Extended block partition type for video coding

By nesting multi-type tree structure and template matching technologies, the partition boundaries of video encoding and decoding are dynamically divided, which solves the problem of difficulty in adapting to local motion and texture features in the existing technology, improves the efficiency and flexibility of video encoding and decoding, and realizes the realization of More efficient codec performance.

CN118985128APending Publication Date: 2024-11-19MEDIATEK INC
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202380033156.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Priority Date
2022-04-13
Filing Date
2023-04-10
Publication Date
2024-11-19

AI Technical Summary

Technical Problem

When dividing pixel blocks, existing video encoding and decoding technologies are difficult to effectively adapt to different local motion and texture features, resulting in low encoding and decoding efficiency. Especially in the case of non-power-second partitioning, the application of transformation cores is limited, which affects the encoding and decoding performance. .

Method used

Using the segmentation methods such as quad-tree, binary-tree and triple-tree nested with multi-type tree structures, dynamically divide partition boundaries through template matching and neural network prediction, optimize partition locations to minimize costs, and support any boundary partitioning, including four The partition boundary of fork trees, binary trees and trident trees, adapts to different types of video content.

Benefits of technology

It improves the flexibility and efficiency of video encoding and decoding, can better adapt to different types of video content, simplify or bypass transformation operations, reduces the complexity and cost of encoding and decoding, and improves the encoding and decoding performance.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118985128A_ABST
    Figure CN118985128A_ABST
Patent Text Reader

Abstract

A method for transmitting arbitrary partition boundaries is provided. A video codec derives a partition structure for segmenting a current block by identifying a partition position with the lowest cost. The video codec divides the current block into first and second partitions according to the identified partition positions. A video codec encodes or decodes the first and second partitions of the current block. The first and second partitions may be associated with first and second templates, the first and second templates being constructed based on reconstructed pixels adjacent to the current block. The video codec may identify the partition location by calculating a first cost based on the first template and a second cost based on the second template and optimizing the partition location to minimize the sum of the first and second costs.
Need to check novelty before this filing date? Find Prior Art

Description

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This disclosure is part of a non-provisional application that claims priority to U.S. Provisional Patent Application No. 63 / 330,343, filed on April 13, 2022. The contents of the above-listed applications are incorporated herein by reference in their entirety. Technical Field

[0003] The present disclosure generally relates to video coding and decoding. In particular, the present disclosure relates to a method for dividing pixel blocks for coding and decoding. Background Art

[0004] Unless otherwise indicated herein, the approaches described in this section are not prior art to the claims listed below and are not admitted to be prior art by inclusion in this section.

[0005] High-Efficiency Video Coding (HEVC) is an international video codec standard developed by the Joint Collaborative Team on Video Coding (JCT-VC). HEVC is based on a hybrid block-based motion compensation DCT transform-like codec architecture. The basic unit of compression, called a coding unit (CU), is a 2Nx2N square pixel block. Each CU can be recursively divided into four smaller CUs until a predetermined minimum size is reached. Each CU contains one or more prediction units (PUs).

[0006] Versatile video coding (VVC) is the latest international video codec standard developed by the Joint Video Expert Team (JVET) of ITU-TSG16 WP3 and ISO / IEC JTC1 / SC29 / WG11. The input video signal is predicted from a reconstructed signal derived from the coded picture region. The prediction residual signal is processed by block transform. The transform coefficients are quantized and entropy coded along with other auxiliary information in the bitstream. The reconstructed signal is generated from the prediction signal and the reconstructed residual signal after inverse transforming the dequantized transform coefficients. The reconstructed signal is further processed by loop filtering to remove codec artifacts. The decoded picture is stored in a frame buffer and used to predict future pictures in the input video signal.

[0007] In VVC, a codec picture is divided into non-overlapping square block areas represented by associated coding tree units (CTUs). A codec picture can be represented by a set of fragments, each fragment containing an integer number of CTUs. The CTUs in a fragment are processed continuously in a raster scan. Intra-frame prediction or inter-frame prediction can be used to decode bi-predictive (B) fragments, where there are at most two motion vectors and reference indices to predict the sample values ​​of each block. Prediction (P) fragments are decoded using intra-frame prediction or inter-frame prediction with at most one motion vector and reference index to predict the sample values ​​of each block. Intra-frame (I) fragments are decoded using only intra-frame prediction.

[0008] A quadtree (QT) with a nested multi-type-tree (MTT) structure can be used to divide a CTU into one or more non-overlapping coding units (CUs) to accommodate various local motion and texture features. A CU can be further divided into smaller CUs using one of five partition types: quadtree partitioning, vertical binary tree partitioning, horizontal binary tree partitioning, vertical center-side ternary tree partitioning, and horizontal center-side ternary tree partitioning.

[0009] Each CU contains one or more prediction units (PUs). The prediction unit, together with the associated CU syntax, serves as a basic unit for sending prediction information. The specified prediction process is used to predict the values ​​of the relevant pixel samples within the PU. Each CU may contain one or more transform units (TUs) for representing prediction residual blocks. A transform unit (TU) consists of a transform block (TB) of luma samples and two corresponding transform blocks of chroma samples, each TB corresponding to a residual block of samples from one color component. Integer transform is applied to the transform block. The level values ​​of the quantization coefficients are entropy encoded and decoded in the bitstream together with other auxiliary information. The terms coding tree block (CTB), coding block (CB), prediction block (PB), and transform block (TB) are defined as 2D sample arrays that specify a color component associated with a CTU, CU, PU, ​​and TU, respectively. Therefore, a CTU consists of a luma CTB, two chroma CTBs, and related syntax elements. Similar relationships are valid for CU, PU, ​​and TU.

[0010] For each inter-predicted CU, motion parameters consisting of motion vector, reference picture index and reference picture list usage index as well as additional information are used for inter-prediction sample generation. Motion parameters can be sent explicitly or implicitly. When a CU is encoded or decoded in skip mode, the CU is associated with one PU and has no significant residual coefficients, no motion vector increments or reference picture indices are encoded or decoded. Merge mode is specified and the motion parameters of the current CU are obtained from neighboring CUs, including spatial and temporal candidates, as well as additional scheduling introduced in VVC. Merge mode can be applied to any inter-predicted CU. An alternative to merge mode is the explicit transmission of motion parameters, where motion vectors, corresponding reference picture indices for each reference picture list and reference picture list usage flags and other required information are explicitly sent per CU. Summary of the invention

[0011] The following summary is illustrative only and is not intended to be binding in any way. That is, the following summary is provided to introduce the concepts, highlights, benefits and advantages of the novel and non-obvious technologies described herein. Selected but not all embodiments are further described in the detailed description below. Therefore, the following summary is not intended to identify the essential features of the claimed subject matter, nor is it intended to be used to determine the scope of the claimed subject matter.

[0012] Some embodiments provide a method for sending arbitrary partition boundaries. The video codec derives a partition structure for segmenting a current block by identifying a partition position with a minimum cost. The video codec may send the identified partition position. The video codec divides the current block into a first and a second partition according to the identified partition position. The video codec encodes or decodes the first and the second partitions of the current block.

