Neural Network-Based Filtering Process for Multiple Color Components in Video Coding

Abstract

A method for processing video data includes receiving a syntax element defining a filtering mode for a neural network (NN) model for both a first color component and a second color component, applying an instance of the NN model to a first block of the first color component in the defined filtering mode to generate a first filtered block, and storing the first filtered block for a coding unit (CU).

Description

Technical Field

[0001] This application claims priority to U.S. Patent Application No. 18 / 331,674, filed Jun. 8, 2023, and U.S. Provisional Patent Application No. 63 / 367,713, filed Jul. 5, 2022, the entire contents of which are incorporated herein by reference. U.S. Patent Application No. 18 / 331,674, filed Jun. 8, 2023, claims the benefit of U.S. Provisional Patent Application No. 63 / 367,713, filed Jul. 5, 2022.

[0002] This disclosure relates to video encoding and video decoding.

Background Art

[0003] Digital video capabilities can be incorporated into a wide range of devices including digital televisions, digital direct broadcast systems, wireless broadcast systems, personal digital assistants (PDAs), laptop or desktop computers, tablet computers, e-book readers, digital cameras, digital recording devices, digital media players, video gaming devices, video game consoles, cellular or satellite radiotelephones, so-called "smartphones", video teleconferencing devices, video streaming devices, etc. Digital video devices implement video coding techniques such as those described in standards defined by MPEG-2, MPEG-4, ITU-T H.263, ITU-T H.264 / MPEG-4, Part 10, Advanced Video Coding (AVC), ITU-T H.265 / High Efficiency Video Coding (HEVC), ITU-T H.266 / Versatile Video Coding (VVC), and extensions of such standards, as well as proprietary video codecs / formats such as AOMedia Video1 (AV1) developed by the Alliance for Open Media. Video devices may more efficiently transmit, receive, encode, decode, and / or store digital video information by implementing such video coding techniques.

[0004] Video coding techniques include spatial (intra-picture) prediction and / or temporal (inter-picture) prediction to reduce or remove redundancy inherent in a video sequence. In the case of block-based video coding, a video slice (e.g., a video picture or a portion of a video picture) may be partitioned into video blocks, which may also be referred to as coding tree units (CTUs), coding units (CUs), and / or coding nodes. Video blocks within an intra-coded (I) slice of a picture are encoded using spatial prediction with respect to reference samples in adjacent blocks within the same picture. Video blocks within an inter-coded (P or B) slice of a picture may use spatial prediction with respect to reference samples in adjacent blocks within the same picture or temporal prediction with respect to reference samples in other reference pictures. A picture may sometimes be referred to as a frame, and a reference picture may sometimes be referred to as a reference frame. SUMMARY OF THE INVENTION MEANS FOR SOLVING THE PROBLEM

[0005] Generally, the present disclosure describes techniques for a filtering process of a distorted picture. The filtering process may be based on machine learning techniques such as neural network technology. Exemplary techniques may be used in the context of advanced video coders such as the Versatile Video Coding (VVC) standard, extensions of VVC, next-generation video coding standards, or anything compliant with any other video codec.

[0006] As will be described in more detail, the present disclosure describes an example of neural network-based filtering in which color components (e.g., chroma components) of video data share the same neural network mode. For example, instead of a video encoder and a video decoder selecting different neural network modes for two chroma components, the video encoder and the video decoder may select the same neural network mode for the two chroma components. In this way, the video encoder and the video decoder may reduce the complexity and processing time because they may perform fewer filter processing selections and / or executions of filter processing compared to the case where different neural network modes for two chroma components may be selected.

[0007] In one example, the present disclosure describes a method of processing video data, the method comprising receiving a syntax element that defines a filter processing mode for a neural network (NN) model for both a first color component and a second color component; applying an instance of the NN model to a first block of the first color component in the defined filter processing mode to generate a first filtered block; and storing the first filtered block for a coding unit (CU).

[0008] In one example, the present disclosure describes a device for processing video data, the device comprising a memory configured to store video data; and one or more processors implemented in a circuit and coupled to the memory, the one or more processors configured to receive a syntax element that defines a filter processing mode for a neural network (NN) model for both a first color component and a second color component, apply an instance of the NN model to a first block of the first color component in the defined filter processing mode to generate a first filtered block, and store the first filtered block in the memory for a coding unit (CU).

[0009] In one example, the present disclosure describes a computer-readable storage medium storing instructions that, when executed, cause one or more processors to receive syntax elements that define a filter processing mode for a neural network (NN) model for both a first color component and a second color component, apply an instance of the NN model in the defined filter processing mode to a first block of the first color component to generate a first filtered block, and store the first filtered block for a coding unit (CU).

[0010] Details of one or more examples are set forth in the accompanying drawings and the description below. Other features, objects, and advantages will be apparent from the description, drawings, and claims. BRIEF DESCRIPTION OF THE DRAWINGS

[0011]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

DETAILED DESCRIPTION OF THE INVENTION

[0012] Video coding techniques include filtering for reducing coding artifacts. In some examples, the filtering techniques can be machine learning-based filtering techniques. For example, the machine learning techniques can be implemented using a neural network. For example, in neural network (NN)-based filtering, the reconstructed sample is the input, and the intermediate output is the residual sample that is subtracted in reverse from the input to refine the input sample.

[0013] Video data can include multiple color components, such as red-green-blue (RGB), or luma and two chroma components (YCbCr). In some cases, there can be different NN-based filters for each of the color components. However, having different NN-based filters for each of the color components can be computationally and processing time intensive.

[0014] In some cases, it may be possible to utilize a single neural network model using different NN model modes. That is, in these cases, a first filter processing mode for the NN model is used to filter a first color component, and a second filter processing mode for the NN model is used to filter a second color component. A video encoder and / or a video decoder may select different modes for each of the two chroma components, which may require the execution of two filter processes on the decoder side. For example, in such an example, the video encoder and the video decoder may execute the NN model in the first filter processing mode to filter a first block of the first color component, and then execute the second NN model in the second filter processing mode to filter a second block of the second color component. Such a double execution of the NN model for filtering can be computationally intensive.

[0015] According to one or more examples described in the present disclosure, due to the similarity between different color components (e.g., similarly, between two chroma components), a video encoder and a video decoder may utilize a filter control mechanism that shares the same neural network model selection for two or more color components (e.g., chroma components). For example, the video encoder and the video decoder may apply an instance of the NN model in a defined filter processing mode to a first block of a first color component, and apply the same instance of the NN model in the defined filter processing mode to a second block of a second color component. In addition to the chroma components (e.g., the first and second color components), the input to the NN model may be the luma component, but the output of the NN model may be the filtered chroma component and not the filtered luma component (however, in some examples, it is possible to output the filtered luma component).

[0016] (For example, when NN model filtering is enabled for both color components), to ensure that the NN model is applied using the same filter processing mode for both the first block of the first color component and the second block of the second color component, the video encoder may signal a syntax element that defines the filter processing mode for the NN model for both the first color component and the second color component, and the video decoder may receive it. In this example, the video decoder may set the filter processing mode for both the first color component and the second color component to be the same filter processing mode.

[0017] In this way, the video encoder and the video decoder can generate filtered blocks using one execution instance of the NN model. By using the exemplary techniques described in this disclosure, the computational complexity can be reduced without sacrificing rate-distortion performance as much as possible compared to techniques where different neural network modes are possible for different chroma components.

[0018] FIG. 1 is a block diagram showing an exemplary video encoding and decoding system 100 in which the techniques of this disclosure may be implemented. The techniques of this disclosure are generally directed to coding (encoding and / or decoding) video data. Generally, video data includes any data for processing video. Thus, video data may include raw unencoded video, encoded video, decoded (e.g., reconstructed) video, and video metadata such as signaling data.

[0019] As shown in FIG. 1, in this example, the system 100 includes a source device 102 that provides encoded video data to be decoded and displayed by a destination device 116. Specifically, the source device 102 provides video data to the destination device 116 via a computer-readable recording medium 110. The source device 102 and the destination device 116 may include any of a wide range of devices, such as desktop computers, notebook (i.e., laptop) computers, mobile devices, tablet computers, set-top boxes, telephone handsets such as smartphones, televisions, cameras, display devices, digital media players, video gaming consoles, video streaming devices, broadcast receiver devices, etc. In some cases, since the source device 102 and the destination device 116 may support wireless communication, they may be referred to as wireless communication devices.

[0020] In the example of FIG. 1, the source device 102 includes a video source 104, a memory 106, a video encoder 200, and an output interface 108. The destination device 116 includes an input interface 122, a video decoder 300, a memory 120, and a display device 118. According to the present disclosure, the video encoder 200 of the source device 102 and the video decoder 300 of the destination device 116 may be configured to apply techniques for a machine learning (e.g., neural network)-based filtering process for a plurality of color components in video coding. Thus, the source device 102 represents an example of a video encoding device, while the destination device 116 represents an example of a video decoding device. In other examples, the source device and the destination device may include other components or configurations. For example, the source device 102 may receive video data from an external video source such as an external camera. Similarly, the destination device 116 may interface with an external display device rather than including an integrated display device.

[0021] System 100 as shown in FIG. 1 is just an example. Generally, any digital video encoding and / or decoding device can perform techniques for a machine learning (e.g., neural network)-based filtering process on multiple color components in video coding. Source device 102 and destination device 116 are just examples of coding devices that generate encoded video data that source device 102 transmits to destination device 116. This disclosure refers to a device that performs data coding (encoding and / or decoding) as a "coding" device. Thus, video encoder 200 and video decoder 300 represent examples of coding devices, specifically, a video encoder and a video decoder, respectively. In some examples, source device 102 and destination device 116 can operate in a substantially symmetric manner such that each of source device 102 and destination device 116 includes video encoding and decoding components. Thus, system 100 can support one-way or two-way video transmission between source device 102 and destination device 116, for example, for video streaming, video playback, video broadcasting, or video telephony.

[0022] Generally, video source 104 represents a source of video data (i.e., raw, unencoded video data), provides a continuous series of pictures of video data (also referred to as "frames") to video encoder 200, and video encoder 200 encodes the data of the pictures. The video source 104 of source device 102 can include a video capture device such as a video camera, a video archive containing previously captured raw video, and / or a video feed interface that receives video from a video content provider. As a further alternative, video source 104 can generate computer graphics-based data as source video, or a combination of live video, archived video, and computer-generated video. In each case, video encoder 200 encodes the captured video data, pre-captured video data, or computer-generated video data. Video encoder 200 can reorder the pictures from the received order (sometimes referred to as "display order") to the coding order for coding. Video encoder 200 can generate a bitstream containing the encoded video data. Source device 102 can then output the encoded video data via output interface 108 to computer-readable recording medium 110 for reception and / or extraction, for example, by input interface 122 of destination device 116.

[0023] The memories 106 of the source device 102 and 120 of the destination device 116 represent general-purpose memories. In some examples, the memories 106, 120 may store raw video data, such as raw video from the video source 104 and raw decoded video data from the video decoder 300. Additionally or alternatively, the memories 106, 120 may store software instructions executable by, for example, the video encoder 200 and the video decoder 300, respectively. Although the memories 106 and 120 are shown separately from the video encoder 200 and the video decoder 300 in this example, it should be understood that the video encoder 200 and the video decoder 300 may also include internal memories for functionally similar or equivalent purposes. Further, the memories 106, 120 may store encoded video data output from the video encoder 200 and input to the video decoder 300. In some examples, portions of the memories 106, 120 may be allocated as one or more video buffers to store, for example, raw decoded video data and / or encoded video data.

[0024] The computer-readable recording medium 110 can represent any type of medium or device capable of transferring the encoded video data from the source device 102 to the destination device 116. In one example, the computer-readable recording medium 110 represents a communication medium that enables the source device 102 to directly transmit the encoded video data to the destination device 116 in real time, for example, via a radio frequency network or a computer-based network. The output interface 108 may modulate the transmission signal including the encoded video data, and the input interface 122 may demodulate the received transmission signal according to a communication standard such as a wireless communication protocol. The communication medium may include any wireless or wired communication medium, such as the radio frequency (RF) spectrum or one or more physical transmission lines. The communication medium may form part of a packet-based network, such as a local area network, a wide area network, or a global network such as the Internet. The communication medium may include routers, switches, base stations, or any other device that may be useful for facilitating communication from the source device 102 to the destination device 116.

[0025] In some examples, the source device 102 may output the encoded data from the output interface 108 to the storage device 112. Similarly, the destination device 116 may access the encoded data from the storage device 112 via the input interface 122. The storage device 112 may include any of a variety of distributed data storage media or locally accessible data storage media, such as a hard drive, a Blu-ray disk, a DVD, a CD-ROM, a flash memory, a volatile or non-volatile memory, or any other suitable digital storage medium for storing the encoded video data.

[0026] In some examples, the source device 102 may output the encoded video data to a file server 114 or another intermediate storage device that may store the encoded video data generated by the source device 102. The destination device 116 may access the stored video data from the file server 114 via streaming or downloading.

[0027] The file server 114 can be any type of server device capable of storing the encoded video data and transmitting the encoded video data to the destination device 116. The file server 114 can represent a web server (e.g., for a website), a server configured to provide a file transfer protocol service (such as the File Transfer Protocol (FTP) or the File Delivery over Unidirectional Transport (FLUTE) protocol), a content delivery network (CDN) device, a hypertext transfer protocol (HTTP) server, a Multimedia Broadcast Multicast Service (MBMS) or Enhanced MBMS (eMBMS) server, and / or a network attached storage (NAS) device. Additionally or alternatively, the file server 114 can implement one or more HTTP streaming protocols such as Dynamic Adaptive Streaming over HTTP (DASH), HTTP Live Streaming (HLS), the Real Time Streaming Protocol (RTSP), HTTP dynamic streaming, etc.

[0028] The destination device 116 can access the encoded video data from the file server 114 through any standard data connection including an Internet connection. This can include a wireless channel (e.g., Wi-Fi connection), a wired connection (e.g., digital subscriber line (DSL), cable modem, etc.), or a combination of both suitable for accessing the encoded video data stored on the file server 114. The input interface 122 can be configured to operate according to any one or more of the various protocols described above for retrieving or receiving media data from the file server 114, or any other such protocol for retrieving media data.

[0029] The output interface 108 and the input interface 122 may represent a wireless transmitter / receiver, a modem, a wired network component (e.g., an Ethernet card), a wireless communication component operating according to any of various IEEE 802.11 standards, or other physical components. In examples where the output interface 108 and the input interface 122 include wireless components, the output interface 108 and the input interface 122 may be configured to transfer data such as encoded video data according to cellular communication standards such as 4G, 4G-LTE (Long Term Evolution), LTE Advanced, 5G, etc. In some examples where the output interface 108 includes a wireless transmitter, the output interface 108 and the input interface 122 may be configured to transfer data such as encoded video data according to other wireless standards such as the IEEE 802.11 specification, the IEEE 802.15 specification (e.g., ZigBee (trademark)), the Bluetooth (trademark) standard, etc. In some examples, the source device 102 and / or the destination device 116 may each include a system-on-a-chip (SoC) device. For example, the source device 102 may include an SoC device that implements functions resulting from the video encoder 200 and / or the output interface 108, and the destination device 116 may include an SoC device that implements functions resulting from the video decoder 300 and / or the input interface 122.

[0030] The techniques of the present disclosure may be applied to video coding that supports any of various multimedia applications such as over-the-air television broadcast, cable television transmission, satellite television transmission, Internet streaming video transmission such as HTTP-based Dynamic Adaptive Streaming over HTTP (DASH), digital video encoded on a data storage medium, decoding of digital video stored on a data storage medium, or other applications.

[0031] The input interface 122 of the destination device 116 receives an encoded video bitstream from a computer-readable recording medium 110 (e.g., a communication medium, a storage device 112, a file server 114, etc.). The encoded video bitstream may include signaling information defined by the video encoder 200 that is also used by the video decoder 300, such as syntax elements having values that describe the characteristics and / or processing of video blocks or other coded units (e.g., slices, pictures, groups of pictures, sequences, etc.). The display device 118 displays the decoded pictures of the decoded video data to the user. The display device 118 may represent any of a variety of display devices, such as a liquid crystal display (LCD), a plasma display, an organic light emitting diode (OLED) display, or another type of display device.

