Secondary transform assignment for NSPT / lfnst coded blocks in video coding

Abstract

A method of decoding video data includes storing information indicative of which coding modes are associated with which category of a plurality of categories, and each category is associated with a respective set of transform kernels, determining that a current block has size for which non-separable primary transform (NSPT) is applied, determining that the current block is coded in a first mode associated with a first category, determining an angular mode for the current block, the angular mode being different than the first mode, determining a transform kernel for NSPT for the current block based on the angular mode and transform kernels associated with a second category, performing, based on the transform kernel, an inverse transform on a transform coefficient block to generate a residual block associated with the current block, and reconstructing the current block based on the residual block.

Description

SECONDARY TRANSFORM ASSIGNMENT FOR NSPT / LFNST CODED BLOCKS IN VIDEO CODING

[0001] This application claims the benefit of U.S. Provisional Application No.63 / 733,954, filed December 13, 2024, the entire contents of which are hereby incorporated by reference.TECHNICAL FIELD

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

[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 radio telephones, so-called “smart phones,” video teleconferencing devices, video streaming devices, and the like. Digital video devices implement video coding techniques, such as those described in the 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 Video 1 (AVI) that was developed by the Alliance for Open Media. The video devices may transmit, receive, encode, decode, and / or store digital video information more efficiently 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 video sequences. For 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 in an intra-coded (1) slice of a picture are encoded using spatial prediction with respect to reference samples in neighboring blocks in the same picture. Video blocks in an inter-coded (P or B) slice of a picture may use spatial prediction with respect to reference samples in neighboring blocks in the same picture or temporal prediction with 1616-600W001respect to reference samples in other reference pictures. Pictures may be referred to as frames, and reference pictures may be referred to as reference frames.SUMMARY

[0005] In general, this disclosure describes techniques for video coding, including techniques for secondary transform assignment for blocks of video data coded using a non-separable primary transform (NSPT) and / or a low-frequency non-separable transform (LFNST). The techniques of the disclosure include storing information that maps coding modes to a plurality of categories, where each category is associated with a respective set of transform kernels. A video coder (e.g., video encoder or video decoder) may determine that a current block has a size for which NSPT is applied and determine that the current block is coded in a first mode associated with a first category. The video coder also determines an angular mode for the current block, where the angular mode is different than the first mode, for example, by using a histogram of gradient (HoG) process. The video coder may determine a transform kernel for NSPT for the current block based on the determined angular mode and the transform kernels associated with a second category, where the second category is different than the first category.

[0006] This approach introduces greater diversity to the available transform kernels, which benefits coding efficiency, especially for prediction blocks lacking clear directionality. For example, even though the current block is coded in a first mode associated with a first category it is possible to determine the transform kernel for the current block based on transform kernels associated with a second, different category.

[0007] A video decoder, for example, may perform, based on the determined transform kernel, an inverse transform on a transform coefficient block to generate a residual block. The video decoder reconstructs the current block based on the residual block. A video encoder may perform, based on the determined transform kernel, a forward transform on a residual block to generate a transform coefficient block and signal information based on the transform coefficient block. The video coder may determine the transform kernel based on the second category in a condition where a secondary transform set is to be used. For instance, the video decoder may determine tire transform kernel based on the second category in response to parsing information indicating that a secondary transform set is to be used.1616-600W001

[0008] In one example, this disclosure describes a method of decoding video data, the method comprising storing information indicative of which coding modes are associated with which category of a plurality of categories, and each category is associated with a respective set of transform kernels, determining that a current block has size for which non-separable primary transform (NSPT) is applied, determining that the current block is coded in a first mode associated with a first category of the plurality of categories, determining an angular mode for the current block, the angular mode being different than the first mode, determining a transform kernel for NSPT for the current block based on the angular mode and transform kernels associated with a second category, the second category being different than the first category, performing, based on the transform kernel, an inverse transform on a transform coefficient block to generate a residual block associated with the current block, and reconstructing the current block based on the residual block.

[0009] In another example, this disclosure describes a device for decoding video data, the device comprising one or more memories configured to store information indicative of which coding modes are associated w ith which category of a plurality of categories, and each category is associated with a respective set of transform kernels, and processing circuitry coupled to the one or more memories, the processing circuitry being configured to determine that a current block has size for which non-separable primary transform (NSPT) is applied, determine that the current block is coded in a first mode associated with a first category of the plurality of categories, determine an angular mode for the current block, the angular mode being different than the first mode, determine a transform kernel for NSPT for the current block based on the angular mode and transform kernels associated with a second category, the second category being different than the first category, perform, based on the transform kernel, an inverse transform on a transform coefficient block to generate a residual block associated with the current block, and reconstruct the current block based on the residual block.