[0013] In some embodiments, the first partition and the second partition may be associated with a first template and a second template, the first template and the second template being constructed based on reconstructed pixels adjacent to the current block. The video codec may identify the partition position by calculating a first cost based on the first template and a second cost based on the second template and optimizing the partition position to minimize the sum of the first and second costs. The video codec may calculate the first cost by matching the first template with a first set of reference pixels, the first set of reference pixels being reference pixels identified by a first intra-frame prediction mode of the first partition in the current picture, and by matching the second template with a second set of reference pixels, the second set of reference pixels being reference pixels identified by a second intra-frame prediction mode of the second partition in the current picture. The video codec may calculate the first cost by matching the first template with a first set of reference pixels, the first set of reference pixels being reference pixels identified by a first motion vector of the first partition in the first reference picture, and by matching the second template with a second set of reference pixels, the second set of reference pixels being reference pixels identified by a second motion vector of the second partition in the second reference picture.

[0014] The partition position can be used to specify the partition boundary of a quadtree (QT), a binary tree (BT), or a ternary tree (TT). The partition structure can specify the offset and angle of the straight line that splits the current block (for example, similar to GPM). The width and height of the current block can be integers that are powers of two, and the partition position is an integer pixel position that is not a power of two. The first partition can be a first rectangular area at the corner of the current block, and the second partition is the L-shaped remaining area of ​​the current block minus the first partition. The first partition can be a first rectangular area that is completely within the current block, and the second partition is the remaining area of ​​the current block that surrounds the first rectangular area on all four sides.

[0015] In some embodiments, when a partition does not have a corresponding transform kernel (e.g., has a dimension that is not a power of two), samples of the first or second partition may be encoded as prediction residuals without being transformed (transform skip), or samples of the first or second partition may be encoded by prediction without sending a prediction residual (residual skip). BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The accompanying drawings are included to provide a further understanding of the present disclosure and are incorporated into and constitute a part of the present disclosure. The accompanying drawings illustrate embodiments of the present disclosure and together with the description are used to explain the principles of the present disclosure. It is worth noting that the drawings are not necessarily drawn to scale, as certain components may be shown out of proportion to size in actual implementations in order to clearly illustrate the concepts of the present disclosure.

[0017] Figure 1 An example coding tree unit (CTU) of recursive partitioning is provided.

[0018] Figure 2 Five types of segmentation of the codec unit are shown.

[0019] Figure 3 Shows the mechanism for sending partition split information.

[0020] Figure 4 Demonstrate binary tree (BT) partitioning with arbitrary boundaries.

[0021] Figure 5 Demonstrate arbitrary boundary partitioning of quadtrees (QT) and ternary trees (TT).

[0022] Fig. 6A -B shows the use of arbitrary straight lines to divide the block.

[0023] Fig. 7A -B conceptually illustrates the template matching method used to derive the partition positions of the current block.

[0024] Figure 8 Shows the partitioning of the parent block into corner blocks and an L-shaped remainder block.

[0025] Fig. 9 Parameters that can be used to select an L-shaped partition of the transform kernel are conceptually shown.

[0026] Fig.10 Conceptually showing a block divided by intra-block large block partitions.

[0027] Fig.11 Conceptual illustration of using a neural network to predict segmentation landmarks for pixel blocks.

[0028] Fig.12 An example video encoder is presented that can implement arbitrary partition boundaries.

[0029] Fig.13 Shows the portion of a video encoder that implements arbitrary partition boundaries.

[0030] Fig.14 A process for deriving partition locations for arbitrary partitions is conceptually presented.

[0031] Fig.15 An example video decoder is presented that can implement arbitrary partition boundaries.

[0032] Fig.16 An example video decoder is presented that can implement arbitrary partition boundaries.

[0033] Fig.17 A process for deriving partition locations for arbitrary partitions is conceptually presented.

[0034] Fig.18 An electronic system that implements some embodiments of the present disclosure is conceptually presented. DETAILED DESCRIPTION

[0035] In the following detailed description, many specific details are set forth by way of example in order to provide a thorough understanding of the relevant teachings. Any variations, derivations, and / or extensions based on the teachings described herein are within the scope of protection of the present disclosure. In some cases, well-known methods, processes, components, and / or circuits associated with one or more example embodiments disclosed herein may be described at a relatively high level without details to avoid unnecessarily obscuring aspects of the teachings of the present disclosure.

[0036] Ⅰ. Block partitioning

[0037] A CTU may be partitioned into one or more non-overlapping codec units (CUs) by using a quadtree (QT) with a nested multi-type tree (MTT) structure to accommodate various local motion and texture characteristics. Figure 1 An example CTU 100 is provided which is recursively partitioned by a QT with a nested MTT. Figure 1 In the figure, the bold solid edges represent quadtree partitioning, and the dashed lines represent multi-type tree (MTT) partitioning. As shown, CTU 100 is divided into CUs 110, 120, 130, and 140 by QT. CU 110 is further divided by QT. CU 120 is not further divided. CU 130 is further divided by MTT. CU 140 is further divided by QT and then by MTT.

[0038] Figure 2 Five types of CU partitions are shown, including QT partitions and MTT partitions. As shown in the figure, a CU can be further split into smaller CUs by using QT partitions (SPLIT_QT) or using one of the four MTT partition types: vertical binary partitions (SPLIT_BT_VER), horizontal binary partitions (SPLIT_BT_HOR), vertical ternary partitions (SPLIT_TT_VER), and horizontal ternary partitions (SPLIT_TT_HOR).

[0039] The following parameters are defined for a quadtree with nested multi-type tree coding tree scheme. These parameters are specified by the sequence parameter set (SPS) syntax element and can be further refined by the picture header syntax element.

[0040] –CTUsize: The root node size of the quadtree

[0041] –MinQTSize: Minimum allowed quadtree leaf node size

[0042] –MaxBtSize: Maximum allowed binary tree root node size

[0043] –MaxTtSize: Maximum allowed ternary tree root node size

[0044] –MaxMttDepth: The maximum allowed hierarchical depth of the multi-type tree split from the leaves of the quad tree

[0045] –MinCbSize: The minimum allowed size of the decoded block node

[0046] The codec tree unit (CTU) is considered as the root of a quadtree (or quadtree) and is first divided by the quadtree structure. Then, each quadtree leaf node (when large enough to allow it) is further divided by a multi-type tree structure. Figure 3 The transmission mechanism of partition split information is shown, specifically for a quadtree with a nested multi-type tree coding tree structure. The figure shows the split flags used to indicate the partition tree structure of a block. Specifically, the first flag (mtt_split_cu_flag) is sent to indicate whether the node is further divided; when a node is further divided, the second flag (mtt_split_cu_vertical_flag) is sent to indicate the split direction, and then the third flag (mtt_split_cu_binary_flag) is sent to indicate whether the split is a binary split or a ternary split. Based on the values ​​of mtt_split_cu_vertical_flag and mtt_split_cu_binary_flag, the multi-type tree split mode (MttSplitMode) of the CU is derived as follows:

[0047]

[0048]

[0049] II. Extended Partition

[0050] A. Arbitrary Boundary Segmentation for BT

[0051] Some embodiments of the present disclosure provide arbitrary BT segmentation. Any BT can have any segmentation boundary position. Figure 4 BT partitions with arbitrary boundaries are shown. As shown in the figure, for a CU 400 of size 32x32, arbitrary horizontal BT partitioning can divide the CU into two sub-CUs 410 and 420, one is 32xN and the other is 32x(32-N), where N is an integer between 1 and 31.