[0032] Although not shown in FIG. 1, in some examples, the video encoder 200 and the video decoder 300 may each be integrated with an audio encoder and / or an audio decoder and may include an appropriate MUX-DEMUX unit, or other hardware and / or software, to handle a multiplexed stream that includes both audio and video in a common data stream.

[0033] The video encoder 200 and the video decoder 300 may each be implemented as any of a variety of suitable encoder and / or decoder circuits, such as one or more microprocessors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs), discrete logic, software, hardware, firmware, or any combination thereof. When the techniques are implemented partially in software, the device may store instructions for the software in a suitable non-transitory computer-readable recording medium and execute the instructions in hardware using one or more processors to perform the techniques of the present disclosure. Each of the video encoder 200 and the video decoder 300 may be included within one or more encoders or decoders, and any of them may be integrated as part of a combined encoder / decoder (CODEC) in their respective devices. Devices including the video encoder 200 and / or the video decoder 300 may include integrated circuits, microprocessors, and / or wireless communication devices such as cellular telephones.

[0034] The video encoder 200 and the video decoder 300 may operate according to a video coding standard such as ITU-T H.265, also known as High Efficiency Video Coding (HEVC), or an extended standard thereof such as multi-view and / or scalable video coding extensions. Alternatively, the video encoder 200 and the video decoder 300 may operate according to other proprietary or industry standards, such as ITU-T H.266, also known as Versatile Video Coding (VVC). In other examples, the video encoder 200 and the video decoder 300 may operate according to a proprietary video codec / format, such as AOMedia Video1 (AV1), an extension of AV1, and / or a successor version of AV1 (e.g., AV2). In other examples, the video encoder 200 and the video decoder 300 may operate according to other proprietary formats or industry standards. However, the techniques of the present disclosure are not limited to any particular coding standard or format. Generally, the video encoder 200 and the video decoder 300 may be configured to execute the techniques of the present disclosure in combination with any video coding technique that uses filtering, such as machine learning (e.g., neural network)-based filtering of video data.

[0035] Generally, video encoder 200 and video decoder 300 may perform block - based coding of pictures. The term "block" generally refers to a structure that contains data to be processed (e.g., encoded, decoded, or otherwise used in the encoding and / or decoding process). For example, a block may include a two - dimensional matrix of samples of luminance and / or chrominance data. Generally, video encoder 200 and video decoder 300 may code video data represented in a YUV (e.g., Y, Cb, Cr) format. That is, instead of coding red, green, and blue (RGB) data for the samples of a picture, video encoder 200 and video decoder 300 may code the luminance component and the chrominance component, and the chrominance component may include both the red and blue chrominance components. In some examples, video encoder 200 converts the received RGB - format data to a YUV representation before encoding, and video decoder 300 converts the YUV representation to an RGB format. Alternatively, a pre - processing unit and a post - processing unit (not shown) may perform these conversions.

[0036] This disclosure may refer to the coding (e.g., encoding and decoding) of pictures as generally including the process of encoding or decoding the data of the pictures. Similarly, this disclosure may refer to the coding of blocks of pictures as including the process of encoding or decoding the data for the blocks, e.g., prediction and / or residual coding. An encoded video bitstream generally includes a series of values of syntax elements that represent coding decisions (e.g., coding modes) and the partitioning of the picture into blocks. Thus, a reference to coding a picture or a block should generally be understood as the coding values of the syntax elements that form the picture or the block.

[0037] HEVC defines various blocks, including coding units (CUs), prediction units (PUs), and transform units (TUs). According to HEVC, a video coder (such as video encoder 200) divides a coding tree unit (CTU) into CUs according to a quadtree structure. That is, the video coder divides the CTU and CU into four equal and non-overlapping squares, and each node of the quadtree has either zero or four child nodes. A node without child nodes may be called a "leaf node", and the CU of such a leaf node may include one or more PUs and / or one or more TUs. The video coder may further divide the PUs and TUs. For example, in HEVC, a residual quadtree (RQT) represents the division of TUs. In HEVC, a PU represents inter-prediction data and a TU represents residual data. An intra-predicted CU includes intra-prediction information such as an intra-mode indication.

[0038] As another example, video encoder 200 and video decoder 300 may be configured to operate according to VVC. According to VVC, a video coder (such as video encoder 200) divides a picture into a plurality of coding tree units (CTUs). Video encoder 200 may divide the CTU according to a tree structure such as a quadtree-binary tree (QTBT) structure or a multi-type tree (MTT) structure. The QTBT structure eliminates the concept of multiple division types, such as the separation between the CUs, PUs, and TUs of HEVC. The QTBT structure includes two levels, that is, a first level divided according to quadtree division and a second level divided according to binary tree division. The root node of the QTBT structure corresponds to the CTU. The leaf node of the binary tree corresponds to the coding unit (CUs).

[0039] In the MTT partitioning structure, blocks can be partitioned using a quadtree (QT) partition, a binary tree (BT) partition, and one or more types of triple trees (also referred to as ternary trees, TT). A triple tree partition or ternary tree partition is a partition in which a block is divided into three sub-blocks. In some examples, the triple tree partition or ternary tree partition divides the block into three sub-blocks without dividing the original block through the center. The partitioning types (e.g., QT, BT, and TT) in the MTT can be symmetric or asymmetric.

[0040] When operating according to the AV1 codec, the video encoder 200 and the video decoder 300 can be configured to code video data within a block. In AV1, the largest coding block that can be processed is called a superblock. In AV1, a superblock can be either 128×128 luma samples or 64×64 luma samples. However, in a successor video coding format (e.g., AV2), a superblock can be defined by a different (e.g., larger) luma sample size. In some examples, a superblock is the top level of a block quadtree. The video encoder 200 can further partition a superblock into smaller coding blocks. The video encoder 200 can use square or non-square partitioning to partition a superblock and other coding blocks into smaller blocks. Non-square blocks can include N / 2×N, N×N / 2, N / 4×N, and N×N / 4 blocks. The video encoder 200 and the video decoder 300 can perform separate prediction and transformation processes for each of the coding blocks.

[0041] AV1 also defines tiles of video data. A tile is a rectangular array of super blocks that can be coded independently of other tiles. That is, the video encoder 200 and the video decoder 300 can encode and decode the coding blocks within a tile, respectively, without using video data from other tiles. However, the video encoder 200 and the video decoder 300 can perform filtering across tile boundaries. Tiles may have uniform or non-uniform sizes. Tile-based coding may enable parallel processing and / or multi-threading for encoder and decoder implementations.

[0042] In some examples, the video encoder 200 and the video decoder 300 may use a single QTBT structure or MTT structure to represent each of the luminance and chrominance components. In other examples, the video encoder 200 and the video decoder 300 may use two or more QTBT or MTT structures, such as one QTBT / MTT structure for the luminance component and another QTBT / MTT structure for both chrominance components (or two QTBT / MTT structures for each chrominance component).

[0043] The video encoder 200 and the video decoder 300 may be configured to use quadtree partitioning, QTBT partitioning, MTT partitioning, super block partitioning, or other partitioning structures.

[0044] In some examples, a CTU includes a coding tree block (CTB) of luma samples, two corresponding CTBs of chroma samples of a picture having three sample arrays, or a CTB of samples of a picture coded using three separate color planes and syntax structures used to code monochrome pictures or samples. A CTB may be an N×N block of samples for some value of N that divides the components into CTBs. A component may be a single sample from one of the one array or three arrays (luma and two chroma) that make up a picture in a 4:2:0, 4:2:2, or 4:4:4 color format, or a single sample of an array that makes up a picture in an array or monochrome format. In some examples, a coding block is an M×N block of samples for some values of M and N that divides the CTB into coding blocks.

[0045] Blocks (e.g., CTUs or CUs) may be grouped in various ways within a picture. As an example, a block may refer to a rectangular region of CTU rows within a particular tile in the picture. A tile may be a rectangular region of CTUs within a particular tile column and a particular tile row in the picture. A tile column refers to a rectangular region of CTUs having a height equal to the height of the picture and a width specified by a syntax element (such as in a picture parameter set). A tile row refers to a rectangular region of CTUs having a height specified by a syntax element (such as in a picture parameter set) and a width equal to the width of the picture.

[0046] In some examples, a tile may be divided into a plurality of blocks, and each of the blocks may include one or more CTU rows within the tile. A tile that is not divided into a plurality of blocks may also be referred to as a block. However, a block that is a true subset of a tile may sometimes not be referred to as a tile. Blocks within a picture may also be arranged as slices. A slice may be an integer number of blocks of a picture that may be exclusively included within a single network abstraction layer (NAL) unit. In some examples, a slice may include either some complete tiles or only a continuous sequence of complete blocks of one tile.

[0047] This disclosure may interchangeably use "N×N" and "N times N", e.g., 16×16 samples or 16 times 16 samples, to refer to the sample dimensions of a block (such as a CU or other video block) in the vertical and horizontal dimensions. Generally, a 16×16 CU has 16 samples in the vertical direction (y = 16) and 16 samples in the horizontal direction (x = 16). Similarly, an N×N CU generally has N samples in the vertical direction and N samples in the horizontal direction, where N represents a non-negative integer value. Samples within a CU may be arranged in rows and columns. Moreover, a CU does not necessarily have to have the same number of samples in the horizontal direction as in the vertical direction. For example, a CU may include N×M samples, and M does not necessarily have to be equal to N.

[0048] Video encoder 200 encodes video data for a CU that represents prediction and / or residual information, as well as other information. Prediction information indicates how the CU is to be predicted to form a prediction block for the CU. Residual information generally represents the sample-by-sample difference between the samples of the CU before encoding and the prediction block.

[0049] To predict a CU, video encoder 200 may generally form a prediction block for the CU through inter prediction or intra prediction. Inter prediction generally refers to predicting a CU from data of a previously coded picture, and intra prediction generally refers to predicting a CU from data coded previously in the same picture. To perform inter prediction, video encoder 200 may use one or more motion vectors to generate a prediction block. Video encoder 200 may generally perform a motion search to identify a reference block that exactly matches the CU with respect to, for example, the difference between the CU and the reference block. Video encoder 200 may calculate a difference metric using sum of absolute difference (SAD), sum of squared differences (SSD), mean absolute difference (MAD), mean squared differences (MSD), or other such difference calculations to determine whether the reference block exactly matches the current CU. In some examples, video encoder 200 may predict the current CU using uni-directional prediction or bi-directional prediction.

[0050] Some examples of VVC may also provide an affine motion compensation mode that can be regarded as an inter prediction mode. In the affine motion compensation mode, video encoder 200 may determine two or more motion vectors representing non-translational motion, such as zoom in or zoom out, rotation, perspective movement, or other irregular motion types.

[0051] To perform intra prediction, video encoder 200 may select an intra prediction mode to generate a prediction block. Some examples of VVC provide 67 intra prediction modes including various direction modes, as well as a planar mode and a DC mode. Generally, video encoder 200 selects an intra prediction mode that describes adjacent samples for the current block (e.g., a block of a CU) and predicts the samples of the current block therefrom. Assuming that video encoder 200 codes CTUs and CUs in raster scan order (left to right, top to bottom), such samples can generally be above, above and to the left, or to the left of the current block in the same picture as the current block.

[0052] Video encoder 200 codes data representing the prediction mode for the current block. For example, in the case of inter prediction mode, video encoder 200 may code data representing which of the various available inter prediction modes is used, as well as motion information for the corresponding mode. In the case of uni - directional or bi - directional inter prediction, for example, video encoder 200 may code motion vectors using an advanced motion vector prediction (AMVP) mode or a merge mode. Video encoder 200 may use a similar mode to code motion vectors for an affine motion compensation mode.

[0053] AV1 includes two general techniques for encoding and decoding coding blocks of video data. The two general techniques are intra prediction (e.g., intra-frame prediction or spatial prediction) and inter prediction (e.g., inter-frame prediction or temporal prediction). In the context of AV1, when predicting a block in the current frame of video data using an intra prediction mode, video encoder 200 and video decoder 300 do not use video data from other frames of the video data. In most intra prediction modes, video encoder 200 encodes a block in the current frame based on the difference between the sample values within the current block and the predicted values generated from reference samples within the same frame. Video encoder 200 determines the predicted values generated from reference samples based on the intra prediction mode.

[0054] Following prediction such as intra prediction or inter prediction of a block, video encoder 200 may calculate the residual data of the block. Residual data, such as a residual block, represents the sample-by-sample difference between a block and its predicted block formed using the corresponding prediction mode. Video encoder 200 may apply one or more transforms to the residual block to generate transform data in the transform domain rather than the sample domain. For example, video encoder 200 may apply a discrete cosine transform (DCT), an integer transform, a wavelet transform, or a conceptually similar transform to the residual video data. Additionally, video encoder 200 may apply a secondary transform, such as a mode-dependent non-separable secondary transform (MDNSST), a signal-dependent transform, a Karhunen-Loeve transform (KLT), etc., following the first transform. Video encoder 200 generates transform coefficients following the application of one or more transforms.

[0055] As described above, following any transformation that generates transformation coefficients, video encoder 200 may perform quantization of the transformation coefficients. Quantization generally refers to the process of quantizing the transformation coefficients to reduce, to the extent possible, the amount of data used to represent the transformation coefficients and perform further compression. By performing the quantization process, video encoder 200 may reduce the bit depth associated with some or all of the transformation coefficients. For example, video encoder 200 may truncate an n-bit value to an m-bit value during quantization, where n is greater than m. In some examples, to perform quantization, video encoder 200 may perform a right shift of each bit of the value to be quantized.

[0056] Following quantization, video encoder 200 may scan the transformation coefficients to generate a one-dimensional vector from the two-dimensional matrix containing the quantized transformation coefficients. The scan may be designed to place transformation coefficients with higher energy (and thus lower frequency) at the front of the vector and transformation coefficients with lower energy (and thus higher frequency) at the back of the vector. In some examples, video encoder 200 may generate a serialized vector and then use a predefined scan order to scan the quantized transformation coefficients in order to entropy code the quantized transformation coefficients of the vector. In other examples, video encoder 200 may perform an adaptive scan. After scanning the quantized transformation coefficients to form a one-dimensional vector, video encoder 200 may entropy code the one-dimensional vector, for example, according to context-adaptive binary arithmetic coding (CABAC). Video encoder 200 may also entropy code the values for syntax elements that describe metadata associated with the encoded video data that video decoder 300 uses when decoding the video data.

[0057] To perform CABAC, the video encoder 200 may assign a context within the context model to the symbol to be transmitted. The context may be related to, for example, whether the adjacent value of the symbol is zero-valued. The probability determination may be based on the context assigned to the symbol.

[0058] The video encoder 200 may further generate syntax data such as block-based syntax data, picture-based syntax data, and sequence-based syntax data for the video decoder 300, for example, within other syntax data such as a picture header, a block header, a slice header, or a sequence parameter set (SPS), a picture parameter set (PPS), or a video parameter set (VPS). The video decoder 300 may similarly decode such syntax data to determine how to decode the corresponding video data.

[0059] In this way, the video encoder 200 may generate a bitstream including encoded video data, for example, syntax elements describing the partitioning of a picture into blocks (e.g., CUs) and prediction information and / or residual information for the blocks. Finally, the video decoder 300 may receive the bitstream and decode the encoded video data.

[0060] Generally, video decoder 300 performs a reverse process to the process performed by video encoder 200 to decode the encoded video data of the bitstream. For example, video decoder 300 may use CABAC to decode the values for the syntax elements of the bitstream in a substantially similar manner that is the reverse of the CABAC encoding process of video encoder 200. The syntax elements may define the partitioning of the picture into CTUs, and the partitioning information for each CTU according to the corresponding partitioning structure such as the QTBT structure, to define the CUs of the CTUs. The syntax elements may further define the prediction information and residual information for blocks (e.g., CUs) of the video data.

[0061] The residual information may be represented, for example, by quantized transform coefficients. Video decoder 300 may inverse quantize and inverse transform the quantized transform coefficients of the block to reproduce the residual block for the block. Video decoder 300 uses the signaling prediction mode (intra prediction or inter prediction) and the related prediction information (e.g., motion information for inter prediction) to form the prediction block for the block. Video decoder 300 can then combine the prediction block and the residual block (sample by sample) to reproduce the original block. Video decoder 300 may perform additional processes such as performing a deblocking process to reduce visual artifacts along the boundaries of the block.

