mode is derived using a probability of a transform mode.
According to claim 1,

The intra prediction mode is derived using a probability of a transform mode set.
inducing an intra prediction mode of a target block;

constructing a reference sample for the intra prediction; and

Performing the intra prediction on the target block using the intra prediction mode and the reference sample

An image encoding method comprising a.
A computer-readable recording medium storing a bitstream for image decoding, wherein the bitstream comprises:

Coded Coding Information

including,

Decoding of the encoded coding information is performed to generate coding information;

An intra prediction mode of a target block is derived using the coding information,

A reference sample for the intra prediction is configured,

A computer-readable recording medium in which the intra prediction of the target block is performed using the intra prediction mode and the reference sample.