[0052] In some embodiments, some partition dimensions for arbitrary boundary BT partitions are not allowed. For example, splitting a 32x32 block into 7x32 and 25x32 partitions for intra-frame coding is not allowed, but splitting a 32x32 block into 16x32 and 16x32 partitions for intra-frame coding is allowed. The allowed partitions are those where the width or height of the block is a power of two, so that existing transform kernels can be applied to these partitions or sub-CUs to complete the coding and decoding process.

[0053] B. Arbitrary Boundary Segmentation for TT and QT

[0054] Arbitrary boundaries can be applied to other types of block partitions, such as TT and QT. Figure 5 Arbitrary boundary partitions for TT and QT are shown. In the figure, the first block 510 is partitioned by arbitrary boundaries 511 and 512 for QT, and the second block 520 is partitioned by arbitrary boundaries 521 and 522 for TT. The arbitrary boundaries are defined by widths M and N. In some embodiments, the partition (or partition) position (M or N value) can be sent from the encoder to the decoder. In some embodiments, the partition position can be derived at the decoder side by a template matching method (see below). Fig. 7A -B for description).

[0055] In some embodiments, blocks for intra-frame coding or inter-frame coding can be divided by arbitrary TT or QT boundaries to simplify coding operations. In some embodiments, partitions created by arbitrary TT or QT boundaries are coded and decoded using residual skip mode or transform skip mode to simplify or bypass transform operations. Partitions coded and decoded by transform skip mode encode and send their prediction residuals without transforming to another domain (e.g., the transform step is skipped). Partitions coded and decoded by residual skip mode only send their prediction information (e.g., inter-frame prediction motion information, intra-frame prediction direction) without sending prediction residuals (the residual generation step is skipped). In some embodiments, some partition dimensions used for arbitrary boundary TT or QT partitioning are not allowed for intra-frame coding. Allowed partitions are partitions where the width or height of the block is a power of two, so that existing transform kernels can be applied to these partitions or sub-CUs to complete the coding process.

[0056] C. Similar to GPM partition

[0057] In some embodiments, the parent block can be divided into two or more parts by an inclined partition boundary (not a horizontal boundary or a vertical boundary). Such a partition boundary can be an arbitrary straight line specified by an angle parameter and an offset parameter, i.e., in a manner similar to a geometric partitioning mode (GPM). Fig. 6A -B shows the division of the block by arbitrary straight lines. Fig. 6AThe block 610 is shown bidirectionally partitioned into parts A and B. Partitions A and B may be different CUs with separate codec streams, such as separate codec modes, separate candidate lists, and / or separate residual codecs. Residual codecs may be applied with transform skipping. Otherwise, residual skipping may be applied to the partitioned block to complete the codec process.

[0058] In some embodiments, the picture / tile / slice / CTU-row / CTU header may be used to turn on or off GPM-like partitioning. The video codec may set the partitioning control based on content-related characteristics. For example, the encoder may perform picture analysis (e.g., MV field analysis) to decide whether to turn on GPM-like partitioning.

[0059] In some embodiments, the TT partition may also be similar to the GPM, so that two inclined straight lines may divide the CU into three parts. Figure 6B Block 620 is shown being partitioned three-way into parts A, B, and C. Since the partitions created by the GPM-like partitioning are non-power-of-two blocks, residual skipping and / or transform skipping can be used to encode these partitions.

[0060] D. Implicitly exported arbitrary partition locations

[0061] In some embodiments, the partition positions (eg, N values) may be explicitly sent from the encoder to the decoder. In some embodiments, the partition positions may be implicitly derived at the decoder side by a template matching method. Fig. 7A -B conceptually shows a template matching method for deriving the partition position of the current block 700. The current block is partitioned by BT into partitions 710 and 720 by an arbitrary partition position N. Partition 710 has a corresponding template 715 of the corresponding reconstructed neighboring samples. Partition 720 has a corresponding template 725 of the reconstructed neighboring samples.

[0062] Fig. 7AThe partition positions are derived when the current block 700 is intra-coded. As shown, partition 710 is intra-coded by a first intra-prediction direction or mode (intra1), while partition 720 is intra-coded by a second intra-prediction direction or mode (intra2). (The two partitions can also share the same intra-prediction direction or mode.) The intra-prediction direction "intra1" is used to identify neighboring pixels 730 to match with the template 715 to generate a matching cost "cost1". The intra-prediction direction "intra2" is used to identify neighboring pixels 730 to match with the template 725 to generate a matching cost "cost2". The matching cost "cost1" is added to "cost2" to calculate the cost of using a specific N value as the partition position of the partitioned block 700. The encoder and decoder will independently derive the N value that results in the lowest cost1+cost2 value as the partition position of the block 700.

[0063] Figure 7B The partition positions are derived when the current block 700 is inter-coded. As shown, partition 710 is inter-coded by a first motion vector (MV1) and partition 720 is inter-coded by a second motion vector (MV2). (The two partitions may also share the same motion vector.) Motion vector MV1 is used to identify a reference pixel set 718, which corresponds to template 715 in a reference picture 719, and motion vector MV2 is used to identify a reference pixel set 728, which corresponds to template 725 in a reference picture 729. Reference pixels 718 are used to generate a matching cost "cost1" corresponding to template 715, and reference pixels 728 are used to generate a matching cost "cost2" corresponding to template 725. Cost1 and Cost2 are added together to calculate the cost of using a particular value of N as the partition position for partitioning block 700. Both the encoder and decoder will independently derive the value of N that results in the lowest cost1+cost2 value as the partition position for block 700.

[0064] BT partitioning with arbitrary boundaries is beneficial for many different types of video content, because object boundaries usually do not fall in the middle of the parent block (traditional VBT, HBT cases cannot match perfectly). In some embodiments, arbitrary boundary BT partitioning can be turned on or off for different granularity levels, for example, CTU / CTU-row / tile / slice / picture / GOP (group of pictures).

[0065] In some embodiments, arbitrary BT boundaries (e.g., non-power of two) are only allowed for blocks encoded by residual or transform skip mode to simplify or bypass transform operations. In some embodiments, arbitrary BT boundaries are allowed for blocks encoded by inter-mode to simplify encoding and decoding operations.

[0066] In some embodiments, some partition dimensions of the implicitly derived arbitrary boundary BT partition are not allowed. For example, splitting a 32x32 block into 7x32 and 25x32 partitions does not allow intra-frame coding, but splitting a 32x32 block into 16x32 and 16x32 partitions does allow intra-frame coding. The allowed partitions are those where the width or height of the block is still a power of two, so that existing transform kernels can be applied to these partitions or sub-CUs to complete the coding process.

[0067] E. Corner partition

[0068] In some embodiments, the parent CU may be divided into 2 partitions such that one partition is a corner rectangular block and the other partition is the remaining L-shaped partition. Figure 8 The parent block 800 is shown to be divided into a corner block 810 and an L-shaped remainder 820. The ratio between the parent block and the corner block 810 can be 2:1, 4:1, or other ratios. In some embodiments, the partition position (N or M value) can be sent from the encoder to the decoder. In some embodiments, the partition position can be determined at the decoder side by referring to the above Fig. 7A -B describes the template matching method to export.