[0062] This disclosure may generally refer to "signaling" some information such as a syntax element. The term "signaling" may generally refer to the communication of values for syntax elements and / or other data used to decode coded video data. That is, video encoder 200 may signal a value for a syntax element within a bitstream. Generally, signaling refers to generating a value within a bitstream. As described above, source device 102 may transfer the bitstream to destination device 116 substantially in real time or non-real time, which may occur, for example, when storing syntax elements in storage device 112 for later retrieval by destination device 116.

[0063] As described above, video coding standards include ITU-T H.261, ISO / IEC MPEG-1 Visual, ITU-T H.262 or ISO / IEC MPEG-2 Visual, ITU-T H.263, ISO / IEC MPEG-4 Visual and ITU-T H.264 (also known as ISO / IEC MPEG-4 AVC), High Efficiency Video Coding (HEVC) including its range extensions such as Multi-View Extension (MV-HEVC) and Scalable Extension (SHVC), or ITU-T H.265. The most recent standard, Versatile Video Coding (VVC) or ITU-T H.266, was developed by the Joint Video Expert Team (JVET) of the ITU-T Video Coding Experts Group (VCEG) and the ISO / IEC Moving Picture Experts Group (MPEG). Version 1 of the VVC specification has been finalized this month (referred to as VVC FDIS below) and is available from http: / / phenix.int-evry.fr / jvet / doc_end_user / documents / 19_Teleconference / wg11 / JVET-S2001-v17.zip.

[0064] The video coding standard is based on the so-called hybrid video coding principle, as shown in FIG. 6 and similar to the description of FIG. 2. The term "hybrid" refers to a combination of two techniques for reducing redundancy in video signals, such as prediction and transform coding using quantization of prediction residuals. Prediction and transform reduce redundancy in video signals by decorrelation, while quantization reduces the data in the transform coefficient representation by reducing its precision, ideally removing only irrelevant details. This hybrid video coding design principle is also used in the two most recent standards, HEVC and VVC. As shown in FIG. 6 and described in more detail with respect to FIG. 2, modern hybrid video coders are composed of various processes.

[0065] As shown in FIG. 6, modern hybrid video coder 130 generally performs block partitioning, motion compensation or inter-picture prediction, intra-picture prediction, transform, quantization, entropy coding, and post / loop filter processing. In the example of FIG. 2, video coder 130 includes addition unit 134, transform unit 136, quantization unit 138, entropy coding unit 140, inverse quantization unit 142, inverse transform unit 144, addition unit 146, loop filter unit 148, decoded picture buffer (DPB) 150, intra prediction unit 152, inter prediction unit 154, and motion estimation unit 156.

[0066] Generally, when encoding video data, video coder 130 may receive input video data 132. Block partitioning is used to divide the received picture (image) of video data into smaller blocks for the operation of the prediction process and the transform process. Early video coding standards used a fixed block size, generally 16×16 samples. More recent standards, such as HEVC and VVC, use a tree-based partitioning structure for flexible partitioning.

[0067] The motion estimation unit 156 and the inter prediction unit 154 can predict the input video data 132, for example, from the previously decoded data in the DPB 150. Motion compensation or inter-picture prediction utilizes the redundancy that exists between pictures in a video sequence (thus "inter"). According to block-based motion compensation used in all modern video codecs, the prediction is obtained from one or more previously decoded pictures, i.e., reference pictures. The corresponding area for generating the inter prediction is indicated by motion information including a motion vector and a reference picture index. In recent video codecs, a hierarchical prediction structure within a group of pictures (GOP) is applied to improve coding efficiency. An example of a GOP 700 having a size equal to 16 is shown in FIG. 7.

[0068] The addition unit 134 can calculate residual data as the difference between the input video data 132 and the predicted data from the intra prediction unit 152 or the inter prediction unit 154. The addition unit 134 provides the residual block to the transform unit 136, and the transform unit 136 applies one or more transforms to the residual block to generate a transform block. The quantization unit 138 quantizes the transform block to form quantized transform coefficients. The entropy coding unit 140 entropy-codes the quantized transform coefficients and other syntax elements such as motion information or intra prediction information to generate the output bitstream 158.

[0069] On one hand, to reproduce the residual block, the inverse quantization unit 142 inverse quantizes the quantized transform coefficients, and the inverse transform unit 144 inverse transforms the transform coefficients. The addition unit 146 synthesizes the residual block (for each sample) with the prediction block to generate the decoded block of the video data. The loop filter unit 148 applies one or more filters (e.g., at least one of a neural network-based filter, a neural network-based loop filter, a neural network-based post-loop filter, an adaptive in-loop filter, or a pre-defined adaptive in-loop filter) to the decoded block to generate the filtered and decoded block.

[0070] According to the techniques of the present disclosure, the neural network filter processing unit of the loop filter unit 148 may receive data for the decoded picture of the video data from the addition unit 146 and from one or more other units of the hybrid video coder 130, such as the transform unit 136, the quantization unit 138, the intra prediction unit 152, the inter prediction unit 154, the motion estimation unit 156, and / or one or more other filter processing units within the loop filter unit 148. For example, the neural network filter processing unit may receive data from the deblocking filter processing unit (also referred to as the “deblocking unit”) of the loop filter unit 148.

[0071] In FIG. 6, intra picture prediction exploits the spatial redundancy present within the picture by deriving a prediction for a block from already coded / decoded, spatially adjacent (reference) samples. Directional angular prediction, DC prediction, and plane or planar prediction are used in recent video coders including AVC, HEVC, and VVC.

[0072] Transformation: The hybrid video coding standard applies block transformation to the prediction residual (regardless of whether the prediction residual is derived from inter-picture prediction or intra-picture prediction). In early standards including H.261 / 262 / 263, the discrete cosine transform (DCT) is adopted. In HEVC and VVC, additional transform kernels other than DCT are applied to consider different statistical values in specific video signals.

[0073] Quantization aims to reduce the accuracy of the input value or set of input values in order to reduce the amount of data required to represent the value. In hybrid video coding, quantization is typically applied to individual transformed residual samples, i.e., the transform coefficients, to yield integer coefficient levels. In recent video coding standards, the step size is derived from the so-called quantization parameter (QP) that controls the fidelity and the bit rate. A larger step size reduces the bit rate but also degrades the quality, resulting in, for example, the video picture exhibiting blocking artifacts and blurred details.

[0074] Entropy coding: Context-adaptive binary arithmetic coding (CABAC) is used in recent video codecs such as AVC, HEVC, and VVC due to its high efficiency.

[0075] Post / In-loop filter processing is a filter processing process (or a combination of such processes) applied to a reconstructed picture in order to reduce coding artifacts. The input to the filter processing process is generally a reconstructed picture that is a combination of a reconstructed residual signal (including quantization error) and a prediction. That is, the input to the filter processing process is the result of adding the prediction to the residual signal. As shown in FIG. 6, the reconstructed picture after in-loop filter processing is stored and used as a reference for inter-picture prediction of subsequent pictures. Coding artifacts are mostly determined by the QP, and thus, QP information is generally used in the design of the filter processing process. In HEVC, the in-loop filter includes deblocking filter processing and sample adaptive offset (SAO) filter processing. In the VVC standard, an adaptive loop filter (ALF) is introduced as a third filter. The filter processing process of the ALF is as shown below. R’(i,j)=R(i,j)+((Σ k≠0 Σ l≠0 f(k,l)×K(R(i+k,j+l)-R(i,j),c(k,l))+64)≫7) (1) Here, canR(i,j) is the sample before the filter processing process, and R’(i,j) is the sample value after the filter processing process. f(k,l) represents the filter coefficient, K(x,y) is the clipping function, and c(k,l) represents the clipping parameter. The variables k and l are

[0076]

Number

[0077] The following describes neural network (NN)-based filtering for video coding. Incorporating a neural network into a hybrid video coding framework can improve the compression efficiency. The neural network is used in the intra-prediction and inter-prediction modules to improve the prediction efficiency. An NN-based in-loop filter is also possible. In some examples, the filtering process is applied as a post-filter. In this case, the filtering process is applied only to the output picture, and the unfiltered pictures are used as reference pictures.

[0078] An NN-based filter can be applied as an addition to existing filters such as the deblocking filter, SAO, and ALF. It can also be applied exclusively if it is designed to replace all existing filters.

[0079] As shown in FIG. 8, the NN-based filter processing process takes the reconstructed sample 800 as input, and the intermediate output is the residual sample (e.g., residual sample 802), and the residual sample is subtracted inversely from the input (e.g., the reconstructed sample 800) to refine the input sample and generate the filtered sample 804. The NN filter can use all color components as input to utilize the inter-component correlation. Different components can share the same filter (including the network structure and model parameters), or each component can have its own unique filter.

[0080] As an example, the NN filter can receive all color components even if a subset of the color components is filtered. For example, in the case of luma and chroma components, the NN filter can receive the luma block and both chroma blocks. In this example, assume that the Cb chroma block is the object to be filtered, and the luma block or the Cr chroma block is not the object to be filtered. The luma block, the Cb chroma block, and the Cr chroma block are all inputs to the NN filter. In such an example, the NN filter can generate the filtered Cb chroma block and the filtered Cr chroma block. The filtered luma block may not be generated, but the present technique does not necessarily require that the filtered luma block is not generated. However, the video encoder 200 and the video decoder 300 can discard the filtered Cr chroma block. That is, the reconstructed coding unit (CU) including the luma block and the chroma block can include the reconstructed luma samples (not filtered), the reconstructed Cr chroma samples (not filtered), and the filtered reconstructed Cb chroma samples.

[0081] The filter processing process can also be generalized as follows. R’(i,j)=R(i,j)+NN_filter_residual_ouput(R) (3)

[0082] The model structure and model parameters of the NN-based filter can be predefined and stored in the video encoder 200 and the video decoder 300. The filter can also be signaled in the bitstream.

[0083] When NN-based filter processing is applied in video coding, the entire video signal may be divided into a plurality of processing components, and each processing component can be processed separately. Possible options for the processing components include frames, slices / tiles, CTUs, or any predefined or signaled shape and size. That is, the term "processing component" can refer to a portion of a picture or frame on which processing can be performed.

[0084] The following describes NN-based filter processing having a multi-mode design. To further improve the performance of NN-based filter processing, a multi-mode solution can be designed. For example, for each processing component, the video encoder 200 can select from a set of modes based on rate-distortion optimization, and the selection can be signaled in the bitstream. Different modes can include different NN models, different values used as input information for the NN model, etc. As an example, JVET-Z0113: Y. Li, K. Zhang, L. Zhang, H. Wang, M. Coban, A. M. Kotra, M. Karczewicz, F. Galpin, K. Andersson, J. Strom, D. Liu, R. Sjoberg, EE1-1.7: Combined Test of EE1-1.6 and EE1-1.3, JVET-Z0113, April 2022 (hereinafter JVET-Z0113) proposed an NN-based filter processing solution that creates multiple modes based on a single NN model by using different QP values as inputs for the NN models of different modes.

[0085] There may be several issues regarding the use of NN-based filter processing techniques. For a video signal (e.g., video data) having multiple color components, the filter processing processes for different color components may be different. However, in the case of a specific type of filter processing (e.g., neural-network-based filter processing), applying different filters for each color component may significantly increase the computational complexity and memory requirements. In these cases, by designing filters that can include multiple color components and applying appropriate control over the way these filters are applied in a video codec, a better trade-off between compression performance and complexity cost can be achieved.

[0086] For example, in JVET-Z0113, an NN-based filter processing solution is proposed that has multiple modes using different QP values for different modes that a video encoder 200 and a video decoder 300 can select for processing components. In such a technique of JVET-Z0113, there is only one model that can be used for filter processing of the chroma component of a slice (e.g., there is only one neural network model) (e.g., here, a slice is an example of a processing component). The two chroma components can select different modes, which requires the decoder side to perform filter processing twice for the prediction for the two chroma components.

[0087] In some techniques, such as the technique of JVET-Z0113, if the color components have different modes of the NN filter, the video encoder 200 and the video decoder 300 will execute two instances of the NN filter in their respective different modes. For different modes, in JVET-Z0113, the input is luma and chroma samples, but for the NN filter intended to filter the chroma components, the output can be only chroma samples. That is, the NN filter mode for the chroma components uses the luma components as input, but does not have to output luma samples. For example, assume that the first NN filter mode is applied to the first color component (e.g., Cb), and the second NN filter mode is applied to the second color component (e.g., Cr).

[0088] In this example, the video encoder 200 and the video decoder 300 will apply the first instance of the NN filter in the first NN filter mode using the input of the luma, Cb, and Cr components (e.g., the reconstructed samples of the luma, Cb, and Cr blocks). The video encoder 200 and the video decoder 300 will ignore the filtered Cr samples and maintain the filtered Cb samples.

[0089] Next, or in parallel, the video encoder 200 and the video decoder 300 will apply the second instance of the NN filter in the second NN filter mode using the input of the luma, Cb, and Cr components (e.g., the reconstructed samples of the luma, Cb, and Cr blocks). The video encoder 200 and the video decoder 300 will ignore the filtered Cb samples and maintain the filtered Cr samples.

[0090] In this example, the video encoder 200 and the video decoder 300 may maintain the filtered Cb samples from the first instance of the NN filter, maintain the filtered Cr samples from the second instance of the NN filter, and discard the rest. Thus, in this example, to generate a filtered CU including luma and chroma components, the video encoder 200 and the video decoder 300 applied two instances of the NN filter in different modes. Such execution of the two instances of the NN filter in different modes becomes processing-intensive and may delay the reconstruction of the picture.

[0091] A CU is an example of a composite block including component blocks. For example, a CU includes a luma block and a chroma block. In the present disclosure, a CU is used as an example to generally mean a composite block including different color components such as luma and chroma components. The term CU should not be limited to the CU defined in the video coding standard or be part of a video codec unless otherwise specified.

[0092] The present disclosure describes exemplary techniques for suppressing different modes for different color components. That is, the NN filter mode for one color component is the same as the filter mode for the other color component. Thus, the video encoder 200 and the video decoder 300 may apply one instance of the NN filter instead of two instances of the NN filter in different modes.

[0093] Furthermore, the present disclosure describes exemplary techniques that enable selectively determining whether to utilize filtered samples from multiple color components or filtered samples from fewer color components (e.g., one color component). Thus, the use of all filtered components is not essential, and by utilizing a subset of the filtered components, instances that enable better coding efficiency or higher image content can be made possible.

[0094] Also, as will be described in more detail, the present disclosure describes exemplary techniques for signaling different levels of parameters for indicating which NN filter mode should be used. For example, a slice level syntax element may have multiple values. A first value may indicate that NN filter processing is disabled for a color component. A subset of the values among the multiple values may each correspond to a different NN filter mode. A last value may indicate whether NN filter processing is enabled or disabled, and if enabled, the NN filter mode is indicated at a lower level (e.g., at the CTU level for all blocks of a CTU, at the CU level, etc.). In such an example, video encoder 200 may signal an additional syntax element for one or more blocks indicating whether NN filter processing is enabled or disabled, and if enabled, the NN filter mode, and video decoder 300 may receive them.

[0095] In one or more examples described in the present disclosure, due to the similarity between two chroma components, it is possible to design a filter control mechanism that shares the same NN mode selection for chroma components in order to reduce computational complexity without sacrificing rate distortion performance as much as possible. That is, the present disclosure describes exemplary techniques for a filter processing process (singular or plural) in which some or all of the filter processing operations (if any) are shared by multiple color components.

[0096] The following describes a first implementation of the exemplary techniques described in the present disclosure. As one implementation, for a color format having one luma component and two chroma components (e.g., YCbCr color format), at most one NN-based filter is enabled for each processing component to complete the filter processing process for the two chroma components.

[0097] In the first example, the NN filter for the chroma color components can be designed to output the filtered chroma components for both after a single filtering process. The filtering control for the two chroma components can be performed together. For example, the two chroma components have the same method of applying the NN filter, including on / off switching for NN filtering, the same method of creating the input data for the NN model, and the same selection of the values of all input elements, so that after a single filtering process, the filtered samples of both chroma components can be obtained.