[0010] In another example, this disclosure describes a device for encoding video data, the device comprising one or more memories configured to store information indicative of which coding modes are associated with which categories of a plurality of categories, and each category is associated with a respective set of transform kernels, and processing circuitry coupled to the one or more memories, the processing circuitry being configured to determine that a current block has size for which non-separable primary transform (NSPT) is applied, determine that the current block is coded in a first mode 1616-600W001associated with a first category of the plurality of categories, determine an angular mode for the current block, the angular mode being different than the first mode, determine a transform kernel for NSPT for the current block based on the angular mode and transform kernels associated with a second category, the second category’ being different than the first category, perform, based on the transform kernel, a transform on a residual block to generate a transform coefficient block associated with the current block, and signal information based on the transform coefficient block.

[0011] The 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 DRAWINGS

[0012] FIG. 1 is a block diagram illustrating an example video encoding and decoding system that may perform the techniques of this disclosure.

[0013] FIG. 2 is a block diagram illustrating an example video encoder that may perform the techniques of this disclosure,

[0014] FIG. 3 is a block diagram illustrating an example video decoder that may perform the techniques of this disclosure.

[0015] FIG. 4 is a flowchart illustrating an example method for encoding a current block in accordance with the techniques of this disclosure.

[0016] FIG. 5 is a flowchart illustrating an example method for decoding a current block in accordance with the techniques of this disclosure.

[0017] FIG. 6 is a conceptual diagram illustrating a region of interest for low-frequency non-separable transform (LFNST) 16.

[0018] FIG. 7 is a conceptual diagram illustrating a region of interest for LFNST8.

[0019] FIG. 8 is a table illustrating a mapping of intra prediction modes to a LFN ST set index,

[0020] FIG. 9 is a conceptual diagram illustrating example Mapping from Intra Prediction (MIP) prediction samples to build a histogram of gradient (HoG).

[0021] FIG. 10 is a conceptual diagram illustrating an overview of non-separable primary’ transforms (NSPTs) among existing LFNSTs.

[0022] FIG. 11 is a conceptual diagram illustrating regular intra prediction modes in VVC Test Model (VTM) 7.0.1616-600W001

[0023] FIG. 12 is a conceptual diagram illustrating an example matrix intra prediction process.

[0024] FIG. 13 is a conceptual diagram illustrating an example histogram of gradient computation.

[0025] FIG. 14 is a conceptual diagram illustrating an example of weight determination and final predictor generation.

[0026] FIG, 15 is a conceptual diagram illustrating an example template and reference samples used in template-based intra model derivation with fusion (TIMD).

[0027] FIG. 16 is a conceptual diagram illustrating an example of spatial geometric partitioning mode.

[0028] FIG. 17 is a flowchart illustrating an example operation of a video decoder, in accordance with one or more techniques of this disclosure.

[0029] FIG. 18 is a flowchart illustrating an example operation of a video encoder, in accordance with one or more techniques of this disclosure.DETAILED DESCRIPTION

[0030] Video coding techniques may apply non-separable transforms, such as a non- separable primary transform (NSPT) or a low-frequency non-separable transform (LFNST), to blocks of video data. For example, for video encoding, a video encoder determines a prediction block for a current block and residual block based on a difference between the current block and the prediction block. Tire video encoder performs a transform to transform the residual block from sample domain to frequency domain by applying a transform kernel for the NSPT or LFNST to generate a transform coefficient block. The video encoder may then signal information based on the transform coefficient block (e.g., after quantization and entropy encoding).

[0031] The video decoder may also determine a prediction block for the current block using the same techniques as the video encoder. The video decoder may generate a transform coefficient block based on tire signaled information. The video decoder may then perform an inverse transform to transform the transform coefficient block from the frequency domain back to the sample domain by applying a transform kernel for the NSPT and LFNST to generate the residual block. The video decoder may reconstruct the current block based on the residual block and the prediction block.1616-600W001

[0032] The manner in which the video encoder and the video decoder determine the prediction block may be referred to as the mode (e.g., coding mode) of the current block. Also, there may be a plurality of transform kernels, where each transform kernel defines a manner in which the transform or inverse transform is performed. The video encoder and the video decoder may perform similar operations to determine tire transform kernel, which may be based on the mode (e.g., coding mode) of the current block, and in some examples, based on an angular mode (e.g., an angular intramode) determined for the current block even if the current block is not coded in an angular intramode.