[0069] In some embodiments, the residual transform of the L-shaped partition 820 can be performed by filling the L-shape into a rectangle and transforming by a conventional transform kernel. In some embodiments, the residual transform can be performed by dividing the L-shaped partition 820 into two rectangular sub-residual blocks and performing the transforms separately. In some embodiments, the residual transform of the L-shaped partition 820 can be performed by using a specially designed L-shaped transform kernel, and the video codec can adaptively select the transform kernel according to the length of the row or column. In some embodiments, the L-shaped partition is coded and decoded by using residual skipping or transform skipping.

[0070] Fig. 9 Parameters of an L-shaped partition that can be used to select a transform kernel are conceptually shown. The figure shows a 2Nx2N CU 900. The CU 900 has an N×N corner partition 910 and an L-shaped remaining portion 920. When a horizontal transform is applied to the CU 900, a transform kernel with a width of 2N is first applied to the upper portion of the L-shape 920, and a transform kernel with a width of N is applied to the lower portion. In a similar manner, when a vertical transform is applied to the CU 900, a transform kernel with a height of 2N is first applied to the left portion, and a transform kernel with a height of N is applied to the right portion.

[0071] F. Blocks within a block

[0072] In some embodiments, "chunk within chunk" partitioning may be used. Fig.10Blocks partitioned by intra-block larger block partitioning are shown. As shown, CU 1000 is partitioned into an inner rectangular portion 1010 and an outer surrounding remainder 1020. Intra-block large block partitioning can promote codec gains for certain types of content that cannot be easily partitioned by BT / QT / TT. In some embodiments, the position of the inner rectangular portion 1010 (e.g., as an offset relative to the upper left corner of the parent block) is sent from the encoder to the decoder.

[0073] In some embodiments, the residual transform of the outer enclosing portion 1020 can be performed by filling the hole in the middle to create a rectangle to be transformed by a conventional transform. In some embodiments, the residual transform can be performed by dividing the outer enclosing portion 1020 into several rectangular sub-residual blocks and transforming them separately. In some embodiments, the residual transform of the outer enclosing portion 1020 can be performed by using a specially designed outer enclosing transform kernel, and the video codec can adaptively select the transform kernel according to the length of the row or column. In some embodiments, the outer enclosing portion 1020 is encoded and decoded by using a skip method, such as residual skip or transform skip.

[0074] G. Segmentation landmark prediction using neural networks

[0075] In some embodiments, neural networks (NN) are used to predict a parent CU's segmentation flags, such as those described in Section I above. Fig.11 The use of a neural network to predict a segmentation flag for partitioning a pixel block is conceptually shown. As shown, the content of CTU 1100 is used as input to neural network 1110. The current CU region 1120 is not available as input to neural network 1110 at the decoder side because CU 1120 has not yet been decoded. After the neural network analysis, the segmentation flag is generated in whole or in part based on the results generated by neural network 1110. The current CU region 1120 can be further segmented based on the predicted segmentation flag.

[0076] The input of the neural network is a large range of pixels of neighboring samples of the current region, and the output of the network is the estimated prediction result of the segmentation mark of the current region. The current parent CU is the current region that needs to predict the segmentation mark. The predicted segmentation mark is used to decide how to split the current "parent CU".

[0077] In some embodiments, a neural network for segmentation flag prediction is implemented and applied on the decoder side. In some embodiments, the decoder makes its segmentation decision based in part on the prediction results. In other words, the decoder relies on information explicitly sent by the encoder and the segmentation flag prediction generated by the neural network. For example, the encoder may only send a BT segmentation flag to the decoder without explicit sending of HBT or VBT. The decoder makes a partition decision between HBT and VBT based on the neural network output.

[0078] In some embodiments, the encoder and decoder use the neural network prediction results to decide the order among all segmentation directions, and then the encoder can send an index to indicate the segmentation direction based on the order determined by the neural network. In some embodiments, the HBT and VBT segmentation decisions are initially ordered by the neural network prediction results. For example, if the HBT is better than the VBT after the order, the encoder and decoder both designate the HBT as index 0 and the VBT as index 1. The encoder can send index 0, and the decoder can understand this to select the HBT. This reordering scheme is used to reduce the syntax rate of sending segmentation flags after context encoding and decoding (e.g., CABAC encoding and decoding).

[0079] Neural networks can also be used to determine the partition positions in asymmetric partition modes (ABT / AQT / ATT). Neural networks provide an alternative to the above template-based method for determining CU partition positions.

[0080] The aforementioned proposed method may be implemented in an encoder and / or a decoder. For example, the proposed method may be implemented in an inter-frame coding and decoding module of an encoder and / or an inter-frame coding and decoding module of a decoder. For example, the proposed method may be implemented in an intra-frame coding and decoding module of an encoder and / or an intra-frame coding and decoding module of a decoder. For example, the proposed method may be implemented in a tree segmentation coding and decoding module of an encoder and / or a tree segmentation coding and decoding module of a decoder.

[0081] III. Example Video Encoder

[0082] Fig.12An example video encoder 1200 that can implement arbitrary partition boundaries is shown. As shown, the video encoder 1200 receives an input video signal from a video source 1205 and encodes the signal into a bitstream 1295. The video encoder 1200 has several components or modules for encoding the signal from the video source 1205, including at least some components selected from the following: a transform module 1210, a quantization module 1211, an inverse quantization module 1214, an inverse transform module 1215, an intra-frame estimation module 1220, an intra-frame prediction module 1225, a motion compensation module 1230, a motion estimation module 1235, a loop filter 1245, a reconstructed picture buffer 1250, an MV buffer 1265, an MV prediction module 1275 and an entropy encoder 1290. The motion compensation module 1230 and the motion estimation module 1235 are part of the inter-frame prediction module 1240.

[0083] In some embodiments, modules 1210-1290 are software instruction modules executed by one or more processing units (e.g., processors) of a computing device or electronic device. In some embodiments, modules 1210-1290 are hardware circuit modules implemented by one or more integrated circuits (ICs) of an electronic device. Although modules 1210-1290 are shown as separate modules, some modules may be combined into a single module.

[0084] The video source 1205 provides a raw video signal that presents the pixel data of each video frame without compression. The subtractor 1208 calculates the difference between the raw video pixel data of the video source 1205 and the predicted pixel data 1213 from the motion compensation module 1230 or the intra-frame prediction module 1225. The transform module 1210 converts the difference (or residual pixel data or residual signal) into transform coefficients (e.g., by performing a discrete cosine transform or DCT). The quantization module 1211 quantizes the transform coefficients into quantization data (or quantization coefficients) 1212, which are encoded into a bitstream 1295 by the entropy encoder 1290.

[0085] The inverse quantization module 1214 dequantizes the quantized data (or quantized coefficients) 1212 to obtain transform coefficients, and the inverse transform module 1215 performs an inverse transform on the transform coefficients to generate a reconstructed residual 1219. The reconstructed residual 1219 is added to the predicted pixel data 1213 to generate reconstructed pixel data 1217. In some embodiments, the reconstructed pixel data 1217 is temporarily stored in a line buffer (not shown) for intra prediction and spatial MV prediction. The reconstructed pixels are filtered by the loop filter 1245 and stored in the reconstructed picture buffer 1250. In some embodiments, the reconstructed picture buffer 1250 is a memory external to the video encoder 1200. In some embodiments, the reconstructed picture buffer 1250 is a memory internal to the video encoder 1200.