[0098] Similar to JVET-Z0113, there is one model to be used for NN filtering for the chroma components, and multiple modes are created by using different QP values in the process of creating the input of the NN model. The selection of the NN filter configuration is signaled and controlled as described below:

[0099] At the slice level (as an example of a processing component, but other examples are also possible), a syntax element with a value range of [0,4] (referred to as the "first-level filter mode syntax element") is signaled for the chroma components. The meanings of the five syntax values are listed below: a. 0: NN filtering for the chroma components for this slice is turned off. b. 1: Turn on NN filtering for the chroma components of this slice, and in the process of creating the input of the NN filter, the first QP in a predefined or signaled set is used. c. 2: Turn on NN filtering for the chroma components of this slice, and in the process of creating the input of the NN filter, the second QP in a predefined or signaled set is used. d. 3: Turn on NN filtering for the chroma components of this slice, and in the process of creating the input of the NN filter, the third QP in a predefined or signaled set is used. e.4: Additional syntax elements are sent to control the NN filter processing separately for each processing component.

[0100] When mode 4 is signaled at the slice level, for each processing component, an additional syntax element (e.g., "second-level filter mode syntax element") having a value range of [0, 3] is signaled for the corresponding chroma component. The meanings of the four syntax values are listed below: a.0: The NN filter processing for the chroma component for this processing component is turned off. b.1: Turn on the NN filter processing for the chroma component of this processing component, and in the process of creating the input to the NN filter, the first QP within a predefined or signaled set is used. c.2: Turn on the NN filter processing for the chroma component of this processing component, and in the process of creating the input to the NN filter, the second QP within a predefined or signaled set is used. d.3: Turn on the NN filter processing for the chroma component of this processing component, and in the process of creating the input to the NN filter, the third QP within a predefined or signaled set is used.

[0101] For example, in the above example, the video encoder 200 may signal a syntax element (e.g., a first-level filter mode syntax element) that defines a filter processing mode for a neural network (NN) model for both the first color component and the second color component, and the video decoder 300 may receive it. For example, in the above example, the value of the first-level syntax element is within the range of [0, 4]. When the value of the first-level syntax element is 1, 2, or 3, the first-level syntax element defines a filter processing mode (e.g., an NN filter mode) for the NN model for both the first and second color components. For example, a value of 1 means that the filter processing mode is based on a first QP within a predefined or signaled set, a value of 2 means that the filter processing mode is based on a second QP within a predefined or signaled set, and a value of 3 means that the filter processing mode is based on a third QP within a predefined or signaled set.

[0102] However, when the first-level filter mode syntax element has a value of 4, the video encoder 200 may signal a syntax element (e.g., a second-level filter mode syntax element) that defines a filter processing mode for a neural network (NN) model for both the first color component and the second color component, and the video decoder 300 may receive it. For example, in the above example, the value of the second-level syntax element is within the range of [0, 3]. When the value of the first-level syntax element is 1, 2, or 3, the second-level syntax element defines a filter processing mode (e.g., an NN filter mode) for the NN model for both the first and second color components. For example, a value of 1 means that the filter processing mode is based on a first QP within a predefined or signaled set, a value of 2 means that the filter processing mode is based on a second QP within a predefined or signaled set, and a value of 3 means that the filter processing mode is based on a third QP within a predefined or signaled set.

[0103] That is, the video encoder 200 may signal a syntax element that defines a filter processing mode for an NN model for both the first color component and the second color component, and the video decoder 300 may receive it. In one example, the syntax element is a first-level filter mode syntax element. In another example, the syntax element is a second-level filter mode syntax element. For example, the video encoder 200 may signal a first syntax element (e.g., a first-level filter mode syntax element) applicable to a plurality of CUs (e.g., CUs of a slice), which indicates that the second syntax element is to be parsed for a subset of the CUs among the plurality of CUs (e.g., the value of the first-level filter mode syntax element is 4, and thus the second-level filter mode syntax element is to be parsed), and the video decoder 300 may receive it. The subset of CUs includes one or more CUs including the reconstructed CUs. That is, the first-level filter mode syntax element is applied to more CUs than the second-level filter mode syntax element.

[0104] The video encoder 200 may signal a second syntax element based on a first syntax element that indicates that the second syntax element is to be parsed for a subset of CUs, and the video decoder 300 may receive it. That is, the video encoder 200 may signal a second-level filter mode syntax element for a subset of CUs on the condition that the value for the first-level filter mode syntax element is 4, and the video decoder 300 may receive it.

[0105] Examples of having a first-level filter mode syntax element and a second-level filter mode syntax element are provided for illustrative purposes only and should not be considered limiting. That is, in some examples, there may be one filter mode syntax element that defines the filter mode for a CU. This one filter mode syntax element may be at the picture level (e.g., for all CUs of a picture), at the slice level (e.g., for all CUs of a slice), at the CTU level (e.g., for all CUs of a CTU), or at the CU level (e.g., for each block).

[0106] The video encoder 200 and the video decoder 300 may apply an instance of the NN model to the first block of the first color component in a defined filter processing mode to generate a first filtered block. For example, the first block may be a portion of the reconstructed sample 800 in FIG. 8, and the first residual value may be the value of the residual sample 802 in FIG. 8. The video encoder 200 and the video decoder 300 may generate a first filtered block (e.g., the filtered sample 804 for the first color component) based on the first block and the first residual value (e.g., by summing the reconstructed sample 800 and the residual sample 802).

[0107] In one or more examples, since a syntax element (e.g., a first-level filter mode syntax element or a second-level filter mode syntax element) defines a filter processing mode for an NN model for both a first color component and a second color component, video encoder 200 and video decoder 300 may apply the same instance of the NN model in the defined filter processing mode to a second block of the second color component to generate a second filtered block residual value. For example, video encoder 200 and video decoder 300 may apply the same instance of the NN model in the defined filter processing mode to a second block of the second color component to generate a second residual value. Video encoder 200 and video decoder 300 may generate a second filtered block based on the second block and the second residual value (e.g., by adding the second block and the second residual value).

[0108] In this disclosure, unless otherwise specified, applying the same instance of the NN model means that one execution of the NN model results in filtered samples for multiple color components. For example, an NN filter may receive luma samples, Cb samples, and Cr samples as inputs. The NN filter mode may be defined by a syntax element (e.g., a value of 1, 2, or 3 for a first-level filter mode syntax element or for a second-level filter mode syntax element). Video encoder 200 and video decoder 300 may apply an instance of the NN filter in the defined filter mode, and the output may be filtered Cb samples and filtered Cr samples.

[0109] To explain that filtered Cb samples and filtered Cr samples are generated by applying the same instance of the NN model, the present disclosure describes the video encoder 200 and the video decoder 300 as applying an instance of the NN model in a defined filtering mode to a first block of a first color component to generate a first filtered block (e.g., based on a first block and a first residual value), and applying the same instance of the NN model in the defined filtering mode to a second block of a second color component to generate a second filtered block (e.g., based on a second block and a second residual value). That is, only one instance of the NN model is executed to generate a first filtered block for a first color component and a second filtered block for a second color component.

[0110] Accordingly, in this example, one syntax element (e.g., a first-level filter mode syntax element or a second-level filter mode syntax element) defines the filtering mode for both chroma components. In some other techniques such as Z-0133, it would be possible for one syntax element to define a filter mode for a first color component and another syntax element to define a filter mode for a second color component. According to the exemplary technique described in the present disclosure, one syntax element defines a filter mode for the NN model for multiple color components.

[0111] In the above example, the video encoder 200 and the video decoder 300 may store sample values for a coding unit (CU) based on the first filtered block and the second filtered block. For example, the video encoder 200 and the video decoder 300 may store the first filtered block and the second filtered block as parts of the CU in their respective buffers, so that when the CU is used for inter prediction, the first filtered block and the second filtered block are used for inter prediction. As another example, the video decoder 300 may store the first filtered block and the second filtered block as parts of the CU in a buffer, so that when a picture including the CU is displayed, the image content is generated based on the first filtered block and the second filtered block.

[0112] As described above, one syntax element defines a filtering mode (e.g., an NN filter mode for an NN filter) for an NN model for both the first color component and the second color component. In some examples, both the first filtered block for the first color component and the second filtered block for the second color component are stored for the CU. However, in some examples, it may be possible not to store both the first filtered block and the second filtered block for the CU.

[0113] In some examples, note that it may be possible to generate a first filtered block for a first color component and a second filtered block for a second color component even if one of them is not stored for a CU. For example, assume that the filtered Cb component is required, but the filtered Cr component is not required for the CU. In this example, the NN model can receive both the luma and chroma components and generate a filtered Cb component and a filtered Cr component. However, the video encoder 200 and the video decoder 300 may discard the filtered Cr component. To store values for the CU, the video encoder 200 and the video decoder 300 may store the filtered Cb component and the original Cr component (e.g., non-filtered, reconstructed samples).

[0114] The following describes an exemplary technique for selecting which filtered components are used to store values for a CU. To facilitate the description, the present disclosure describes filter control for switching an NN filter on or off. Such a disclosure of turning the NN filter on or off may refer to an example of whether a filtered color component is used. Although the filtered color component is still generated, in an example where the filtering of the color component is turned off, the filtered color component can be discarded.

[0115] When the filtering process is enabled, the above example where both the first filtered block for the first color component and the second filtered block for the second color component are stored for the CU is referred to as the first exemplary technique. That is, in the first exemplary technique, one syntax element defines the filtering mode for the NN model for both the first color component and the second color component. Also, in this first exemplary technique, the video encoder 200 and the video decoder 300 store the first filtered block for the first color component and the second filtered block for the second color component as values for the CU. That is, the video encoder 200 and the video decoder 300 may apply the same instance of the NN model in the defined filtering mode to the first block of the first color component and the second block of the second color component to generate the first filtered block and the second filtered block, respectively.

[0116] For example, the video encoder 200 and the video decoder 300 may apply the same instance of the NN model in the defined filtering mode to the first block of the first color component and the second block of the second color component to generate the first residual value and the second result value, respectively. The video encoder 200 and the video decoder 300 may generate the first filtered block based on the first block and the first residual value (e.g., by adding them together), and generate the second filtered block based on the second block and the second residual value (e.g., by adding them together).

[0117] In the second example, similar to the first example, the NN filter for the chroma color components is designed such that both chroma components can be output after a single filtering process. When the NN filtering is on for the two chroma components, in addition to the output of the two chroma components having to be generated by a single filtering process, the filter control enables the on / off switching of the NN filter for the two chroma components to be signaled separately. That is, both filtered chroma blocks can be the output of the NN filter, but at least one of the filtered chroma blocks can be discarded.

[0118] At the slice level (as an example of a processing component), a syntax element having a value range of [0,4] is signaled for the first chroma component (referred to as first_chroma_mode_slice in the following description). For the second chroma component, when first_chroma_mode_slice == 0, another syntax element having a value range of [0,4] is signaled for the second chroma component to indicate the selection of NN filtering for the second chroma component (referred to as second_chroma_mode_slice in the following description). Otherwise (first_chroma_mode_slice!= 0), a flag having a value range of [0,1] is signaled, and when the flag is 0, at this time, second_chroma_mode_slice is set to 0, and otherwise, second_chroma_mode_slice is set to first_chroma_mode_slice.

[0119] The meanings of the five syntax values of first_chroma_mode_slice and second_chroma_mode_slice are the same as those in the first example, except that the applicable range of each mode can be limited to the corresponding chroma component. That is, the values of first_chroma_mode_slice and second_chroma_mode_slice can be the same as the values for the first-level filter mode syntax element or the second-level filter mode syntax element described above. The decoding / derivation logic of first_chroma_mode_slice and second_chroma_mode_slice can be summarized as the following pseudo-code. In the following pseudo-code, the syntax elements coded in the bitstream are written in bold font, and all syntax values are assumed to be 0 before the decoding / derivation logic starts.

[0120]

Table 1

[0121] Similar to the first example where the first-level filter mode syntax element and the second-level filter mode syntax element exist, when first_chroma_mode_slice or second_chroma_mode_slice is equal to 4 (for example, the first-level filter mode syntax element has a value of 4), an additional syntax element (for example, the second-level filter mode syntax element) is signaled separately for the selection of NN filter processing for each processing component. The decoding / derivation logic can be summarized in the following pseudo-code. The syntax elements coded in the bitstream are written in bold font, and all syntax values are assumed to be 0 before the decoding / derivation logic starts.

[0122]

Table 2

[0123] In this example, the interpretation of the mode values at the slice level and the processing component level is the same as that described above for the first-level filter mode syntax elements and the second-level filter mode syntax elements. At the slice level, 0 means that the NN filter processing is off, 1 to 3 correspond to three QP selections that can be used in the process of creating input information for the NN model, and 4 means that additional information for indicating the NN filter processing selection for each processing component is signaled at the processing component level (e.g., at the CTB level for all CUs of the CTB, at the CU level, etc.). Similarly, at the processing component level, 0 means that the NN filter processing is off, and 1 to 3 correspond to three QP selections that can be used in the process of creating input information for the NN model.

[0124] Therefore, in one or more examples, the syntax element signaled by the video encoder 200 and received by the video decoder 300 may have a value indicating that the NN-based filter processing is enabled for the first color component (e.g., first_chroma_mode_slice is true). In this example, the video encoder 200 may signal a flag (e.g., second_chroma_flag_slice) indicating that the NN-based filter processing is enabled for the second color component, and the video decoder 300 may receive it (e.g., second_chroma_flag_slice is true). In this example, since the video encoder 200 and the video decoder 300 apply the same instance of the NN model in the defined filter processing mode to the second block, a second filtered block can be generated based on the second block of the second color component. To store the value for the CU, the video encoder 200 and the video decoder 300 may store the value for the CU based on the first filtered block and the second filtered block.

[0125] In some examples, the filtered second color component may not be used as a value for the CU. For example, video encoder 200 may signal a flag (e.g., second_chroma_flag_slice) that indicates that NN-based filtering is disabled for the second color component, and video decoder 300 may receive it (e.g., (e.g., second_chroma_flag_slice is false). In this example, video encoder 200 and video decoder 300 may store a value for the CU based on a first filtered block of the first color component (e.g., a block filtered using NN), and a second block of the second color component (e.g., a block not filtered using NN).

[0126] In some examples, the first color component may not be filtered (e.g., first_chroma_mode_slice == 0). In this example, video encoder 200 may signal another syntax element (e.g., second_chroma_mode_slice) that indicates a filter mode for the second color component, and video decoder 300 may receive it. In this example, video encoder 200 and video decoder 300 may store a value for the CU based on a first block of the first color component (e.g., a block not filtered for use with NN), and a second filtered block of the second color component (e.g., a block filtered using NN).

[0127] Generally, for the second example, the first_chroma_mode_slice syntax element can be considered as a syntax element that defines a filtering mode for a neural network (NN) model for both the first color component and the second color component. For example, if the first_chroma_mode_slice syntax element is not 0, then at this time, the NN filtering is on for the first color component, and the second_chroma_flag_slice indicates that the filtering is on for the second color component, and at this time, the first_chroma_mode_slice defines what the filtering mode is for both the first color component and the second color component. Even in the example where the second_chroma_flag_slice is false, since no additional syntax element is needed to indicate the filtering mode, the first_chroma_mode_slice can be considered as defining the filtering mode for both the first color component and the second color component.

[0128] As a third example, the third example is the same as the second example except that an alternative way of decoding / deriving the selection of NN filtering is used. For each color component, a flag indicating an on / off selection is coded, and if at least one of the flags indicates that the NN filtering is on, an additional syntax element is signaled. The decoding at the slice level (as an example of a processing component, but others are possible) is summarized below. The syntax elements coded in the bitstream are written in bold font, and all syntax values are assumed to be 0 before the decoding / deriving logic starts.

[0129]

Table 3

[0130] In this example, first_chroma_flag_slice and second_chroma_flag_slice have values in the range [0,1], and chroma_mode_slice has values in the range [0,3]. Thus, the value ranges of first_chroma_mode_slice and second_chroma_mode_slice are kept the same as in the second example. The range [0,4], and the interpretation of the values are also aligned. That is, chroma_mode_slice can be considered an exemplary syntax element that defines the filtering mode for the neural network (NN) model for both the first color component and the second color component. The values for chroma_mode_slice can be the same as the first-level filter mode syntax elements described above.