[0033] For blocks coded with certain modes, such as non-directional intra-prediction modes (e.g., non-angular modes) or inter-prediction modes, providing a greater diversity of transform kernels can benefit coding efficiency. That is, there may be benefits in having a diversity of transform kernels from which a transform kernel can be determined. However, with a large number of transform kernels, the memory and signaling overhead increases. Accordingly, there may be a balance between the increasing diversity of transform kernels, while minimizing increase in memory and signaling overhead.

[0034] Accordingly, in video coding, the selection of a transform kernel for a block of video data may depend on the coding mode used for that block. For blocks coded using conventional directional intra-prediction modes (e.g., angular modes), a video coder (e.g., video encoder or video decoder) may select a set of transform kernels by directly mapping the signaled or derived intra-prediction mode to one of several transform set indexes where each transform set index maps to a set of transform kernels. There may¬ be a plurality of sets of transform kernels (e.g., plurality of transform kernel sets), and each set of transform kernels may include a plurality of transform kernels.

[0035] However, video coding standards also include non-directional intra-prediction modes (e.g., non-angular modes), such as matrix intra prediction (M1P), template-based intramode derivation (TIMD), decoder side intramode derivation (DIMD), extrapolation intra prediction (EIP), and spatial geometric partitioning mode (SGPM), as well as intra template matching prediction (IntraTMP) and inter-prediction modes. For blocks coded with these non-directional or inter-prediction modes, a direct mapping from the coding mode is not available.

[0036] To address this, a video coder may perform a separate intra-mode derivation process for the purpose of transform selection. This derivation process, such as a 1616-600W001DIMD-like process, may be applied to a prediction block associated with current block coded in the non-directional mode, or to neighboring reconstructed samples. This DIMD-like process often computes a histogram of gradient (HoG) to identify dominant directions in the prediction block or template. This HoG process may output one or more dominant angular modes, such as a primary intramode (e.g., the direction with the highest gradient) and a secondary' intramode (e.g., the direction with the second-highest gradient).

[0037] The video encoder may signal information indicating whether a primary' transform set (e.g., a first set of transform kernels) is to be used or a secondary transform set (e.g., a second set of transform kernels). In some conventional techniques, the primary transform set is determined based on the primary’ intramode and the secondary- transform set is determined based on the secondary' intramode.

[0038] A challenge arises, however, because these non-directional blocks often lack clear directionality. Although the above example techniques provide for ways to determine two different sets of transform kernels, there may be benefit in increasing the number of transform kernels that are available for transform or inverse transform. However, as noted above, simply increasing transform kernels can lead to additional memory utilization and increased signaling overhead (e.g., more bits are needed to identify which transform kernel to use if there are more transform kernels).

[0039] The techniques of this disclosure provide an improved and more flexible non- separable transform kernel selection, thereby increasing the diversity of available transform kernels and improving coding efficiency, particularly for blocks that do not have clear directionality, while minimizing additional memory utilization and signaling overhead. The techniques group coding modes into a plurality of categories. A video coder, such as a video encoder or a video decoder, may store information indicative of which coding modes are associated with which category of the plurality of categories. Each category is associated with a respective set of transform kernels. For example, a video coder may maintain multiple mapping tables or one common mapping table. In the example of multiple mapping tables, one for each category, each table maps a derived angular mode index to a specific set of transform kernels (also called transform kernel set, transform set, or kernel set). In the example of a common mapping table, the set of transform kernels associated with each angular mode index may be different for different categories. For instance, angular mode X in the common mapping table may map to index Y for a set of transform kernels. For one category, index Y may map to a 1616-600W001particular set of transform kernels, but for another category index Y may map to another particular set of transform kernels.

[0040] These sets of transform kernels may include newly designed, unique kernels, or may share or overlap with kernels from other categories. For instance, the total number of transform sets (e.g., set of transform kernels) may be 35, with 3 candidate kernels per set, but the specific kernel associated with a given set index (e.g., index 8) may be different in a first category’ versus a second category. This structure provides a larger total pool of available transform kernels without necessarily signaling a larger number of transform sets.

[0041] In some examples, the categories are defined based on the type of coding mode. A first category' (e.g,. Category’ 1) may be associated with a set of non-directional intra modes, including a template-based intramode derivation (TIMD) mode, a decoder side intramode derivation (DIMD) mode, an extrapolation intra prediction (EIP) mode, a matrix intra prediction (MIP) mode, and a spatial geometric partitioning mode (SGPM). In this case, the first mode (e.g., the signaled coding mode) of the current block is one of the TIMD mode, the DIMD mode, the EIP mode, the MIP mode, or the SGPM. In some examples, the first category (e.g., Category 2) is associated with an intra template matching prediction (IntraTMP) mode or an inter-coding mode (e.g., an Inter case NSPT). The second category (e.g., Category 0), which serves as the alternative category, may be associated with conventional angular modes, a position dependent prediction (PDP) mode, and a neural network (IntraNN) mode.