[0086] The intra estimation module 1220 performs intra prediction based on the reconstructed pixel data 1217 to generate intra prediction data. The intra prediction data is provided to the entropy encoder 1290 to be encoded into a bitstream 1295. The intra prediction data is also used by the intra prediction module 1225 to generate predicted pixel data 1213.

[0087] The motion estimation module 1235 performs inter-frame prediction by generating MVs to refer to pixel data of a previously decoded frame stored in the reconstructed picture buffer 1250. These MVs are provided to the motion compensation module 1230 to generate predicted pixel data.

[0088] Instead of encoding the complete actual MV in the bitstream, the video encoder 1200 uses MV prediction to generate a predicted MV, and the difference between the MV used for motion compensation and the predicted MV is encoded as residual motion data and stored in the bitstream 1295.

[0089] The MV prediction module 1275 generates a predicted MV based on a reference MV generated for encoding a previous video frame, i.e., a motion compensated MV for performing motion compensation. The MV prediction module 1275 obtains the reference MV from the previous video frame from the MV buffer 1265. The video encoder 1200 stores the MV generated for the current video frame in the MV buffer 1265 as a reference MV for generating a predicted MV.

[0090] The MV prediction module 1275 uses the reference MV to create a predicted MV. The predicted MV can be calculated by spatial MV prediction or temporal MV prediction. The difference (residual motion data) between the predicted MV and the motion compensation MV (MC MV) of the current frame is encoded into the bit stream 1295 by the entropy encoder 1290.

[0091] The entropy encoder 1290 encodes various parameters and data into a bitstream 1295 by using an entropy coding technique such as context-adaptive binary arithmetic coding (CABAC) or Huffman coding. The entropy encoder 1290 encodes various header elements, flags, along with quantized transform coefficients 1212 and residual motion data as syntax elements into the bitstream 1295. The bitstream 1295 is then stored in a storage device or transmitted to a decoder via a communication medium such as a network.

[0092] The loop filter 1245 performs filtering or smoothing operations on the reconstructed pixel data 1217 to reduce encoding and decoding artifacts, particularly at the boundaries of pixel blocks. In some embodiments, the filtering operation performed includes sample adaptive offset (SAO). In some embodiments, the filtering operation includes an adaptive loop filter (ALF).

[0093] Fig.13 A portion of a video encoder 1200 that implements arbitrary partition boundaries is shown. A partition engine 1310 can generate a partition information set 1320 to an entropy encoder 1290. At different levels of the video hierarchy (e.g., sequence, picture, slice, block), the entropy encoder 1290 encodes or sends the partition information set 1320 as a syntax element to a bitstream 1295. The partition engine 1310 also provides a partition structure 1330 to the transform module 1210 so that the transform module can perform a transform operation on the prediction residual 1209 according to the partition structure 1330 to produce quantized coefficients 1212. For example, the transform module 1210 can use different transform kernels for different parts of the L-shaped partition.

[0094] The partition engine 1310 can test different partition positions by using the template matching engine 1315 to calculate the costs associated with those different partition positions. Specifically, the partition engine 1310 can specify the partition 1350 created by the partition position to the template matching engine 1315, and receive the template cost 1355 of the specified partition 1350. Based on the received template costs of different partitions created by the different partition positions, the partition engine 1310 determines the best (lowest cost) partition position. The best partition position is provided as part of the partition structure 1330.

[0095] The template matching engine 1315 obtains samples stored in the reconstructed picture buffer 1250 to construct a template for the specified partition 1350. The template matching engine 1315 also obtains reference samples corresponding to the template from the reconstructed picture buffer 1250, based on the prediction mode assigned to the partition. The prediction mode can be an intra prediction direction provided by the intra estimation module 1220 or a motion vector provided by the motion estimation module 1235. The template matching module 1315 then compares the constructed template with the obtained reference samples to generate a template cost 1355 for the specified partition 1350.

[0096] In some embodiments, if the partition structure 1330 indicates a partition that does not have a corresponding transform kernel (e.g., does not have a dimension that is not a power of two), the prediction residual 1209 for the partition is not transformed (transform skip), but is directly provided to the entropy encoder 1290 to be encoded and sent in the bitstream 1295. In some embodiments, the prediction residual 1209 is also not sent (residual skip), and the encoding of the partition is based entirely on the prediction mode of the partition.

[0097] Fig.14 The process 1400 for deriving the partition location for an arbitrary partition is conceptually presented. In some embodiments, one or more processing units (e.g., processors) of a computing device implementing the encoder 1200 perform the process 1400 by executing instructions stored in a computer-readable medium. In some embodiments, an electronic device implementing the encoder 1200 performs the process 1400.

[0098] The encoder receives (at block 1410) data for a block of pixels that is to be encoded as a current block in a current picture.

[0099] The encoder derives (at block 1420) a partition structure for partitioning the current block by identifying the partition position with the lowest cost. The encoder partitions (at block 1430) the current block into a first partition and a second partition according to the identified partition position. In some embodiments, the encoder also sends the derived partition position. The partition position can be used to specify a partition boundary of a quadtree (QT), a binary tree (BT), or a ternary tree (TT). The partition structure can specify an offset and angle of a straight line for partitioning the current block (e.g., a GPM-like).

[0100] The width and height of the current block may be integers of a power of two, and the partition positions are integer pixel positions that are not powers of two. The first partition may be a first rectangular area at a corner of the current block, and the second partition is an L-shaped remaining area of ​​the current block minus the first partition. The first partition may be a first rectangular area completely within the current block, and the second partition is a remaining area of ​​the current block surrounded by four sides of the first rectangular area.

[0101] In some embodiments, the first partition and the second partition are associated with a first template and a second template (e.g., templates 715 and 725), respectively, and the first template and the second template are reconstructed based on pixels adjacent to the current block. Therefore, the encoder can identify the partition position by calculating a first cost based on the first template and a second cost based on the second template and optimizing the partition position to minimize the sum of the first and second costs. The encoder can calculate the first cost by matching the first template with a first reference pixel set, which is the reference pixel identified by the first intra-frame prediction mode of the first partition in the current picture, and by matching the second template with a second reference pixel set to calculate the second cost, which is the reference pixel identified by the second intra-frame prediction mode of the second partition in the current picture. The encoder can calculate the first cost by matching the first template with a first reference pixel set, which is the reference pixel identified by the first motion vector of the first partition in the first reference picture, and by matching the second template with a second reference pixel set to calculate the second cost, which is the reference pixel identified by the second motion vector of the second partition in the second reference picture.

[0102] The encoder encodes the first and second partitions of the current block (at block 1440). When the partition does not have a corresponding transform kernel (e.g., has a dimension that is not a power of two), the samples of the first or second partition may be encoded as a prediction residual without being transformed (transform skip), or the samples of the first or second partition may be encoded by prediction without sending a prediction residual (residual skip).

[0103] IV. Sample Video Decoder

[0104] In some embodiments, the encoder may send (or generate) one or more syntax elements in a bitstream so that the decoder may parse the one or more syntax elements from the bitstream.

[0105] Fig.15 An example video decoder 1500 is shown that can implement arbitrary partition boundaries. As shown, the video decoder 1500 is an image decoding or video decoding circuit that receives a bitstream 1595 and decodes the contents of the bitstream into pixel data of a video frame for display. The video decoder 1500 has several components or modules for decoding the bitstream 1595, including some components selected from the following: an inverse quantization module 1511, an inverse transform module 1510, an intra-frame prediction module 1525, a motion compensation module 1530, a loop filter 1545, a decoded picture buffer 1550, an MV buffer 1565, an MV prediction module 1575 and a parser 1590. The motion compensation module 1530 is part of the inter-frame prediction module 1540.