[0131] For example, video encoder 200 may signal a flag indicating that NN-based filtering is enabled for the first color component, and video decoder 300 may receive it. In this example, video encoder 200 may signal first_chroma_flag_slice indicating that NN-based filtering is enabled for the first color component, and video decoder 300 may receive it. Similarly, video encoder 200 may signal a flag indicating that NN-based filtering is enabled for the second color component, and video decoder 300 may receive it. In this example, video encoder 200 may signal second_chroma_flag_slice indicating that NN-based filtering is enabled for the second color component, and video decoder 300 may receive it.

[0132] The video encoder 200 may signal a syntax element, and the video decoder 300 may receive it. Based on a flag indicating that NN-based filtering is enabled for the first color component, a filtering mode for the NN model for both the first and second color components is defined. For example, the video encoder 200 may signal the chroma_mode_slice syntax element based on first_chroma_flag_slice being true, and the video decoder 300 may receive it. Similarly, the video encoder 200 may signal a syntax element, and the video decoder 300 may receive it. Based on a flag indicating that NN-based filtering is enabled for the second color component, a filtering mode for the NN model for both the first and second color components is defined. For example, the video encoder 200 may signal the chroma_mode_slice syntax element based on second_chroma_flag_slice being true, and the video decoder 300 may receive it.

[0133] If NN filtering is enabled (e.g., on) for one or both of the first and second color components, the video decoder 300 may determine the filtering mode based on the value of chroma_mode_slice. As shown above, the formula for determining the filtering mode for both color components is the same (e.g., the filter mode is chroma_mode_slice + 1 for both). Thus, chroma_mode_slice defines the filtering mode for the NN model for both the first and second color components.

[0134] Equivalents of this example can be created by changing the way the syntax element is derived and interpreted. An example of such an equivalent design is described in a later example.

[0135] In this example, when first_chroma_mode_slice or second_chroma_mode_slice is equal to 4, additional syntax elements (such as, for example, the second-level filter mode syntax element) are signaled separately for each processing component for the selection of the NN filtering. The decoding / derivation logic can be summarized as pseudocode. The syntax elements coded in the bitstream are written in bold font, and all syntax values are assumed to be 0 before the decoding / derivation logic starts.

[0136]

Table 4

[0137] In this example, first_chroma_flag_processing_unit and second_chroma_flag_processing_unit have range values of [0,1], chroma_mode_processing_unit has range values of [0,2], and thus the value ranges of first_chroma_mode_processing_unit and second_chroma_mode_processing_unit are kept the same as those in the second example. The range [0,3], and the interpretation of the values are also aligned. By changing the way the syntax elements are derived and interpreted, an equivalent of this example can be created. An example of such an equivalent design is described in a later example.

[0138] As a fourth example, the fourth example is equivalent to the third example, with different derivation / interpretation of the syntax values but the same actual NN control. The slice-level decoding can be summarized as follows. The syntax elements coded in the bitstream are written in bold font, and all syntax values are assumed to be initialized with 0 before the decoding / derivation logic starts.

[0139]

Table 5

[0140] In this case, first_chroma_flag_slice and second_chroma_flag_slice have values in the range [0, 1], and chroma_mode_slice has values in the range [0, 3]. As a result, the value ranges of first_chroma_mode_slice and second_chroma_mode_slice become [0, 3]. According to the following interpretation, this example is equivalent to the third example: a. When first_chroma_flag_slice == 0, at this time, the NN filtering process is off for the first chroma component. b. When first_chroma_flag_slice == 1, the selection of the NN filtering process is further signaled / derived as follows: i. When first_chroma_mode_slice == 0, in the process of creating the input to the NN filter, the first QP within a pre-defined or signaled set is used. ii. When first_chroma_mode_slice == 1, in the process of creating the input to the NN filter, the second QP within a pre-defined or signaled set is used. iii. When first_chroma_mode_slice == 2, in the process of creating the input to the NN filter, the third QP within a pre-defined or signaled set is used. iv. When first_chroma_mode_slice == 4, additional syntax elements are sent to control the NN filtering process separately for each processing component.

[0141] Similar to the slice level, when the selection of NN filter processing is signaled at the processing component level, when the NN filter processing is on (as shown in the following pseudo-code), the first_chroma_mode_processing_unit and the second_chroma_mode_processing_unit are given a chroma_mode_processing_unit. Syntax elements coded in the bitstream are written in bold font, and all syntax values are assumed to be initialized to 0 before the decoding / derivation logic starts.

[0142]

Table 6

[0143] The interpretation can be the same as at the slice level, except that there may not be a mode value of 4 at the processing component level: a. When first_chroma_flag_processing_unit == 0, at this time, the NN filter processing is off for the first chroma component. b. When first_chroma_flag_processing_unit == 1, the selection of NN filter processing is further signaled / derived as follows: i. When first_chroma_mode_processing_unit == 0, in the process of creating the input of the NN filter, the first QP in the predefined or signaled set is used. ii. When first_chroma_mode_processing_unit == 1, in the process of creating the input of the NN filter, the second QP in the predefined or signaled set is used. iii. When first_chroma_mode_processing_unit == 2, in the process of creating the input of the NN filter, the third QP in the predefined or signaled set is used.

[0144] Examples 1 to 4 can be extended to any multi-mode NN filter processing design, and the exemplary techniques described in this disclosure can still be applied. For example, in one or more examples, a syntax element that defines a filter processing mode for an NN model for both a first color component and a second color component defines which parameter to select from a plurality of parameters, and the defined parameter defines the filter processing mode. That is, when a syntax element defines a filter processing mode, it may mean that the syntax element defines which parameter to use and the parameter defines the filter processing mode.

[0145] For example, the syntax element can have a value of 1, 2, or 3, where a value of 1 indicates that a first parameter from a predefined set or a signaled set defines the filter processing mode. A value of 2 indicates that a second parameter from a predefined set or a signaled set defines the filter processing mode, and a value of 3 indicates that a third parameter from a predefined set or a signaled set defines the filter processing mode. In some examples, the parameter is a quantization parameter.

[0146] Some of the additional parameters or sets of parameters for which the parameter is the selected example are: a. Instead of selecting a QP value from a predefined / signaled set having a size of 3. The method can be extended when using a candidate set of N values. b. Instead of using different QP values for different modes, any value used in the process of creating an input to the NN filter processing model can be utilized to design a multi-mode NN filter processing method. c. Instead of using different QP values for different modes, a multi-mode design can use different NN models for different modes. d. The different ways of designing the multi-mode filter processing method described above can be combined to create a more sophisticated multi-mode design. For example, the techniques can be combined to create M models, and for each model, there are N QP values that can be used when creating the input to the model. At this time, a total of M * N modes are created. By using two models and two QPs to be selected for each model, a four-model design can be created.

[0147] As a second implementation form, the exemplary techniques of the first implementation form can be extended to include more color components. For example, all examples of the first implementation form can be designed in such a way that the NN model can output all three color components (RGB, YUV, etc.), and in order to achieve the goal that at most one NN filter processing operation is executed for all three color components, the filter control described in the present disclosure can be used.

[0148] FIG. 2 is a block diagram showing an exemplary video encoder 200 that can implement the techniques of the present disclosure. FIG. 2 is provided for explanation and should not be considered as limiting the techniques widely exemplified and described in the present disclosure. For explanation purposes, the present disclosure will describe the video encoder 200 according to the techniques of VVC (ITU-T H.266 under development) and HEVC (ITU-T H.265). However, the techniques of the present disclosure can be implemented by video encoding devices configured for other video coding standards, as well as video coding formats such as successors to the AV1 and AV1 video coding formats.

[0149] In the example of FIG. 2, the video encoder 200 includes a video data memory 230, a mode selection unit 202, a residual generation unit 204, a conversion processing unit 206, a quantization unit 208, an inverse quantization unit 210, an inverse conversion processing unit 212, a reconstruction unit 214, a filter unit 216, a decoded picture buffer (DPB) 218, and an entropy encoding unit 220. Any or all of the video data memory 230, the mode selection unit 202, the residual generation unit 204, the conversion processing unit 206, the quantization unit 208, the inverse quantization unit 210, the inverse conversion processing unit 212, the reconstruction unit 214, the filter unit 216, the DPB 218, and the entropy encoding unit 220 may be implemented in one or more processors or in processing circuitry. For example, the units of the video encoder 200 may be implemented as one or more circuits or logic elements as part of a hardware circuit, or as part of a processor, an ASIC, or an FPGA. Additionally, the video encoder 200 may include additional or alternative processors or processing circuitry to perform these and other functions.

[0150] The video data memory 230 may store video data to be encoded by components of the video encoder 200. The video encoder 200 may receive video data stored within the video data memory 230, for example, from the video source 104 (FIG. 1). The DPB 218 may function as a reference picture memory that stores reference video data used in the prediction of subsequent video data by the video encoder 200. The video data memory 230 and the DPB 218 may be formed by any of various memory devices such as a dynamic random access memory (DRAM) including a synchronous dynamic random access memory (SDRAM), a magnetoresistive RAM (MRAM), a resistive RAM (RRAM), or other types of memory devices. The video data memory 230 and the DPB 218 may be provided by the same memory device or by separate memory devices. In various examples, the video data memory 230 may be on-chip with other components of the video encoder 200 as shown, or may be off-chip with respect to those components.

[0151] In the present disclosure, references to the video data memory 230 should not be construed as being limited to the memory internal to the video encoder 200 or the memory external to the video encoder 200, unless otherwise specifically described as such. Rather, references to the video data memory 230 should be understood as a reference memory that stores video data received by the video encoder 200 for encoding (e.g., video data for the current block to be encoded). The memory 106 of FIG. 1 may also provide temporary storage of outputs from various units of the video encoder 200.

[0152] The various units in FIG. 2 are shown to assist in understanding the operations performed by video encoder 200. The units may be implemented as fixed function circuits, programmable circuits, or combinations thereof. A fixed function circuit refers to a circuit that provides a specific function, and the operations that can be performed are pre-set. A programmable circuit refers to a circuit that can be programmed to perform various tasks and provides a flexible function in the operations that can be performed. For example, a programmable circuit may execute software or firmware that operates the programmable circuit in a manner defined by software or firmware instructions. A fixed function circuit may execute software instructions (e.g., receive or output parameters), but the type of operations performed by the fixed function circuit is generally invariant. In some examples, one or more of the units may be different circuit blocks (fixed function or programmable), and in some examples, one or more of the units may be integrated circuits.

[0153] Video encoder 200 may include an arithmetic logic units (ALUs), elementary function units (EFUs), digital circuits, analog circuits, and / or a programmable core formed from programmable circuits. In an example where the operations of video encoder 200 are performed using software executed by a programmable circuit, memory 106 (FIG. 1) may store software instructions (e.g., object code) that video encoder 200 receives and executes, or another memory (not shown) within video encoder 200 may store such instructions.

[0154] The video data memory 230 is configured to store the received video data. The video encoder 200 may extract a picture of the video data from the video data memory 230 and provide the video data to the residual generation unit 204 and the mode selection unit 202. The video data in the video data memory 230 may be raw video data to be encoded.

[0155] The mode selection unit 202 includes a motion estimation unit 222, a motion compensation unit 224, and an intra prediction unit 226. The mode selection unit 202 may include additional functional units that perform video prediction according to other prediction modes. By way of example, the mode selection unit 202 may include a palette unit, an intra block copy unit (which may be part of the motion estimation unit 222 and / or the motion compensation unit 224), an affine unit, a linear model (LM) unit, and the like.

[0156] The mode selection unit 202 generally coordinates a plurality of encoding paths to test a combination of encoding parameters and the resulting rate distortion value for such a combination. The encoding parameters may include the partitioning of CTUs into CUs, the prediction mode for the CUs, the type of transform for the residual data of the CUs, the quantization parameter for the residual data of the CUs, and the like. The mode selection unit 202 may ultimately select a combination of encoding parameters that has a better rate distortion value than other tested combinations.

[0157] The video encoder 200 may divide the picture taken from the video data memory 230 into a series of CTUs, and encapsulate one or more CTUs within a slice. The mode selection unit 202 may divide the CTUs of the picture according to a tree structure, such as the MTT structure, QTBT structure, superblock structure, or quad-tree structure described above. As described above, since the video encoder 200 divides the CTUs according to a tree structure, one or more CUs may be formed. Such CUs are sometimes generally referred to as "video blocks" or "blocks".

[0158] Generally, the mode selection unit 202 also controls its components (e.g., the motion estimation unit 222, the motion compensation unit 224, and the intra prediction unit 226) to generate a prediction block for the current block (e.g., the current CU, or in HEVC, the overlapping part of the PU and TU). In the case of inter prediction of the current block, the motion estimation unit 222 may perform a motion search to identify one or more exactly matching reference blocks among one or more reference pictures (e.g., one or more previously coded pictures stored in the DPB 218). Specifically, the motion estimation unit 222 may calculate a value representing how similar a potential reference block is to the current block according to, for example, the sum of absolute differences (SAD), the sum of squared differences (SSD), the mean absolute difference (MAD), the mean squared difference (MSD), etc. The motion estimation unit 222 may generally perform these calculations using the sample-by-sample differences between the current block and the reference block under consideration. The motion estimation unit 222 may identify the reference block having the lowest value obtained from these calculations, which indicates the reference block that most closely matches the current block.

[0159] The motion estimation unit 222 may form one or more motion vectors (MVs) that define the position of a reference block in a reference picture relative to the position of a current block in the current picture. The motion estimation unit 222 may then provide the motion vectors to the motion compensation unit 224. For example, in the case of uni-directional inter prediction, the motion estimation unit 222 may provide a single motion vector, while in the case of bi-directional inter prediction, the motion estimation unit 222 may provide two motion vectors. The motion compensation unit 224 may then generate a prediction block using the motion vectors. For example, the motion compensation unit 224 may retrieve data of the reference block using the motion vectors. As another example, if the motion vectors have fractional sample accuracy, the motion compensation unit 224 may interpolate values for the prediction block according to one or more interpolation filters. Moreover, in the case of bi-directional inter prediction, the motion compensation unit 224 may retrieve data for two reference blocks identified by the respective motion vectors and may combine the retrieved data, e.g., through per-sample averaging or weighted averaging.

[0160] When operating according to the AV1 video coding format, the motion estimation unit 222 and the motion compensation unit 224 may be configured to encode coding blocks of video data (e.g., both luma coding blocks and chroma coding blocks) using translational motion compensation, affine motion compensation, overlapped block motion compensation (OBMC), and / or combined inter-intra prediction.

[0161] As another example, in the case of intra prediction or intra prediction coding, the intra prediction unit 226 may generate a prediction block from samples adjacent to the current block. For example, in the case of the directional mode, the intra prediction unit 226 may generally mathematically combine the values of the adjacent samples to generate a prediction block and populate these calculated values in a direction defined across the current block. As another example, in the case of the DC mode, the intra prediction unit 226 may calculate the average of the adjacent samples for the current block and generate a prediction block that includes this obtained average for each sample of the prediction block.

[0162] When operating according to the AV1 video coding format, the intra prediction unit 226 may be configured to encode coding blocks of video data (e.g., both luma coding blocks and chroma coding blocks) using directional intra prediction, non-directional intra prediction, recursive filter intra prediction, chroma-from-luma (CFL) prediction, intra block copy (IBC), and / or color palette mode. The mode selection unit 202 may include additional functional units that perform video prediction according to other prediction modes.

[0163] The mode selection unit 202 provides the prediction block to the residual generation unit 204. The residual generation unit 204 receives the raw, unencoded version of the current block from the video data memory 230 and the prediction block from the mode selection unit 202. The residual generation unit 204 calculates the sample-by-sample difference between the current block and the prediction block. The resulting sample-by-sample differences define the residual block for the current block. In some examples, the residual generation unit 204 may also determine the differences between the sample values in the residual block in order to generate the residual block using residual differential pulse code modulation (RDPCM). In some examples, the residual generation unit 204 may be formed using one or more subtractor circuits that perform binary subtraction.