[0042] In accordance with one or more examples, a “category switching” method is used to increase the diversity of available transform kernels. A video coder determines that a current block has a size for which a non-separable primary transform (NSPT) is applied. The video coder also determines that the current block is coded in a first mode (e.g., MIP mode) that is associated with a first category (e.g., Category 1). The video coder determines an angular mode for the current block. This angular mode is different than the first mode. For example, the first mode is a non-directional mode (MIP), while the determined angular mode is a directional mode (e.g., a primary- intramode) derived from a HoG process. The video coder determines a transform kernel for NSPT for the current block based on this determined angular mode and the transform kernels associated with a second category' (e.g,. Category’ 0). The second category' is different than the first category. This allow s the block to use a transform kernel from an1616-600W001alternative “bucket” of kernels, increasing the likelihood of finding an efficient transform.

[0043] This selection of the alternative category' may be conditional. In some examples, a video decoder parses information, such as a flag signaled at the coding unit (CU) level, indicating whether a primary transform set or a secondary transform set is to be used for transform kernel selection. In some examples, a video encoder and a video decoder may determine multiple transform sets (e.g., two transform sets). A first transform set is referred to as a primary transform set and a second transform set is referred to as a secondary transform set. The video encoder may signal information indicating whether the primary or secondary transform set is to be selected, and signals information indicating which transform kernel to use within the selected one of the primary or secondary transform set.

[0044] For instance, in some conventional techniques, the video encoder and the video decoder may determine two angular modes for the current block, referred to as primary' intramode and secondary intramode. In conventional techniques, if the information indicates tlie primary transform set is to be used, the video coder may determine the transform kernel based on the primary' angular mode and the transform kernels associated with the first category (the block's original category, which is Category 1 in this example). In conventional techniques, if the information indicates the secondary transform set is to be used, the video coder determines the transform kernel based on the secondary angular mode (e.g., the secondary' intramode) and the transform kernels associated with the first category’ (e.g,, Category 1 in this example),

[0045] In accordance with one or more examples described in this disclosure, if the information indicates the primary transform set is to be used, the video coder may determine the transform kernel based on the primary angular mode (e.g., primary' intramode) and the transform kernels associated with the first category' (the block's original category, which is Category 1 in this example). In contrast with the conventional techniques, if the information indicates the secondary transform set is to be used, the video coder determines the transform kernel based on the primary’ angular mode (e.g., the primary intramode), and not the secondary' intramode, and the transform kernels associated with the second category (e.g., Category 0 in this example).

[0046] For example, if the primary transform set is to be used, the video coder may input the primary intramode (e.g., primary' angular mode) into the mapping table for Category 1 and determine a first set of transform kernels, and then from the first set of 1616-600W001transform kernels, determine the transform kernel to use. If the secondary transform set is to be used, the video coder may input the primary intramode (e.g., primary angular mode) into the mapping table for Category 0 (e.g., the category that does not include the coding mode used to code the current block) and determine a second set of transform kernels, and then from the second set of transform kernels, determine the transform kernel to use. As described, the mapping table for Category 0 and Category 1 may be different mapping tables or may be the same mapping table, but with mapping to different sets of transform kernels for the same index.

[0047] This is different from conventional techniques. In the conventional techniques, if the secondary transform set is to be used, tire video coder may input the secondary intramode (e.g., secondary’ angular mode) into the mapping table for Category' I (e.g., the category that does include the coding mode used to code the current block) and determine a second set of transform kernels, and then from the second set of transform kernels, determine the transform kernel to use.

[0048] The video coder performs different actions based on whether the coder is an encoder or a decoder. A device for decoding video data, such as a device including a video decoder, includes processing circuitry configured to perform these steps. The processing circuitry performs, based on the determined transform kernel, an inverse transform on a transform coefficient block to generate a residual block associated with the current block. The processing circuitry then reconstructs the current block based on the residual block and a corresponding prediction block. A device for encoding video data, such as a device including a video encoder, includes processing circuitry configured to perform a reciprocal process. The processing circuitry determines the transform kernel (e.g., by testing multiple kernels and selecting one based on a rate-distortion cost) and performs, based on the selected transform kernel, a transform on a residual block to generate a transform coefficient block associated with the current block. The processing circuitry then signals information based on the transform coefficient block (e.g., the quantized coefficients and the information indicating which transform kernel or set to use).