[0106] In some embodiments, modules 1510-1590 are software instruction modules executed by one or more processing units (e.g., processors) of a computing device. In some embodiments, modules 1510-1590 are hardware circuit modules implemented by one or more ICs of an electronic device. Although modules 1510-1590 are shown as separate modules, some modules may be combined into a single module.

[0107] The parser 1590 (or entropy decoder) receives the bitstream 1595 and performs initial parsing according to the syntax defined by the video coding or image coding standard. The parsed syntax elements include various header elements, flags, and quantization data (or quantization coefficients) 1512. The parser 1590 parses out various syntax elements by using entropy coding techniques (e.g., context-adaptive binary arithmetic coding (ABAC) or Huffman encoding).

[0108] The inverse quantization module 1511 dequantizes the quantized data (or quantized coefficients) 1512 to obtain transform coefficients, and the inverse transform module 1510 inversely transforms the transform coefficients 1516 to generate a reconstructed residual signal 1519. The reconstructed residual signal 1519 is added to the predicted pixel data 1513 from the intra prediction module 1525 or the motion compensation module 1530 to generate decoded pixel data 1517. The decoded pixel data is filtered by the loop filter 1545 and stored in the decoded picture buffer 1550. In some embodiments, the decoded picture buffer 1550 is a memory external to the video decoder 1500. In some embodiments, the decoded picture buffer 1550 is a memory internal to the video decoder 1500.

[0109] The intra prediction module 1525 receives intra prediction data from the bitstream 1595 and, based thereon, generates predicted pixel data 1513 from decoded pixel data 1517 stored in the decoded picture buffer 1550. In some embodiments, the decoded pixel data 1517 is also stored in a line buffer (not shown) for intra prediction and spatial MV prediction.

[0110] In some embodiments, the contents of the decoded picture buffer 1550 are used for display. The display device 1555 either obtains the contents of the decoded picture buffer 1550 for direct display or obtains the contents of the decoded picture buffer to a display buffer. In some embodiments, the display device receives pixel values ​​from the decoded picture buffer 1550 via pixel transfer.

[0111] The motion compensation module 1530 generates predicted pixel data 1513 from the decoded pixel data 1517 stored in the decoded picture buffer 1550 according to the motion compensated MV (MC MV). These motion compensated MVs are decoded by adding the residual motion data received from the bitstream 1595 and the predicted MV received from the MV prediction module 1575.

[0112] The MV prediction module 1575 generates a predicted MV based on a reference MV generated for decoding a previous video frame (e.g., a motion compensated MV for performing motion compensation). The MV prediction module 1575 obtains the reference MV of the previous video frame from the MV buffer 1565. The video decoder 1500 stores the motion compensated MV generated for decoding the current video frame in the MV buffer 1565 as a reference MV for generating a predicted MV.

[0113] The loop filter 1545 performs filtering or smoothing operations on the decoded pixel data 1517 to reduce encoding and decoding artifacts, particularly at the boundaries of pixel blocks. In some embodiments, the filtering operation performed includes sample adaptive offset (SAO). In some embodiments, the filtering operation includes an adaptive loop filter (ALF).

[0114] Fig.16 A portion of a video decoder 1500 that implements arbitrary partition boundaries is shown. A partition engine 1610 may receive a partition information set 1620 from an entropy decoder 1590. The entropy decoder 1590 may decode the partition information set 1620 based on syntax elements of a bitstream 1595 at different levels of the video hierarchy (e.g., sequence, picture, slice, block). The partition engine 1610 provides a partition structure 1630 to an inverse transform module 1510 so that the inverse transform module may perform an inverse transform operation based on the quantized coefficients 1512 to produce a prediction residual 1509 according to the partition structure 1630. For example, the inverse transform module 1510 may use different transform kernels for different parts of an L-shaped partition.

[0115] The partition engine 1610 can test different partition positions by using the template matching engine 1615 to calculate the costs associated with those different partition positions. Specifically, the partition engine 1610 can specify the partition 1650 created by the partition position to the template matching engine 1615, and receive the template cost 1655 of the specified partition 1650. Based on the received template costs of different partitions created by the different partition positions, the partition engine 1610 determines the best (lowest cost) partition position. The best partition position is provided as part of the partition structure 1630.

[0116] The template matching engine 1615 obtains samples stored in the reconstructed picture buffer 1550 to construct a template for the specified partition 1650. The template matching engine 1615 also obtains reference samples corresponding to the template from the reconstructed picture buffer 1550, based on the prediction mode assigned to the partition. The prediction mode can be an intra-frame prediction direction or a motion vector provided by the entropy decoder 1590. The template matching module 1615 then compares the constructed template with the obtained reference samples to generate a template cost 1655 for the specified partition 1650.

[0117] In some embodiments, if the partition structure 1630 indicates that the partition does not have a corresponding transform kernel (e.g., has a dimension that is not a power of two), the prediction residual 1509 for the partition is not transformed (transform skip), but is received directly from the entropy decoder 1590. In some embodiments, the prediction residual 1509 is not sent (residual skip), and the decoding of the partition is completely based on the prediction mode of the partition.

[0118] Fig.17 The process 1700 for deriving the partition location of an arbitrary partition is conceptually shown. In some embodiments, one or more processing units (e.g., processors) of a computing device implementing the decoder 1500 perform the process 1700 by executing instructions stored in a computer-readable medium. In some embodiments, an electronic device implementing the decoder 1500 performs the process 1700.

[0119] The decoder receives (at block 1710) data for a block of pixels that is to be decoded as a current block in a current picture of a video.

[0120] The decoder derives (at block 1720) a partition structure for partitioning the current block by identifying the partition position with the lowest cost. The decoder partitions (at block 1730) the current block into a first partition and a second partition according to the identified partition position. In some embodiments, the decoder receives the partition position from the encoder. The partition position can be used to specify a partition boundary of a quadtree (QT), a binary tree (BT), or a ternary tree (TT). The partition structure can specify an offset and angle of a straight line for partitioning the current block (e.g., a GPM-like).

[0121] The width and height of the current block may be integers of a power of two, and the partition positions are integer pixel positions that are not powers of two. The first partition may be a first rectangular area at a corner of the current block, and the second partition is an L-shaped remaining area of ​​the current block minus the first partition. The first partition may be a first rectangular area completely within the current block, and the second partition is a remaining area of ​​the current block surrounded by four sides of the first rectangular area.

[0122] In some embodiments, the first partition and the second partition are associated with a first template and a second template (e.g., templates 715 and 725), respectively, and the first template and the second template are reconstructed based on pixels adjacent to the current block. Therefore, the encoder can identify the partition position by calculating a first cost based on the first template and a second cost based on the second template and optimizing the partition position to minimize the sum of the first and second costs. The decoder can calculate the first cost by matching the first template with a first reference pixel set, which is the reference pixel identified by the first intra-frame prediction mode of the first partition in the current picture, and by matching the second template with a second reference pixel set to calculate the second cost, which is the reference pixel identified by the second intra-frame prediction mode of the second partition in the current picture. The decoder can calculate the first cost by matching the first template with a first reference pixel set, which is the reference pixel identified by the first motion vector of the first partition in the first reference picture, and by matching the second template with a second reference pixel set to calculate the second cost, which is the reference pixel identified by the second motion vector of the second partition in the second reference picture.