[0164] In an example where the mode selection unit 202 divides a CU into PUs, each PU may be associated with a luma prediction unit and a corresponding chroma prediction unit. The video encoder 200 and the video decoder 300 can support PUs having various sizes. As described above, the size of a CU may refer to the size of the luma coding block of the CU, and the size of a PU may refer to the size of the luma prediction unit of the PU. Assuming that the size of a particular CU is 2N×2N, the video encoder 200 may support a PU size of 2N×2N or N×N for intra prediction and a PU size of 2N×2N, 2N×N, N×2N, N×N, or similar, symmetric PU sizes for inter prediction. The video encoder 200 and the video decoder 300 may also support an asymmetric partition of PU sizes of 2N×nU, 2N×nD, nL×2N, and nR×2N for inter prediction.

[0165] In an example where the mode selection unit 202 does not further divide a CU into PUs, each CU may be associated with a luma coding block and a corresponding chroma coding block. As described above, the size of a CU may refer to the size of the luma coding block of the CU. The video encoder 200 and the video decoder 300 may support CU sizes of 2N×2N, 2N×N, or N×2N.

[0166] As some examples, in the case of other video coding techniques such as intra block copy mode coding, affine mode coding, and linear model (LM) mode coding, the mode selection unit 202 generates a prediction block for the current block being encoded via respective units associated with the coding technique. In some examples such as palette mode coding, the mode selection unit 202 may not generate a prediction block. Instead, the mode selection unit 202 may generate syntax elements indicating a way to reconstruct the block based on the selected palette. In such a mode, the mode selection unit 202 may provide these syntax elements to the entropy coding unit 220 to be encoded.

[0167] As described above, the residual generation unit 204 receives video data for the current block and the corresponding prediction block. The residual generation unit 204 then generates a residual block for the current block. To generate the residual block, the residual generation unit 204 calculates the sample - by - sample difference between the prediction block and the current block.

[0168] The transformation processing unit 206 applies one or more transformations to the residual block to generate a block of transformation coefficients (referred to herein as the "transformation coefficient block"). The transformation processing unit 206 may apply various transformations to the residual block to form the transformation coefficient block. For example, the transformation processing unit 206 may apply a discrete cosine transform (DCT), a directional transform, a Karhunen - Loève transform (KLT), or a conceptually similar transform to the residual block. In some examples, the transformation processing unit 206 may perform multiple transformations, such as a secondary transformation like a primary transformation and a rotation transformation, on the residual block. In some examples, the transformation processing unit 206 does not apply a transformation to the residual block.

[0169] When the transformation processing unit 206 operates according to AV1, it may apply one or more transformations to the residual block to generate a block of transformation coefficients (referred to herein as the "transformation coefficient block"). The transformation processing unit 206 may apply various transformations to the residual block to form the transformation coefficient block. For example, the transformation processing unit 206 may apply a horizontal / vertical transformation combination that may include a discrete cosine transform (DCT), an asymmetric discrete sine transform (ADST), an inverted ADST (e.g., ADST in reverse order), and an identity transform (IDTX). When using the identity transform, the transformation is skipped in one of the vertical or horizontal directions. In some examples, the transformation processing may be skipped.

[0170] The quantization unit 208 may quantize the transform coefficients in the transform coefficient block to generate a quantized transform coefficient block. The quantization unit 208 may quantize the transform coefficients of the transform coefficient block according to the quantization parameter (QP) value associated with the current block. The video encoder 200 may adjust the degree of quantization applied to the transform coefficient block associated with the current block by adjusting the QP value associated with the CU (e.g., via the mode selection unit 202). Quantization may introduce information loss, and thus, the quantized transform coefficients may be less accurate than the original transform coefficients generated by the transform processing unit 206.

[0171] The inverse quantization unit 210 and the inverse transform processing unit 212 may apply inverse quantization and inverse transform, respectively, to the quantized transform coefficient block to reconstruct a residual block from the transform coefficient block. The reconstruction unit 214 may generate a reconstructed block corresponding to the current block (although there may be some distortion) based on the reconstructed residual block and the prediction block generated by the mode selection unit 202. For example, the reconstruction unit 214 may add samples of the reconstructed residual block to the corresponding samples from the prediction block generated by the mode selection unit 202 to generate the reconstructed block.

[0172] The filter unit 216 may perform one or more filtering operations on the reconstructed block. For example, the filter unit 216 may perform a deblocking operation to reduce blocking artifacts along the edges of the CU. The operation of the filter unit 216 may be skipped in some examples.

[0173] According to one or more examples described in the present disclosure, the filter unit 216 may be configured to perform a neural network-based filtering technique. For example, the filter unit 216 may perform a neural network-based filtering technique in addition to, or instead of, other filtering techniques such as ALF and SAO. For example, the filter unit 216 may be part of a reconstruction loop that includes a decoding process performed by the video encoder 200. The filter unit 216 may perform the neural network-based filtering described in the present disclosure as part of the in-loop filtering within the reconstruction loop of the decoding process of the video encoder 200.

[0174] When operating according to AV1, the filter unit 216 may perform one or more filtering operations on the reconstructed block. For example, the filter unit 216 may perform a deblocking operation to reduce blockiness artifacts along the edges of the CU. In other examples, the filter unit 216 may apply a constrained directional enhancement filter (CDEF), which may be applied after deblocking and may include the application of an inseparable, non-linear, low-pass directional filter based on the estimated edge direction. The filter unit 216 may also include a loop restoration filter, which may be applied after the CDEF and may include a separable symmetric normalized Wiener filter or a dual self-induced filter.

[0175] Video encoder 200 stores the reconstructed blocks in DPB 218. For example, in an example where the operation of filter unit 216 is not performed, reconstruction unit 214 may store the reconstructed blocks in DPB 218. In an example where the operation of filter unit 216 is performed, filter unit 216 may store the filtered reconstructed blocks in DPB 218. Motion estimation unit 222 and motion compensation unit 224 may retrieve a reference picture formed from the reconstructed (and possibly filtered) blocks from DPB 218 to inter-predict blocks of a picture to be encoded later. In addition, intra prediction unit 226 may use the reconstructed blocks of the current picture in DPB 218 to intra-predict other blocks in the current picture.

[0176] Generally, the entropy encoding unit 220 may entropy-encode syntax elements received from other functional components of the video encoder 200. For example, the entropy encoding unit 220 may entropy-encode the quantized transform coefficient blocks from the quantization unit 208. As another example, the entropy encoding unit 220 may entropy-encode prediction syntax elements (e.g., motion information for inter prediction or intra mode information for intra prediction) from the mode selection unit 202. The entropy encoding unit 220 may perform one or more entropy encoding operations on syntax elements, which are another example of video data, to generate entropy-encoded data. For example, the entropy encoding unit 220 may perform context-adaptive variable length coding (CAVLC) operations, CABAC operations, variable-to-variable (V2V) coding operations, syntax-based context-adaptive binary arithmetic coding (SBAC) operations, probability interval partitioning entropy (PIPE) coding operations, exponential Golomb coding operations, or another type of entropy encoding operation on the data. In some examples, the entropy encoding unit 220 may operate in a bypass mode where the syntax elements are not entropy-encoded.

[0177] The video encoder 200 may output a bitstream including entropy-encoded syntax elements required to reconstruct blocks of a slice or a picture. Specifically, the entropy encoding unit 220 may output the bitstream.

[0178] The entropy encoding unit 220 may be configured as a symbol-to-symbol adaptive multi-symbol arithmetic coder according to AV1. The syntax elements in AV1 include an alphabet of N elements, and the context (e.g., probability model) includes a set of N probabilities. The entropy encoding unit 220 may store the probabilities as n-bit (e.g., 15-bit) cumulative distribution functions (CDFs). The entropy encoding unit 22 may perform recursive scaling using an update factor based on the alphabet size to update the context.

[0179] The operations described above are described with respect to blocks. Such descriptions should be understood as operations on luma coding blocks and / or chroma coding blocks. As described above, in some examples, the luma coding block and the chroma coding block are the luma component and the chroma component of the CU. In some examples, the luma coding block and the chroma coding block are the luma component and the chroma component of the PU.

[0180] In some examples, the operations performed on the luma coding block need not be repeated for the chroma coding block. As an example, the operation of identifying the motion vector (MV) and the reference picture for the luma coding block need not be repeated to identify the MV and the reference picture for the chroma block. Rather, the MV of the luma coding block may be scaled to determine the MV of the chroma block, and the reference picture may be the same. As another example, the intra prediction process may be the same for the luma coding block and the chroma coding block.

[0181] Video encoder 200 represents an example of a device configured to encode video data, including a memory configured to store video data and one or more processing units implemented within a circuit, the one or more processing units being configured to filter two or more color components of the video data using a neural network (NN)-based filter from a single NN model and to output two or more color components after a single filtering process using the NN-based filter. In some examples, filtering includes how to apply the NN-based filter, whether to turn the NN-based filter on or off, how to create input data for a single NN model, and determining the values of all input elements in the same manner for two or more color components.

[0182] In some examples, video encoder 200 may be configured to signal information indicating whether the NN-based filter is on or off for each of two or more color components. In such examples, filtering includes filtering based on whether the NN-based filter is on or off for each of two or more color components.

[0183] For example, video encoder 200 may signal a syntax element that defines a filtering mode for a neural network (NN) model for both a first color component and a second color component. Examples of syntax elements include the first-level filter mode syntax element, the second-level filter mode syntax element, first_chroma_mode_slice (e.g., where first_chroma_mode_slice is not zero and second_chroma_flag_slice is true), and chroma_mode_slice (e.g., where first_chroma_flag_slice and / or second_chroma_flag_slice is true), as described above.

[0184] The filter unit 216 may be configured to apply an instance of the NN model to the first block of the first color component in a defined filter processing mode to generate a first filtered block. For example, the filter unit 216 may use the reconstructed samples of the first color component (e.g., the first chroma block) to generate the first residual samples, and add the first residual samples to the reconstructed samples of the first color component to generate the first filtered block.

[0185] The filter unit 216 may be configured to apply the same instance of the NN model to the second block of the second color component in a defined filter processing mode to generate a second filtered block. For example, the filter unit 216 may use the reconstructed samples of the second color component (e.g., the second chroma block) to generate the second residual samples, and add the second residual samples to the reconstructed samples of the second color component to generate the second filtered block.

[0186] In one or more examples, the filter unit 216 may store the first filtered block and the second filtered block. For example, the sample values stored by the filter unit 216 in the DPB 218 for the CU may be the first filtered block and the second filtered block. Thus, when the CU is used for inter prediction, the filtered blocks will be used for inter prediction.

[0187] In some cases, the filter unit 216 may store the second filtered block only when the filtering process is enabled for the second color component. For example, using the same (e.g., single) instance of the NN model, the filter unit 216 may generate a first filtered block and a second filtered block. However, the second filtered block may sometimes not be stored (e.g., discarded).

[0188] As an example, the video encoder 200 may signal the above-mentioned second_chroma_flag_slice flag as true, indicating that the NN-based filtering process is enabled for the second color component. In this example, the filter unit 216 may store a value for the CU based on the first filtered block (for the first color component) and the second filtered block (for the second color component). In another example, the video encoder 200 may signal the above-mentioned second_chroma_flag_slice flag as false, indicating that the NN-based filtering process is disabled for the second color component. In this example, the filter unit 216 may store a value for the CU based on the first filtered block (for the first color component) and the second block (for the second color component) without performing filtering. Also in this case, the second filtered block can be generated by applying an instance of the NN model. However, when the second_chroma_flag_slice flag is false, this second filtered block may be discarded.

[0189] FIG. 3 is a block diagram showing an exemplary video decoder 300 that can implement the techniques of the present disclosure. FIG. 3 is provided for illustration and is not intended to limit the techniques widely exemplified and described in the present disclosure. For illustration purposes, the present disclosure will describe the video decoder 300 according to the techniques of VVC (ITU-T H.266 under development) and HEVC (ITU-T H.265). However, the techniques of the present disclosure can be implemented by video coding devices configured for other video coding standards.

[0190] In the example of FIG. 3, the video decoder 300 includes a coded picture buffer (CPB) memory 320, an entropy decoding unit 302, a prediction processing unit 304, an inverse quantization unit 306, an inverse transform processing unit 308, a reconstruction unit 310, a filter unit 312, and a decoded picture buffer (DPB) 314. Any or all of the CPB memory 320, the entropy decoding unit 302, the prediction processing unit 304, the inverse quantization unit 306, the inverse transform processing unit 308, the reconstruction unit 310, the filter unit 312, and the DPB 314 can be implemented in one or more processors or in processing circuitry. For example, the units of the video decoder 300 can be implemented as one or more circuits or logic elements as part of a hardware circuit, or as part of a processor, ASIC, or FPGA. Additionally, the video decoder 300 may include additional or alternative processors or processing circuitry to perform these and other functions.

[0191] The prediction processing unit 304 includes a motion compensation unit 316 and an intra prediction unit 318. The prediction processing unit 304 may include additional units that perform predictions according to other prediction modes. By way of example, the prediction processing unit 304 may include a palette unit, an intra block copy unit (which may form part of the motion compensation unit 316), an affine unit, a linear model (LM) unit, and the like. In other examples, the video decoder 300 may include more, fewer, or different functional components.

[0192] When operating according to AV1, the compensation unit 316 may be configured to decode coding blocks of video data (e.g., both luma coding blocks and chroma coding blocks) using translational motion compensation, affine motion compensation, OBMC, and / or combined inter-intra prediction, as described above. The intra prediction unit 318 may be configured to decode coding blocks of video data (e.g., both luma coding blocks and chroma coding blocks) using directional intra prediction, non-directional intra prediction, recursive filter intra prediction, CFL, intra block copy (IBC), and / or color palette mode, as described above.

[0193] The CPB memory 320 may store video data such as an encoded video bitstream to be decoded by components of the video decoder 300. The video data stored in the CPB memory 320 may be obtained, for example, from the computer-readable recording medium 110 (FIG. 1). The CPB memory 320 may include a CPB that stores encoded video data (e.g., syntax elements) from the encoded video bitstream. Also, the CPB memory 320 may store video data other than syntax elements of the coded picture, such as temporary data representing outputs from various units of the video decoder 300. The DPB 314 generally stores decoded pictures that the video decoder 300 may output and / or use as reference video data when decoding subsequent data or pictures of the encoded video bitstream. The CPB memory 320 and the DPB 314 may be formed by any of various memory devices such as DRAM, including SDRAM, MRAM, RRAM, or other types of memory devices. The CPB memory 320 and the DPB 314 may be provided by the same memory device or separate memory devices. In various examples, the CPB memory 320 may be on-chip with other components of the video decoder 300 or off-chip with respect to those components.

[0194] Additionally or alternatively, in some examples, the video decoder 300 may retrieve the encoded video data from the memory 120 (FIG. 1). That is, the memory 120 may store data as discussed above for the CPB memory 320. Similarly, the memory 120 may store instructions to be executed by the video decoder 300 when some or all of the functionality of the video decoder 300 is implemented in software to be executed by the processing circuitry of the video decoder 300.

[0195] The various units shown in FIG. 3 are shown to assist in understanding the operations performed by the video decoder 300. The units may be implemented as fixed function circuitry, programmable circuitry, or a combination thereof. Similar to FIG. 2, fixed function circuitry refers to circuitry that provides a specific function and the operations that may be performed are preset. Programmable circuitry refers to circuitry that may be programmed to perform various tasks and provides a flexible function in the operations that may be performed. For example, the programmable circuitry may execute software or firmware that operates the programmable circuitry in a manner defined by instructions of the software or firmware. Fixed function circuitry may execute software instructions (e.g., receive or output parameters), but the type of operations performed by the fixed function circuitry is generally invariant. In some examples, one or more of the units may be different circuit blocks (fixed function or programmable), and in some examples, one or more of the units may be an integrated circuit.

[0196] The video decoder 300 may include a programmable core formed from an ALU, EFU, digital circuitry, analog circuitry, and / or programmable circuitry. In an example where the operations of the video decoder 300 are implemented by software executed on the programmable circuitry, on-chip memory or off-chip memory may store instructions (e.g., object code) of the software that the video decoder 300 receives and executes.

[0197] The entropy decoding unit 302 may receive the encoded video data from the CPB and entropy decode the video data in order to reproduce the syntax elements. The prediction processing unit 304, the inverse quantization unit 306, the inverse transform processing unit 308, the reconstruction unit 310, and the filter unit 312 may generate the decoded video data based on the syntax elements extracted from the bitstream.

[0198] Generally, the video decoder 300 reconstructs the picture block by block. The video decoder 300 may perform the reconstruction operation individually for each block (the currently reconstructed block, i.e., the decoded block, may be referred to as the "current block").