[0049] The angular mode used for the transform selection may be determined in various ways. In one example, the video decoder determines the angular mode by parsing information indicating the angular mode, where the mode is explicitly signaled in the bitstream. In another example, the video decoder determines the angular mode based on1616-600W001a histogram of gradient (HoG) computed from neighboring samples of the current block or from a prediction block of the current block.

[0050] In other examples, the transform selection logic is more complex and depends on multiple derived angular modes. For instance, the video coder may derive both a primary intramode and a secondary intramode from the HoG process. In these examples, the determined angular mode comprises a first angular mode (the primary intramode). The processing circuitry is further configured to determine a second angular mode for the current block (the secondary intramode). The processing circuitry then determines a difference value between the first angular mode and the second angular mode, for example, by calculating an absolute difference. This difference value is compared to a predetermined threshold to control the transform selection,

[0051] Several cases may apply based on this comparison. In a first example (Case 1 described further below), the processing circuitry is configured to determine the transform kernel for NSPT for the current block based on the first angular mode and the transform kernels associated with the second category (e.g., Category 0) in response to the difference value being less than the threshold. If the difference value is not less than the threshold, the processing circuitry may determine the transform kernel based on the second angular mode and the transform kernels associated with the first category (e.g., Category 1 or 2).

[0052] In a second example (related to Case 4 described further below), the processing circuitry is configured to determine the transform kernel for NSPT for the current block based on the first angular mode and the transform kernels associated with the second category (e.g., Category 0) in response to the difference value being greater than the threshold.

[0053] Other alternative selection processes may be used. In another example (Case 2 described further below), in response to the absolute difference being less than the threshold, the video coder determines the transform kernel for the secondary transform set from the alternative NSPT category (e.g., Category 0) using the secondary derived transform (the second angular mode, which is the secondary intramode in this example). Otherwise, the video coder may select the transform kernel from the original intramode category (e.g., Category 1 or 2) using the second angular mode.

[0054] In another example (Case 3 described further below), in response to the absolute difference being greater than the threshold, the video coder determines the transform kernel for the secondary transform set from the alternative category (e.g., Category 0) 1616-600W001using the primary derived transform (the first angular mode, which is the primary' intramode in this example). Otherwise, the video coder may select the transform kernel from the original intramode category' using the second angular mode.

[0055] In another example (Case 4), in response to the absolute difference being greater than the threshold, the video coder determines the transform kernel for the secondary transform set from the alternative NSPT category' (e.g., Category 0) using the secondary derived transform (the second angular mode). Otherwise, the video coder may select the transform kernel from the original intramode category using the second angular mode.

[0056] The alternative NSPT category' is not limited to Category 0. In some examples, if the current mode belongs to Category 2, the alternative category may be Category' 0 or Category' 1. If the current mode belongs to Category' 1, the alternative category' may be Category 0 or Category' 2. This alternative category assignment may be predetermined and fixed in the coding standard, or may be signaled at a high level, such as in a slice header, picture parameter set (PPS), or sequence parameter set (SPS), The category’ switching technique may also be extended to modes that belong to Category' 0 (e.g., intraNN), where the alternative sets may come from Category 1 or Category 2 transform sets.

[0057] Other techniques for NSPT / LFNST kernel set selection may be used. In one example, a dominant intra direction (primary' intramode) is derived using a DIMD-like process, and a transform index is computed based on this derived primary intramode. A video encoder may test two or more transforms. These transforms are associated with the same computed transform index, but are selected from two (or more) different kernel sets (e.g., different categories). The video encoder uses a rate-distortion process to select one of the kernel sets and signals an index of the selected kernel set in the bitstream. For example, if three sets are tested, the video encoder may signal an index (e.g., 0, 1, or 2) using unary' binarization with separate contexts for each bin.

[0058] In another example, the selection logic depends on block size. A video coder derives both a primary intramode and a secondary intramode. For transform blocks smaller than a predefined threshold, a video encoder tests a transform index derived from the primary' intramode from two different kernel sets. A signaled index (e.g., 0 or 1) indicates from which kernel set the transform is selected. For transform blocks larger than or equal to the predefined threshold, the video encoder tests two different transform indexes (one derived from the primary intramode, one from the secondary' intramode)1616-600W001from the same kernel set. The signaled index (e.g., 0 or 1) indicates whether the transform index derived from the primary or secondary intra direction is to be used.