[0123] The decoder reconstructs the first and second partitions of the current block (at block 1740). When the partition does not have a corresponding transform kernel (e.g., has a dimension that is not a power of two), the samples of the first or second partition may be encoded and decoded as prediction residuals without being transformed (transform skipping), or the samples of the first or second partition may be encoded and decoded by prediction without sending prediction residuals (residual skipping). In these cases, the decoder may bypass the inverse transform operation. The decoder may then provide the reconstructed current block for display as part of the reconstructed current picture.

[0124] V. Example Electronic System

[0125] Many of the above features and applications are implemented as software processes, which are specified as a set of instructions recorded on a computer-readable storage medium (also referred to as a computer-readable medium). When these instructions are executed by one or more computing or processing units (e.g., one or more processors, processor cores, or other processing units), they cause the processing units to perform the actions indicated in the instructions. Examples of computer-readable media include, but are not limited to, compact disc read-only memory (CD-ROM), flash drives, random access memory (RAM) chips, hard drives, erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), etc. Computer-readable media do not include carrier waves and electronic signals transmitted by wireless or wired connections.

[0126] In this specification, the term "software" is intended to include firmware residing in read-only memory or application programs stored in magnetic memory, which can be read into memory for processing by a processor. In addition, in some embodiments, multiple software inventions can be implemented as sub-parts of a larger program while retaining different software inventions. In some embodiments, multiple software inventions can also be implemented as separate programs. Finally, any combination of separate programs that jointly implement the software inventions described herein are within the scope of this disclosure. In some embodiments, the software program, when installed to run on one or more electronic systems, defines one or more specific machine implementations that process and execute the operations of the software program.

[0127] Fig.18 An electronic system 1800 is conceptually shown to implement some embodiments of the present disclosure. The electronic system 1800 may be a computer (e.g., a desktop computer, a personal computer, a tablet computer, etc.), a phone, a PDA, or any other type of electronic device. Such an electronic system includes various types of computer-readable media and interfaces for various other types of computer-readable media. The electronic system 1800 includes a bus 1805, a processing unit 1810, a graphics processing unit (GPU) 1815, a system memory 1820, a network 1825, a read-only memory 1830, a permanent storage device 1835, an input device 1840, and an output device 1845.

[0128] Bus 1805 collectively represents all system, peripheral, and chipset buses that communicatively connect numerous internal devices of electronic system 1800. For example, bus 1805 communicatively connects processing unit 1810 with GPU 1815, read-only memory 1830, system memory 1820, and permanent storage device 1835.

[0129] The processing unit 1810 obtains instructions to be executed and data to be processed from these various memory units in order to perform the processing of the present disclosure. In different embodiments, the processing unit can be a single processor or a multi-core processor. Some instructions are passed to and executed by the GPU 1815. The GPU 1815 can offload various calculations or supplement the image processing provided by the processing unit 1810.

[0130] Read-only-memory (ROM) 1830 stores static data and instructions used by processing unit 1810 and other modules of the electronic system. On the other hand, permanent storage device 1835 is a read-write storage device. This device is a non-volatile storage unit that stores instructions and data even when electronic system 1800 is turned off. Some embodiments of the present disclosure use a large-capacity memory device (such as a magnetic disk or optical disk and its corresponding magnetic disk drive) as permanent storage device 1835.

[0131] Other embodiments use removable storage devices (e.g., floppy disks, flash memory devices, etc., and their corresponding disk drives) as permanent storage devices. Like the permanent storage device 1835, the system memory 1820 is a read-write memory device. However, unlike the permanent storage device 1835, the system memory 1820 is a volatile read-write memory, such as a random access memory. The system memory 1820 stores some instructions and data used by the processor at runtime. In some embodiments, the processing according to the present disclosure is stored in the system memory 1820, the permanent storage device 1835 and / or the read-only memory 1830. For example, according to some embodiments of the present disclosure, various memory units include instructions for processing multimedia clips. From these various memory units, the processing unit 1810 obtains instructions to be executed and data to be processed in order to perform the processing of some embodiments.

[0132] The bus 1805 is also connected to an input device 1840 and an output device 1845. The input device 1840 enables a user to convey information and select commands to the electronic system. The input device 1840 includes an alphanumeric keyboard and a pointing device (also referred to as a "cursor control device"), a camera (e.g., a webcam), a microphone, or a similar device for receiving voice commands, etc. The output device 1845 displays an image generated by the electronic system or outputs data. The output device 1845 includes a printer and a display device, such as a cathode ray tube (CRT) or a liquid crystal display (LCD), and a speaker or similar audio output device. Some embodiments include devices used as input and output devices, such as a touch screen.

[0133] Finally, if Fig.18 As shown, bus 1805 also couples electronic system 1800 to a network 1825 via a network adapter (not shown). In this manner, the computer may be part of a computer network, such as a local area network ("LAN"), a wide area network ("WAN"), or an intranet, or a network of multiple networks, such as the Internet. Any or all components of electronic system 1800 may be used in conjunction with the present disclosure.

[0134] Some embodiments include electronic components, such as microprocessors, storage devices, and memory, which store computer program instructions in machine-readable or computer-readable media (or otherwise referred to as computer-readable storage media, machine-readable media, or machine-readable storage media). Some examples of such computer-readable media include RAM, ROM, read-only compact discs (CD-ROM), recordable compact discs (CD-R), rewritable compact discs (CD-RW), read-only digital versatile discs (e.g., DVD-ROM, dual-layer DVD-ROM), various recordable / rewritable DVDs (e.g., DVD-RAM, DVD-RW, DVD+RW, etc.), flash memory (e.g., SD card, mini SD card, micro SD card, etc.), magnetic and / or solid-state hard drives, read-only and recordable A computer program or computer code may be a computer program that is executable by at least one processing unit and includes a set of instructions for performing various operations. Examples of computer programs or computer codes include machine code such as produced by a compiler, and files including high-level code that is executed by a computer, an electronic component, or a microprocessor using an interpreter.

[0135] Although the above discussion mainly involves microprocessors or multi-core processors executing software, many of the above features and applications are performed by one or more integrated circuits, such as application specific integrated circuits (ASICs) or field programmable gate arrays (FPGAs). In some embodiments, such integrated circuits execute instructions stored on the circuits themselves. In addition, some embodiments execute software stored in programmable logic devices (PLDs), ROM, or RAM devices.

[0136] As used in this specification and any claims of this application, the terms "computer," "server," "processor," and "memory" refer to electronic or other technical devices. These terms do not include people or groups of people. For the purposes of this specification, the terms display or display refer to displaying on an electronic device. As used in this specification and any claims of this application, the terms "computer-readable medium," "computer-readable medium," and "machine-readable medium" are entirely limited to tangible physical objects that store information in a computer-readable form. These terms do not include any wireless signals, wired download signals, and any other transient signals.