[0199] The entropy decoding unit 302 may entropy decode the syntax elements that define the quantized transform coefficients of the quantized transform coefficient block, as well as the transform information such as the quantization parameter (QP) and / or the transform mode indication(s). The inverse quantization unit 306 may use the QP associated with the quantized transform coefficient block to determine the degree of quantization and, similarly, the degree of inverse quantization to be applied by the inverse quantization unit 306. The inverse quantization unit 306 may perform, for example, a left shift operation bit by bit to inverse quantize the quantized transform coefficients. Thereby, the inverse quantization unit 306 may form a transform coefficient block including the transform coefficients.

[0200] After the inverse quantization unit 306 forms the transform coefficient block, the inverse transform processing unit 308 may apply one or more inverse transforms to the transform coefficient block to generate the residual block associated with the current block. For example, the inverse transform processing unit 308 may apply an inverse DCT, an inverse integer transform, an inverse Karhunen - Loève transform (KLT), an inverse rotation transform, an inverse direction transform, or another inverse transform to the transform coefficient block.

[0201] Furthermore, the prediction processing unit 304 generates a prediction block according to the prediction information syntax element entropy-decoded by the entropy decoding unit 302. For example, if the prediction information syntax element indicates that the current block is inter-predicted, the motion compensation unit 316 may generate a prediction block. In this case, the prediction information syntax element may indicate the reference picture in the DPB 314 from which the reference block is to be taken, as well as a motion vector that identifies the location of the reference block in the reference picture relative to the location of the current block in the current picture. The motion compensation unit 316 can generally perform the inter-prediction process in a manner substantially similar to the manner described with respect to the motion compensation unit 224 (FIG. 2).

[0202] As another example, if the prediction information syntax element indicates that the current block is intra-predicted, the intra-prediction unit 318 may generate a prediction block according to the intra-prediction mode indicated by the prediction information syntax element. Also in this case, the intra-prediction unit 318 can generally perform the intra-prediction process in a manner substantially similar to the manner described with respect to the intra-prediction unit 226 (FIG. 2). The intra-prediction unit 318 may retrieve data of adjacent samples for the current block from the DPB 314.

[0203] The reconstruction unit 310 may reconstruct the current block using the prediction block and the residual block. For example, the reconstruction unit 310 may add the samples of the residual block to the corresponding samples of the prediction block to reconstruct the current block.

[0204] The filter unit 312 may perform one or more filter operations on the reconstructed block. For example, the filter unit 312 may perform a deblocking operation to reduce blockiness artifacts along the edges of the reconstructed block. The operation of the filter unit 312 is not necessarily performed in all cases.

[0205] According to one or more examples described in the present disclosure, the filter unit 312 may be configured to perform a neural network-based filter processing technique. For example, the filter unit 312 may perform a neural network-based filter processing technique in addition to or instead of other filter processing techniques such as ALF and SAO. In some examples, the filter unit 312 may perform the neural network-based filter processing described in the present disclosure as part of in-loop filter processing. Although not shown, in some examples, a filter unit may be coupled to the output of the DPB 314, and the output from this filter unit may be the decoded video. In some examples, the filter unit coupled to the output of the DPB 314 may be configured to perform an exemplary neural network-based filter processing technique described in the present disclosure as part of post-loop filter processing.

[0206] The video decoder 300 may store the reconstructed block in the DPB 314. For example, in an example where the operation of the filter unit 312 is not performed, the reconstruction unit 310 may store the reconstructed block in the DPB 314. In an example where the operation of the filter unit 312 is performed, the filter unit 312 may store the filtered and reconstructed block in the DPB 314. As described above, the DPB 314 may provide reference information such as samples of the current picture for intra prediction and previously decoded pictures for subsequent motion compensation to the prediction processing unit 304. Further, the video decoder 300 may output the decoded picture (e.g., the decoded video) from the DPB 314 for later display on a display device such as the display device 118 of FIG. 1.

[0207] As such, video decoder 300 represents an example of a video decoding device that includes a memory configured to store video data, and one or more processing units implemented within a circuit, the one or more processing units being configured to filter two or more color components of the video data using a neural network (NN)-based filter from a single NN model and output two or more color components after a single filtering process using the NN-based filter. In some examples, filtering includes how to apply the NN-based filter, whether to turn on or off the NN-based filter, how to create input data for a single NN model, and determining the values of all input elements in the same manner for two or more color components.

[0208] In some examples, video decoder 300 may be configured to receive information indicating whether the NN-based filter is on or off for each of two or more color components. In such examples, filtering includes filtering based on whether the NN-based filter is on or off for each of two or more color components.

[0209] For example, filter unit 312 may receive a syntax element that defines a filtering mode for a neural network (NN) model for both a first color component and a second color component. Examples of syntax elements include the first-level filter mode syntax element, the second-level filter mode syntax element, first_chroma_mode_slice (e.g., where first_chroma_mode_slice is not zero and second_chroma_flag_slice is true), and chroma_mode_slice (e.g., where first_chroma_flag_slice and / or second_chroma_flag_slice is true), as described above.

[0210] In some examples, a syntax element defines which parameter to select from a plurality of parameters, and the defined parameter defines a filtering mode. As an example, the parameter is a quantization parameter, and the plurality of parameters is one of a predefined set or a signaled set. For example, a value of 1 for a first-level filter mode syntax element indicates that the filtering mode is defined by a first parameter within a predefined set or a signaled set, and a value of 2 for the first-level filter mode syntax element indicates that the filtering mode is defined by a second parameter within a predefined set or a signaled set, and a value of 3 for the first-level filter mode syntax element indicates that the filtering mode is defined by a third parameter within a predefined set or a signaled set. Values for second-level filter mode syntax elements, first_chroma_mode_slice, and chroma_mode_slice may similarly define a filtering mode as described in the example above.

[0211] The filter unit 312 may apply an instance of the NN model to a first block of a first color component in a defined filtering mode to generate a first filtered block. Also, the filter unit 312 may apply the same instance of the NN model to a second block of a second color component in the defined filtering mode to generate a second filtered block. The filter unit 312 may store a value for the CU based on the first filtered block. In instances where NN-based filtering is enabled for both the first color component and the second color component, the filter unit 312 may store a value for the CU based on the first filtered block and the second filtered block. As described above, a CU is used as a term to describe a composite block that includes a plurality of different color components.

[0212] Thus, instead of a syntax element defining a filtering mode separately for each of the color components, one syntax element defines a filtering mode for two or more color components. For example, in the execution of the NN model in the defined filtering mode, the filter unit 312 may generate a first filtered block for the first color component and a second filtered block for the second color component. The filter unit 312 stores a value for the CU based on the first filtered block, and may store a value for the CU based on the first filtered block and the second filtered block if filtering is enabled for the second color component.

[0213] Storing a value for the CU based on the first filtered block and / or the second filtered block may mean that the samples stored in the DPB 314 for the CU are the first filtered block and the second filtered block, or, if filtering is not enabled for the second color component, the first filtered block and the second block (without filtering). As another example, the decoded video data output by the DPB 314 may include the first filtered block and the second filtered block if filtering is enabled for both, or the first filtered block and the second block (without filtering) if filtering is not enabled for the second color component.

[0214] As an example, the syntax element can be a syntax element applicable to a plurality of CUs. For example, the syntax element can be a first-level filter mode syntax element that defines a filtering mode for a plurality of CUs (e.g., CUs within a slice). As another example, the syntax element can be a second-level filter mode syntax element that defines a filtering mode for a subset of CUs. For example, the video decoder 300 can receive a first syntax element applicable to a plurality of CUs that instructs that a second syntax element is to be parsed for a subset of the plurality of CUs (e.g., in the above example, the first-level filter mode syntax element is 4). In this example, the subset of CUs includes one or more CUs including the CU in which the value is stored. The video decoder 300 can receive a second syntax element (e.g., a second-level filter mode syntax element) based on the first syntax element (e.g., the first-level filter mode syntax element) that instructs that the second syntax element is to be parsed for a subset of CUs. That is, when the value for the first-level syntax element is 4 in the above example, the video decoder 300 can receive a second-level syntax element having values from 0 to 3.

[0215] As described above, whether the second filtered block (e.g., generated from NN-based filter processing) is used or not can be set based on the signaled parameter. For example, the syntax element received by the video decoder 300 that defines the filter processing mode for the NN model for both the first color component and the second color component may have a value indicating that the NN-based filter processing is enabled for the first color component. For example, the value for first_chroma_mode_slice may be non-zero. In this example, if a flag (e.g., second_chroma_flag_slice) indicates that the NN-based filter processing is enabled for the second color component, the filter unit 312 can generate a second filtered block based on the second block of the second color component, since it applies the same instance of the NN model to the second block in the defined filter processing mode.

[0216] That is, the first_chroma_mode_slice value may define the filter processing mode for the NN model, and if the filter processing is enabled for the second color component, the value of first_chroma_mode_slice defines the filter processing mode of the NN model for the second color component. Thus, in this example, the filter unit 312 can utilize both the first filtered block and the second filtered block generated by applying the same instance of the NN model in the defined filter processing mode to store the value for the CU.

[0217] However, in an example where the second_chroma_flag_slice is false, indicating that the NN-based filtered one is invalidated for the second color component, the filter unit 312 may store values for the CU based on the first filtered block and the second block of the second color component. That is, the filter unit 312 may discard the second filtered block of the second color component and instead store the first filtered block and the second block (e.g., without NN filtering) in the DPB 314.

[0218] The use of the second_chroma_flag_slice is an example. In some examples, such as an example where the syntax defining the filter processing mode for both color components is chroma_mode_slice, the video decoder 300 may receive a flag indicating that the NN-based filtering is enabled for the first color component (e.g., receive the first_chroma_flag_slice or the second_chroma_flag_slice). In such an example, the video decoder 300 may receive a syntax element (e.g., chroma_mode_slice) based on the flag indicating that the NN-based filtering is enabled for the first color component.

[0219] FIG. 4 is a flowchart showing an exemplary method of encoding a current block according to the techniques of the present disclosure. The current block may include a current CU. It should be understood that, although described with respect to the video encoder 200 (FIGS. 1 and 2), other devices may be configured to implement a method similar to the method of FIG. 4.

[0220] In this example, video encoder 200 first predicts (350) the current block. For example, video encoder 200 may form a predicted block for the current block. Video encoder 200 may then calculate (352) a residual block for the current block. To calculate the residual block, video encoder 200 may calculate the difference between the original unencoded block and the predicted block for the current block. Video encoder 200 may then transform the residual block and quantize the transform coefficients of the residual block (354). Next, video encoder 200 may scan (356) the quantized transform coefficients of the residual block. During or following the scan, video encoder 200 may entropy encode (358) the transform coefficients. For example, video encoder 200 may encode the transform coefficients using CAVLC or CABAC. Video encoder 200 may then output (360) the entropy encoded data of the block.

[0221] FIG. 5 is a flowchart illustrating an exemplary method of decoding a current block of video data in accordance with the techniques of the present disclosure. The current block may include a current CU. Although described with respect to video decoder 300 (FIGS. 1 and 3), it should be understood that other devices may be configured to implement a method similar to the method of FIG. 5.

[0222] Video decoder 300 may receive entropy-coded data for the current block, such as entropy-coded prediction information and entropy-coded data for the transform coefficients of the residual block corresponding to the current block (370). Video decoder 300 may entropy-decode the entropy-coded data to determine prediction information for the current block and reproduce the transform coefficients of the residual block (372). Video decoder 300 may predict the current block using, for example, an intra prediction mode or an inter prediction mode indicated by the prediction information for the current block to calculate a prediction block for the current block (374). Video decoder 300 may then inverse scan the reproduced transform coefficients to generate a block of quantized transform coefficients (376). Video decoder 300 may then inverse quantize the transform coefficients and apply an inverse transform to the transform coefficients to generate a residual block (378). Video decoder 300 may finally decode the current block by combining the prediction block and the residual block (380).

[0223] In one or more examples, video decoder 300 may also be configured to perform neural network-based filtering according to one or more examples described in this disclosure. For example, filter unit 312 may be configured to perform in-loop filtering using neural network-based filtering. In some examples, a filter unit coupled to the output of DPB 314 may be configured to perform filtering using exemplary neural network-based filtering described in this disclosure as part of post-loop filtering.

[0224] Although FIG. 5 is described with respect to video decoder 300, in some examples, video encoder 200 also includes a reconstruction process, and video encoder 200 may be configured to perform the exemplary techniques of FIG. 5, for example, through inverse quantization unit 210, inverse transform processing unit 212, and reconstruction unit 214. In some examples, filter unit 216 may be configured to perform neural network-based filter processing techniques described in this disclosure. For example, filter unit 216 may perform in-loop filter processing as part of the reconstruction loop of video encoder 200.

[0225] FIG. 9 is a flowchart illustrating an exemplary method of processing video data according to the techniques of this disclosure. For ease of explanation, the example is described with respect to video decoder 300, but video encoder 200 may perform similar operations.

[0226] Video decoder 300 may receive (900) a syntax element that defines a filter processing mode for a neural network (NN) model for both a first color component and a second color component. Examples of syntax elements include the first-level filter mode syntax element, the second-level filter mode syntax element, first_chroma_mode_slice (e.g., where first_chroma_mode_slice is not 0 and second_chroma_flag_slice is true), and chroma_mode_slice (e.g., where first_chroma_flag_slice and / or second_chroma_flag_slice is true) described above. In some examples, the syntax element defines which parameter to select from a plurality of parameters, and the defined parameter defines the filter processing mode. As an example, the parameter is a quantization parameter, and the plurality of parameters is one of a predefined set or a signaled set.

[0227] As described above, the first-level filter mode syntax element can have a value from 0 to 4. A value of 0 may indicate that (as an example) the NN-based filtering is disabled for the color components of the blocks within a slice. A value of 1, 2, or 3 may indicate the filtering mode for the NN model for both the first color component and the second color component (e.g., which of the first QP, the second QP, or the third QP from a set should be used, and defining the filtering mode defined by the first, second, or third QP from the set). A value of 4 may indicate that the filtering mode for the NN model is defined at a lower level (e.g., at the CTU level or the CU level) by the second-level filter mode syntax element. A value of 0 for the second-level filter mode syntax element may indicate that (as an example) the NN-based filtering is disabled for the color components. A value of 1, 2, or 3 may indicate the filtering mode for the NN model for both the first color component and the second color component (e.g., which of the first QP, the second QP, or the third QP from a set should be used, and defining the filtering mode defined by the first, second, or third QP from the set).

[0228] first_chroma_mode_slice is another example of a syntax element that defines the filtering mode for a neural network (NN) model for both the first color component and the second color component. For example, if first_chroma_mode_slice has a non-zero value and the NN-based filtering is enabled for the second color component, then first_chroma_mode_slice defines the filtering mode for both the first color component and the second color component at this time.

[0229] chroma_mode_slice is another example of a syntax element that defines the filtering mode of a neural network (NN) model for both a first color component and a second color component. For example, if first_chroma_flag_slice and / or second_chroma_flag_slice is true, the value of chroma_mode_slice can define the filtering mode (e.g., whether a first QP value from a set is used to define the filtering mode, a second QP value is used, or a third QP value is used).

[0230] The filter unit 312 can apply an instance of the NN model to a first block of the first color component in a defined filtering mode to generate a first filtered block (902). For example, the filter unit 312 can apply an instance of the NN model to a first block of the first color component in a defined filtering mode to generate a first residual value, and generate a first filtered block based on the first block and the first residual value (e.g., by summing the first residual value and the first filtered block).

[0231] In some examples, when NN-based filtering is enabled for the first color component, at this time, the NN-based filtering is automatically enabled for the second color component. That is, in some examples, it may not be possible to selectively not use the filtering for the second color component. Thus, in one or more examples, the filter unit 312 may apply the same instance of the NN model in a defined filtering mode to a second block of the second color component to generate a second filtered block (906). As described above, applying the same instance of the NN model means that the NN model receives both a first block of the first color component and a second block of the second color component (including, for example, the luma component) as inputs and generates as outputs a first filtered block of the first color component and a second filtered block of the second color component.