[0059] In a related example, the selection logic depends on the difference between the derived modes. A video coder derives the primary and secondary intramodes and determines a difference between them. If the difference is smaller than a predefined threshold, the video coder tests transforms using the index from tire primary intramode from two different kernel sets, and a decision (e.g., a kernel set index) is signaled. If the difference is not smaller than the threshold, the video coder may test two different transform indexes (derived from the primary and secondary intramodes) from the same kernel set, and a decision (e.g., a transform index selection) is signaled.

[0060] FIG. 1 is a block diagram illustrating an example video encoding and decoding system 100 that may perform the techniques of this disclosure. Tire techniques of this disclosure are generally directed to coding (encoding and / or decoding) video data. In general, video data includes any data for processing a video. Thus, video data may include raw, unencoded video, encoded video, decoded (e.g., reconstructed) video, and video metadata, such as signaling data.

[0061] As shown in FIG. 1, system 100 includes a source device 102 that provides encoded video data to be decoded and displayed by a destination device 116, in this example. In particular, source device 102 provides the video data to destination device 116 via a computer-readable medium 110. Source device 102 and destination device 116 may be or 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 device, broadcast receiver devices, or the like. In some cases, source device 102 and destination device 116 may¬ be equipped for wireless communication, and thus may be referred to as wireless communication devices.

[0062] In the example of FIG. 1, source device 102 includes video source 104, memory 106, video encoder 200, and output interface 108, Destination device 116 includes input interface 122, video decoder 300, memory 120, and display device 118. In accordance with this disclosure, video encoder 200 of source device 102 and video decoder 300 of destination device 116 may be configured to apply the techniques for transform coding. Thus, source device 102 represents an example of a video encoding device, while destination device 116 represents an example of a video decoding device.1616-600W001In other examples, a source device and a destination device may include other components or arrangements. For example, source device 102 may receive video data from an external video source, such as an external camera. Likewise, destination device 116 may interface with an external display device, rather than include an integrated display device.

[0063] System 100 as shown in FIG. 1 is merely one example. In general, any digital video encoding and / or decoding device may perform techniques for transform coding. Source device 102 and destination device 116 are merely examples of such coding devices in which source device 102 generates coded video data for transmission to destination device 116. This disclosure refers to a “coding” device as a device that performs coding (encoding and / or decoding) of data. Thus, video encoder 200 and video decoder 300 represent examples of coding devices, in particular, a video encoder and a video decoder, respectively. In some examples, source device 102 and destination device 116 may operate in a substantially symmetrical manner such that each of source device 102 and destination device 116 includes video encoding and decoding components. Hence, system 100 may support one-way or two-way video transmission between source device 102 and destination device 116, e.g., for video streaming, video playback, video broadcasting, or video telephony.

[0064] In general, video source 104 represents a source of video data (i.e., raw, unencoded video data) and provides a sequential series of pictures (also referred to as “frames”) of the video data to video encoder 200, which encodes data for the pictures. Video source 104 of source device 102 may include a video capture device, such as a video camera, a video archive containing previously captured raw video, and / or a video feed interface to receive video from a video content provider. As a further alternative, video source 104 may generate computer graphics-based data as the source video, or a combination of live video, archived video, and computer-generated video. In each case, video encoder 200 encodes the captured, pre-captured, or computer-generated video data. Video encoder 200 may rearrange the pictures from the received order (sometimes referred to as “display order”) into a coding order for coding. Video encoder 200 may generate a bitstream including encoded video data. Source device 102 may then output the encoded video data via output interface 108 onto computer-readable medium 110 for reception and / or retrieval by, e.g., input interface 122 of destination device 116.

[0065] Memory 106 of source device 102 and memory' 120 of destination device 116 represent general purpose memories. In some examples, memories 106, 120 may store 1616-600W001raw video data, e.g., raw video from video source 104 and raw, decoded video data from video decoder 300. Additionally or alternatively, memories 106, 120 may store software instructions executable by, e.g., video encoder 200 and video decoder 300, respectively. Although memory 106 and memory 120 are shown separately' from video encoder 200 and video decoder 300 in this example, it should be understood that video encoder 200 and video decoder 300 may also include internal memories for functionally similar or equivalent purposes. Furthermore, memories 106, 120 may store encoded video data, e.g., output from video encoder 200 and input to video decoder 300. In some examples, portions of memories 106, 120 may be allocated as one or more video buffers, e.g., to store raw, decoded, and / or encoded video data.

[0066] Computer-readable medium 110 may represent any type of medium or device capable of transporting the encoded video data from source device 102 to destination device 116. In one example, computer-readable medium 110 represents a communication medium to enable source device 102 to transmit encoded video data directly to destination device 116 in real-time, e.g., via a radio frequency network or computer-based network. Output interface 108 may modulate a transmission signal including the encoded video data, and 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 a radio frequency (RF) spectrum or one or more phy sical 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 equipment that may be useful to facilitate communication from source device 102 to destination device 116.