[0137] Although the present disclosure has been described with reference to many specific details, those skilled in the art will recognize that the present disclosure may be implemented in other specific forms without departing from the spirit of the present disclosure. Fig.14 and Fig.17 ) conceptually illustrates the processes. The specific operations of these processes may not be performed in the exact order shown and described. The specific operations may not be performed in a continuous series of operations, and different specific operations may be performed in different embodiments. In addition, the process may be implemented using several sub-processes or as part of a larger macro process. Therefore, it will be understood by those skilled in the art that the present disclosure is not limited by the foregoing illustrative details, but is defined by the appended claims.

[0138] Additional Notes

[0139] The subject matter described herein sometimes represents different components, which are contained in or connected to other different components. It is understood that the described structure is only an example, and in fact, it can be implemented by many other structures to achieve the same function. Conceptually, the arrangement of any components that achieve the same function is actually "associated" to achieve the desired function. Therefore, regardless of the structure or intermediate components, any two components combined to achieve a specific function are considered to be "interrelated" to achieve the desired function. Similarly, any two associated components are considered to be "operably connected" or "operably coupled" to achieve a specific function. Any two components that can be associated with each other are also considered to be "operably coupled" to achieve a specific function. Any two components that can be associated with each other are also considered to be "operably coupled" to achieve a specific function. Specific examples of operable connections include, but are not limited to, physically pairable and / or physically interacting components, and / or wirelessly interactive and / or wirelessly interactive components, and / or logically interactive and / or logically interactive components.

[0140] Furthermore, with respect to the use of substantially any plural and / or singular terms, those skilled in the art can translate from the plural to the singular and / or from the singular to the plural as appropriate to the context and / or application. For sake of clarity, the present invention expressly sets forth various singular / plural arrangements.

[0141] In addition, it will be understood by those skilled in the art that, in general, the terms used in the present invention, especially in the claims, and the subject matter of the claims, are generally used as "open" terms, for example, "including" should be interpreted as "including but not limited to", "having" should be interpreted as "at least having", "including" should be interpreted as "including but not limited to", etc. It will be further understood by those skilled in the art that if a specific number of claim contents are intended to be introduced, it will be clearly indicated in the claim, and will not be displayed in the absence of such contents. For example, to aid understanding, the following claims may contain the phrases "at least one" and "one or more" to introduce the contents of the claims. However, the use of these phrases should not be understood to imply the use of the indefinite article "a" or "an" to introduce the claim contents and restrict any particular claim. Even when the same claim includes the introductory phrases "one or more" or "at least one", the indefinite article, such as "a" or "an", should be interpreted to mean at least one or more, and the same is true for the use of the explicit description used to introduce the claim. In addition, even if the introductory content explicitly refers to a specific number, a person of ordinary skill in the art will recognize that such content should be interpreted as indicating the number of references, for example, "two references" without other modifications means at least two references, or two or more references. In addition, when a statement similar to "at least one of A, B, and C" is used, it is usually stated so that a person of ordinary skill in the art can understand the statement, for example, "a system includes at least one of A, B, and C" will include but is not limited to a system with A alone, a system with B alone, a system with C alone, a system with A and B, a system with A and C, a system with B and C, and / or a system with A, B, and C, etc. A person of ordinary skill in the art will further understand that any separated words and / or phrases represented by two or more alternative terms, whether in the specification, the claims, or the drawings, should be understood to include the possibility of one of these terms, one of them, or both of these terms. For example, "A or B" should be understood as the possibility of "A", or "B", or "A and B".

[0142] As can be seen from the foregoing, various embodiments have been described for illustrative purposes, and various modifications may be made without departing from the scope and spirit of the invention. Therefore, the various embodiments disclosed herein are not intended to be limiting, and the true scope and application are indicated by the claims.

Claims

1. A video encoding and decoding method, comprising: Receiving data of a pixel block, wherein the data of the pixel block is to be encoded or decoded into a current block of a current picture of a video; deriving a partition structure for splitting the current block by identifying a partition location with a minimum cost; Splitting the current block into a first partition and a second partition according to the identified partition position; as well as The first partition and the second partition of the current block are encoded or decoded.

2. The video encoding and decoding method according to claim 1, characterized in that: The width and height of the current block are multiple integers of a power of two, and the partition positions are non-power-of-two integer pixel positions.

3. The video encoding and decoding method according to claim 1, characterized in that: A plurality of samples of the first partition or the second partition are encoded as prediction residuals without being transformed.

4. The video encoding and decoding method according to claim 1, characterized in that: A plurality of samples of the first partition or the second partition are encoded and decoded by prediction without sending a prediction residual.

5. The video encoding and decoding method according to claim 1, characterized in that: The first partition and the second partition are associated with a first template and a second template, respectively, the first template and the second template are constructed based on a plurality of reconstructed pixels adjacent to the current block, wherein identifying the partition position comprises: calculating a first cost based on the first template and calculating a second cost based on the second template; The partition location is optimized to minimize the sum of the first cost and the second cost.

6. The video encoding and decoding method according to claim 5, characterized in that: Calculating the first cost includes matching the first template to a first set of reference pixels, the first set of reference pixels being a plurality of reference pixels in the current picture identified by a first intra prediction mode of the first partition; as well as Calculating the second cost includes matching the second template to a second set of reference pixels, where the second set of reference pixels is a plurality of reference pixels in the current picture identified by a second intra prediction mode of the second partition.

7. The video encoding and decoding method according to claim 5, characterized in that: Calculating the first cost includes matching the first template to a first set of reference pixels, the first set of reference pixels being a plurality of reference pixels in a first reference picture identified by a first motion vector of the first partition; as well as Calculating the second cost includes matching the second template to a second set of reference pixels, the second set of reference pixels being a plurality of reference pixels in a second reference picture identified by a second motion vector of the second partition.

8. The video encoding and decoding method according to claim 1, characterized in that: The first partition is a first rectangular area at a corner of the current block, and the second partition is an L-shaped remaining area of ​​the current block minus the first partition.

9. The video encoding and decoding method according to claim 1, characterized in that: The first partition is a first rectangular area completely located in the current block, and the second partition is a remaining area of ​​the current block that surrounds the first rectangular area on four sides.

10. The video encoding and decoding method according to claim 1, characterized in that: Further comprising transmitting the identified partition location.

11. The video encoding and decoding method according to claim 1, characterized in that: The partition positions are used to specify partition boundaries of a quadtree, binary tree, or ternary tree.

12. The video encoding and decoding method according to claim 1, characterized in that: The partition structure specifies the offset and angle of the lines used to partition the current block.

13. An electronic device, comprising: The video codec circuit is configured to perform a plurality of operations including: Receiving data of a pixel block, wherein the data of the pixel block is to be encoded or decoded into a current block of a current picture of a video; deriving a partition structure for splitting the current block by identifying a partition location with a minimum cost; dividing the current block into a first partition and a second partition according to the identified partition position; and The first partition and the second partition of the current block are encoded or decoded.

14. A video decoding method, comprising: Receiving data of a pixel block, wherein the data of the pixel block is to be decoded into a current block of a current picture of a video; deriving a partition structure for splitting the current block by identifying a partition location with a minimum cost; Splitting the current block into a first partition and a second partition according to the identified partition position; as well as The first partition and the second partition of the current block are reconstructed.

15. A video encoding method, comprising: Receiving data of a pixel block, wherein the data of the pixel block is to be encoded as a current block of a current picture of a video; deriving a partition structure for splitting the current block by identifying a partition location with a minimum cost; Splitting the current block into a first partition and a second partition according to the identified partition position; as well as The first partition and the second partition of the current block are encoded.