[0232] That is, in some examples, the NN model for the chroma components (e.g., the first color component and the second color component) receives the luma component as an input as well, but may output a first filtered block of the first color component and a second filtered block of the second color component. The NN model may not generate a filtered luma block, but in some examples, it may be possible for the NN model to generate a filtered luma block.

[0233] In this case, the filter unit 312 may store the first filtered block and the second filtered block (908). For example, the sample values that the filter unit 312 stores in the DPB 314 for a CU may be the values of the first filtered block and the second filtered block.

[0234] However, in some examples, it may not always be true to utilize the second filtered block and it may be selectable. For example, in some examples, the filter unit 312 may determine whether NN-based filtering for the second color component is enabled (904). For example, the filter unit 312 may determine the value of first_chroma_flag_slice or second_chroma_flag_slice.

[0235] For example, the syntax element defining the filtering mode may be first_chroma_mode_slice. If the value of first_chroma_mode_slice is not 0, then at this time, the filter unit 312 may determine the value of a flag (for example, second_chroma_flag_slice). That is, the filter unit 312 may receive a flag indicating that NN-based filtering is enabled for the second color component (yes in 904). In this example, the filter unit 312 applies the same instance of the NN model to the second block of the second color component in the defined filtering mode to generate a second filtered block (906) and may store the first filtered block and the second filtered block (908).

[0236] However, in some examples, the filter unit 312 may receive a flag indicating that NN-based filtering is disabled for the second color component (no in 904). In this example, the filter unit 312 may store the first filtered block and the second block (for example, a block not filtered by NN) (910).

[0237] The following describes exemplary techniques according to one or more examples described in the present disclosure.

[0238] Clause 1A. A method for processing video data, comprising: filtering two or more color components of the video data using a neural network (NN)-based filter from a single NN model; and outputting two or more color components after a single filtering process using a single NN-based filter.

[0239] Clause 2A. The method according to clause 1A, wherein filtering includes how to apply the NN-based filter, whether to turn on or turn off the NN-based filter, how to create input data for a single NN model, and determining the values of all input elements in the same way for two or more color components.

[0240] Clause 3A. Further comprising receiving or signaling information indicating whether the NN-based filter is on or off for each of two or more color components, and filtering being based on whether the NN-based filter is on or off for each of two or more color components.

[0241] Clause 4A. The method according to any one of clauses 1A to 3A, further comprising reconstructing sample values of a picture, and filtering including filtering two or more color components of the reconstructed sample values of the picture.

[0242] Clause 5A. The method according to clause 4A, wherein filtering includes in-loop filtering in a video encoder or a video decoder.

[0243] Clause 6A. The method according to clause 4A, wherein filtering includes post-loop filtering in a video decoder.

[0244] Clause 7A. A device for processing video data, comprising a memory configured to store video data and a processing circuit configured to perform the method according to any one of Clauses 1A to 6A.

[0245] Clause 8A. The device according to Clause 7A, further comprising a display configured to display the decoded video data.

[0246] Clause 9A. The device according to Clause 7A or 8A, wherein the device comprises one or more of a camera, a computer, a mobile device, a broadcast receiver device, or a set-top box.

[0247] Clause 10A. The device according to any one of Clauses 7A to 9A, wherein the device comprises a video decoder.

[0248] Clause 11A. The device according to any one of Clauses 7A to 10A, wherein the device comprises a video encoder.

[0249] Clause 12A. A computer-readable storage medium storing instructions which, when executed, cause one or more processors to perform the method according to any one of Clauses 1A to 6A.

[0250] Clause 13A. A device for processing video data, comprising means for performing the method according to any one of Clauses 1A to 6A.

[0251] Clause 1B. A method for processing video data, comprising receiving a syntax element defining a filter processing mode for a neural network (NN) model for both a first color component and a second color component; applying an instance of the NN model in the defined filter processing mode to a first block of the first color component to generate a first filtered block; and storing the first filtered block for a coding unit (CU).

[0252] Clause 2B. Further including applying the same instance of the NN model to the second block of the second color component in a defined filtering mode to generate a second filtered block, and storing includes storing the first filtered block and the second filtered block for the CU, the method according to Clause 1B.

[0253] Clause 3B. Receiving a syntax element that is applicable to a plurality of CUs, the method according to Clause 1B or 2B, where the CU is one of the plurality of CUs.

[0254] Clause 4B. The syntax element is a second syntax element, and the method further includes receiving a first syntax element that is a first syntax element applicable to a plurality of CUs, and the second syntax element is parsed for a subset of the plurality of CUs including one or more CUs including the CU, and receiving the syntax element includes receiving the second syntax element based on the first syntax element that instructs that the second syntax element is parsed for a subset of the CUs, the method according to any one of Clauses 1B to 3B.

[0255] Clause 5B. The syntax element has a value that indicates that NN-based filtering is enabled for the first color component, the method according to any one of Clauses 1B to 4B.

[0256] Clause 6B. Further including receiving a flag that indicates that NN-based filtering is enabled for the second color component, and applying the same instance of the NN model to the second block in a defined filtering mode to generate a second filtered block based on the second block of the second color component, and storing includes storing the first filtered block and the second filtered block for the CU, the method according to Clause 5B.

[0257] The method according to Clause 5B, further comprising receiving and storing a flag indicating that the NN-based filtering process is disabled for the second color component, wherein storing comprises storing the first filtered block and the second block of the second color component for the CU.

[0258] The method according to any one of Clauses 1B to 7B, further comprising receiving a flag indicating that the NN-based filtering process is enabled for the first color component, wherein receiving a syntax element comprises receiving the syntax element based on a flag indicating that the NN-based filtering process is enabled for the first color component.

[0259] The method according to any one of Clauses 1B to 8B, wherein the syntax element defines which parameter to select from a plurality of parameters, and the defined parameter defines a filtering mode.

[0260] The method according to Clause 9B, wherein the parameter is a quantization parameter and the plurality of parameters are one of a predefined set or a signaled set.

[0261] The method according to any one of Clauses 1B to 10B, wherein the first color component is the first chroma component and the second color component is the second chroma component.

[0262] The method according to any one of Clauses 1B to 11B, wherein to generate the first filtered block, applying an instance of the NN mode in the defined filtering mode to the first block of the first color component, to generate the first residual value, applying an instance of the NN model in the defined filtering mode to the first block of the first color component, and generating the first filtered block based on the first block and the first residual value.

[0263] Clause 13B. A device for processing video data, comprising a memory configured to store video data, and one or more processors implemented in a circuit and coupled to the memory, the one or more processors being configured to receive a syntax element that defines a filtering mode for a neural network (NN) model for both a first color component and a second color component, and to apply an instance of the NN model in the defined filtering mode to a first block of the first color component to generate a first filtered block and store the first filtered block in the memory for a coding unit (CU).

[0264] Clause 14B. The device according to Clause 13B, wherein one or more processors are configured to apply the same instance of the NN model in the defined filtering mode to a second block of the second color component to generate a second filtered block, and the one or more processors are configured to store the first filtered block and the second filtered block for the CU.

[0265] Clause 15B. The device according to Clause 13B or 14B, wherein one or more processors are configured to receive a syntax element applicable to a plurality of CUs, and the CU is one of the plurality of CUs.

[0266] Clause 16B. The syntax element is a second syntax element, and one or more processors are configured to receive a first syntax element applicable to a plurality of CUs, where the second syntax element is analyzed for a subset of the plurality of CUs, including one or more CUs containing a CU, and the first syntax element instructs that the second syntax element be analyzed for the subset of CUs. To receive the syntax element, one or more processors are configured to receive the second syntax element based on the first syntax element that instructs that the second syntax element be analyzed for the subset of CUs. The device according to any one of Clauses 13B to 15B.

[0267] Clause 17B. The device according to any one of Clauses 13B to 16B, where the syntax element has a value that instructs that NN-based filtering is enabled for a first color component.

[0268] Clause 18B. One or more processors are configured to receive a flag that instructs that NN-based filtering is enabled for a second color component, and apply the same instance of the NN model to a second block in a defined filtering mode, based on the second block of the second color component, to generate a second filtered block. To store, one or more processors are configured to store the first filtered block and the second filtered block for the CU. The device according to Clause 17B.

[0269] Clause 19B. One or more processors are configured to receive a flag that instructs that NN-based filtering is disabled for a second color component, and to store, for the CU, the first filtered block and the second block of the second color component. The device according to Clause 17B.

[0270] Clause 20B. One or more processors are configured to receive a flag indicating that NN-based filtering is enabled for a first color component, and to receive a syntax element based on the flag indicating that NN-based filtering is enabled for the first color component, for receiving the syntax element, the device according to any one of Clauses 13B to 19B.

[0271] Clause 21B. The device according to any one of Clauses 13B to 20B, wherein the syntax element defines which parameter to select from a plurality of parameters, and the defined parameter defines a filtering mode.

[0272] Clause 22B. The device according to Clause 21B, wherein the parameter is a quantization parameter, and the plurality of parameters is one of a predefined set or a signaled set.

[0273] Clause 23B. The device according to any one of Clauses 13B to 22B, wherein the first color component is a first chroma component and the second color component is a second chroma component.

[0274] Clause 24B. For generating a first filtered block, one or more processors are configured to apply an instance of the NN mode in a defined filtering mode to a first block of a first color component to generate a first residual value, and to generate a first filtered block based on the first block and the first residual value. The device according to any one of Clauses 13B to 23B.

[0275] Clause 25B. A computer-readable storage medium storing instructions which, when executed, cause one or more processors to receive syntax elements that define a filter processing mode for a neural network (NN) model for both a first color component and a second color component, apply an instance of the NN model in the defined filter processing mode to a first block of the first color component to generate a first filtered block, and store the first filtered block for a coding unit (CU).

[0276] It should be recognized that, in some examples, some of the acts or events of any of the techniques described herein may be performed in a different order, may be added, merged, or entirely excluded (e.g., not all acts or events described are necessary for the practice of the technique). Additionally, in some examples, the acts or events may not be performed sequentially, e.g., may be performed in parallel through multi-threaded processing, interrupt processing, or multiple processors.

[0277] In one or more examples, the functions described may be implemented in hardware, software, firmware, or any combination thereof. If implemented in software, the functions may be stored on or transmitted via a computer-readable recording medium as one or more instructions or code and executed by a hardware-based processing unit. The computer-readable recording medium may include a computer-readable storage medium corresponding to a tangible medium such as a data storage medium, or a communication medium including any medium that facilitates transfer of a computer program from one place to another, for example, according to a communication protocol. Thus, the computer-readable recording medium generally may correspond to (1) a tangible computer-readable storage medium that is non-transitory, or (2) a communication medium such as a signal or carrier wave. The data storage medium may be any available medium that can be accessed by one or more computers or one or more processors to retrieve instructions, code, and / or data structures for implementing the techniques described in this disclosure. A computer program product may include a computer-readable recording medium.

[0278] By way of example and not limitation, such computer-readable storage media can include RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage, or other magnetic storage devices, flash memory, or any other medium that can be used to store the desired program code in the form of instructions or data structures and that can be accessed by a computer. Also, any connection can be properly termed a computer-readable recording medium. For example, if the instructions are transmitted from a website, server, or other remote source using coaxial cable, fiber optic cable, twisted pair, digital subscriber line (DSL), or wireless technologies such as infrared, radio, and microwave, then the coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwave are included in the definition of the medium. However, it should be understood that computer-readable storage media and data storage media do not include connections, carrier waves, signals, or other transient media, but instead are directed to non-transitory tangible storage media. As used herein, disk and disc include compact disc (CD), laser disc, optical disc, digital versatile disc (DVD), floppy disk, and Blu-ray disc, where disk typically magnetically reproduces data, while disc optically reproduces data using a laser. Combinations of the above should also be included within the scope of computer-readable recording media.

[0279] The commands may be executed by one or more processors, such as one or more DSPs, general-purpose microprocessors, ASICs, FPGAs, or other equivalent integrated logic circuits or discrete logic circuits. Thus, the terms "processor" and "processing circuit" as used herein may refer to either the foregoing structures, or any other structure suitable for implementation of the techniques described herein. Additionally, in some aspects, the functions described herein may be provided within dedicated hardware modules and / or software modules configured for encoding and decoding, or may be incorporated within a combined codec. Also, the techniques may be implemented entirely in one or more circuits or logic elements.

[0280] The techniques of the present disclosure may be implemented in a variety of devices or apparatuses, including wireless handsets, integrated circuits (ICs), or sets of ICs (e.g., chip sets). To emphasize the functional aspects of devices configured to implement the disclosed techniques, various components, modules, or units have been described in this disclosure, but they do not necessarily require realization by different hardware units. Rather, as described above, the various units may be combined in a codec hardware unit, or provided by a set of interoperable hardware units that include one or more of the processors described above in conjunction with suitable software and / or firmware.

[0281] Various examples have been described. These and other examples fall within the scope of the following claims.

Claims

1. A method for processing video data, It receives syntax elements that define filtering modes for a neural network (NN) model for both the first and second color components, To generate a first filtered block, the execution instance of the NN model is applied to the first block of the first color component in the defined filtering mode, To generate a second filtered block, the same execution instance of the NN model is applied to the second block of the second color component in the defined filtering mode, wherein both the first filtered block and the second filtered block are generated after a single filtering by the application of the execution instance of the NN model. The first filtered block and the second filtered block are stored for the coding unit (CU), Methods that include...

2. The method according to claim 1, wherein receiving the syntax element includes receiving the syntax element applicable to a plurality of CUs, and the CU is one of the plurality of CUs.

3. The syntax element is a second syntax element, and the method is A first syntax element applicable to multiple CUs, further comprising receiving a first syntax element indicating that the second syntax element is to be parsed for a subset of the multiple CUs, including one or more CUs, which include the CU, The method according to claim 1, wherein receiving the syntax element includes receiving the second syntax element based on the first syntax element indicating that the second syntax element is to be parsed for the subset of CUs.

4. The method according to claim 1, wherein the syntax element has a value indicating that NN-based filtering is enabled for the first color component.

5. Receiving a flag indicating that NN-based filtering is enabled for the second color component, Applying the same execution instance of the NN model to the second block in the defined filtering mode generates a second filtered block based on the second block of the second color component, It further includes, The method according to claim 4, wherein storing includes storing the first filtered block and the second filtered block for the CU.

6. The further includes receiving a flag indicating that NN-based filtering is disabled for the second color component, The method according to claim 4, wherein storing includes storing the first filtered block and the second block of the second color component for the CU.

7. The further includes receiving a flag indicating that NN-based filtering is enabled for the first color component, The method according to claim 1, wherein receiving the syntax element includes receiving the syntax element based on a flag indicating that NN-based filtering is enabled for the first color component.

8. The method according to claim 1, wherein the syntax element defines which parameter should be selected from a plurality of parameters, and the defined parameter defines the filtering mode.

9. The method according to claim 8, wherein the parameter is a quantization parameter, and the plurality of parameters are one of a predefined set or a signaled set.

10. The method according to claim 1, wherein the first color component is a first chroma component and the second color component is a second chroma component.

11. To generate the first filtered block, the execution instance of the NN model is applied to the first block of the first color component in the defined filtering mode. To generate a first residual value, the execution instance of the NN model is applied to the first block of the first color component in the defined filtering mode, and To generate the first filtered block based on the first block and the first residual value, The method according to claim 1, including the method described in claim 1.

12. A device that processes video data, A memory configured to store the aforementioned video data, One or more processors implemented within the circuit and coupled to the memory, It receives a syntax element that defines a filtering mode for a neural network (NN) model for both the first and second color components, To generate a first filtered block, the running instance of the NN model is applied to the first block of the first color component in the defined filtering mode, To generate a second filtered block, the same execution instance of the NN model is applied to the second block of the second color component in the defined filtering mode, so that both the first filtered block and the second filtered block are generated after a single filtering by the application of the execution instance of the NN model. The memory stores the first filtered block and the second filtered block for the coding unit (CU). One or more processors configured as such, A device equipped with the following features.

13. The device according to claim 12, wherein, in order to receive the syntax elements, one or more processors are configured to receive the syntax elements applicable to a plurality of CUs, and the CU is one of the plurality of CUs.

14. The device according to claim 12, further comprising means for performing the method described in any one of claims 3 to 11.

15. A computer-readable storage medium storing instructions, wherein when the instructions are executed, one or more processors cause the method according to any one of claims 1 to 11 to be executed.

Images