[0067] In some examples, source device 102 may output encoded data from output interface 108 to storage device 112. Similarly, destination device 116 may access encoded data from storage device 112 via input interface 122. Storage device 112 may include any of a variety of distributed or locally accessed data storage media such as a hard drive, Blu-ray discs, DVDs, CD-ROMs, flash memory; volatile or non-volatile memory; or any other suitable digital storage media for storing encoded video data.

[0068] In some examples, source device 102 may output encoded video data to file server 114 or another intermediate storage device that may store the encoded video data1616-600W001generated by source device 102. Destination device 116 may access stored video data from file server 114 via streaming or download.

[0069] File server 114 may be any type of server device capable of storing encoded video data and transmitting that encoded video data to the destination device 116. File server 114 may represent a web server (e.g., for a website), a server configured to provide a file transfer protocol sendee (such as File Transfer Protocol (F TP) or 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 (MAS) device. File server 114 may, additionally or alternatively, implement one or more HTTP streaming protocols, such as Dynamic Adaptive Streaming over HTTP (DASH), HTTP Live Streaming (HLS), Real Time Streaming Protocol (RTSP), HTTP Dynamic Streaming, or the like.

[0070] Destination device 116 may access encoded video data from file server 114 through any standard data connection, including an Internet connection. This may include a wireless channel (e.g., a Wi-Fi connection), a wired connection (e.g., digital subscriber line (DSL), cable modem, etc.), or a combination of both that is suitable for accessing encoded video data stored on file server 114. Input interface 122 may be configured to operate according to any one or more of the various protocols discussed above for retrieving or receiving media data from file server 114, or other such protocols for retrieving media data.

[0071] Output interface 108 and input interface 122 may represent wireless transmitters / receivers, modems, wired networking components (e.g., Ethernet cards), wireless communication components that operate according to any of a variety' of IEEE 802.11 standards, or other physical components. In examples where output interface 108 and input interface 122 include wireless components, output interface 108 and input interface 122 may be configured to transfer data, such as encoded video data, according to a cellular communication standard, such as 4G, 4G-LTE (Long-Term Evolution), LTE Advanced, 5G, or the like. In some examples where output interface 108 includes a wireless transmitter, output interface 108 and input interface 122 may be configured to transfer data, such as encoded video data, according to other wireless standards, such as an IEEE 802.11 specification, an IEEE 802.15 specification (e.g., ZigBee™), a Bluetooth™ standard, or the like. In some examples, source device 102 and / or destination device 116 may include respective system-on-a-chip (SoC) devices. For 1616-600W001example, source device 102 may include an SoC device to perform the functionality attributed to video encoder 200 and / or output interface 108, and destination device 116 may include an SoC device to perform the functionality attributed to video decoder 300 and / or input interface 122.

[0072] Tire techniques of this disclosure may be applied to video coding in support of any of a variety of multimedia applications, such as over-the-air television broadcasts, cable television transmissions, satellite television transmissions, Internet streaming video transmissions, such as dynamic adaptive streaming over HTTP (DASH), digital video that is encoded onto a data storage medium, decoding of digital video stored on a data storage medium, or other applications.

[0073] Input interface 122 of destination device 116 receives an encoded video bitstream from computer-readable medium 110 (e.g., a communication medium, storage device 112, file server 114, or the like). Hie encoded video bitstream may include signaling information defined by video encoder 200, which is also used by video decoder 300, such as syntax elements having values that describe characteristics and / or processing of video blocks or other coded units (e.g., slices, pictures, groups of pictures, sequences, or the like). Display device 118 displays decoded pictures of the decoded video data to a user. 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.

[0074] Although not shown in FIG. 1, in some examples, video encoder 200 and video decoder 300 may each be integrated with an audio encoder and / or audio decoder (e.g,, audio codec), and may include appropriate MUX-DEMUX units, or other hardware and / or software, to handle multiplexed streams including both audio and video in a common data stream. Example audio codecs may include AAC, AC-3, AC-4, ALAC, ALS, AMBE, AMR, AMR-WB (G.722.2), AMR-WB+, aptx (various versions), ATRAC, BroadVoice (BV16, BV32), CELT, Enhanced AC-3 (E-AC-3), EVS, FLAG, G.711, G.722, G.722.1, G.722.2 (AMR-WB). G.723.1, G.726, G.728, G.729, G.729.1, GSM-FR, HE-AAC, iLBC, iSAC, LA Lyra, Monkey's Audio, MP1, MP2 (MPEG-1, 2 Audio Layer II), MP3, Musepack, Nellymoser Asao, OptimFROG, Opus, Sac, Satin, SBC, SILK, Siren 7, Speex, SVOPC, True Audio (TTA), TwinVQ, USAC, Vorbis (Ogg), WavPack, and Windows Media Audio,

[0075] Video encoder 200 and video decoder 300 each may be implemented as any of a variety of suitable encoder and / or decoder circuitry that includes a processing system, 1616-600W001such 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 combinations thereof. When the techniques are implemented partially in software, a device may store instructions for the software in a suitable, non-transitory computer-readable medium and execute the instructions in hardware using one or more processors to perform the techniques of this disclosure. Each of video encoder 200 and video decoder 300 may be included in one or more encoders or decoders, either of which may be integrated as part of a combined encoder / decoder (CODEC) in a respective device. A device including video encoder 200 and / or video decoder 300 may implement video encoder 200 and / or video decoder 300 in processing circuitry such as an integrated circuit and / or a microprocessor. Such a device may be a wireless communication device, such as a cellular telephone, or any other type of device described herein.

[0076] Video encoder 200 and video decoder 300 may operate according to a video coding standard, such as ITU-T H.265, also referred to as High Efficiency Video Coding (HE VC) or extensions thereto, such as the multi -view and / or scalable video coding extensions. Alternatively, video encoder 200 and video decoder 300 may operate according to other proprietary’ or industry' standards, such as ITU-T H.266, also referred to as Versatile Video Coding (VVC). In other examples, video encoder 200 and video decoder 300 may operate according to a proprietary video codec / format, such as AOMedia Video 1 (AV1), extensions of AV1, and / or successor versions of AV1 (e.g., AV2). In other examples, video encoder 200 and video decoder 300 may operate according to other proprietary formats or industry standards. The techniques of this disclosure, however, are not limited to any particular coding standard or format. In general, video encoder 200 and video decoder 300 may be configured to perform the techniques of this disclosure in conjunction with any video coding techniques that use transform coding.

[0077] In general, video encoder 200 and video decoder 300 may perform block-based coding of pictures. Tire term “block” generally refers to a structure including 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. In general, video encoder 200 and video decoder 300 may code video data represented in a YUV (e.g., Y, Cb, Cr) format. That is, rather than coding red, green, and blue (RGB) data for samples of a picture, video encoder 200 1616-600W001and video decoder 300 may code luminance and chrominance components, where the chrominance components may include both red hue and blue hue chrominance components. In some examples, video encoder 200 converts received RGB formatted data to a YUV representation prior to encoding, and video decoder 300 converts the YUV representation to the RGB format. Alternatively, pre- and post-processing units (not shown) may perform these conversions.

[0078] This disclosure may generally refer to coding (e.g,, encoding and decoding) of pictures to include tire process of encoding or decoding data of the picture. Similarly, this disclosure may refer to coding of blocks of a picture to include the process of encoding or decoding data for the blocks, e.g., prediction and / or residual coding. An encoded video bitstream general ly includes a series of values for syntax elements representative of coding decisions (e.g., coding modes) and partitioning of pictures into blocks. Thus, references to coding a picture or a block should generally be understood as coding values for syntax elements forming the picture or block.

[0079] 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) partitions a coding tree unit (CTU) into CU s according to a quadtree structure. That is, the video coder partitions CTUs and CUs into four equal, nonoverlapping squares, and each node of the quadtree has either zero or four child nodes. Nodes without child nodes may be referred to as “leaf nodes,” and CUs of such leaf nodes may include one or more PUs and / or one or more TUs. lire video coder may further partition PUs and TUs, For example, in HEV C, a residual quadtree (RQT) represents partitioning of TUs. In HEVC, PUs represent inter-prediction data, while TUs represent residual data. CUs that are intra-predicted include intra-prediction information, such as an intra-mode indication.

[0080] 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) partitions a picture into a plurality of CTUs. Video encoder 200 may partition a CTU according to a tree structure, such as a quadtree -binary tree (QTBT) structure or Multi-Type Tree (MTT) structure. The QTBT structure removes the concepts of multiple partition types, such as the separation between CUs, PUs, and TUs of HEVC. A QTBT structure includes two levels: a first level partitioned according to quadtree partitioning, and a second level partitioned according to binary' tree1616-600W001partitioning. A root node of the QTBT structure corresponds to a CTU. Leaf nodes of the binary trees correspond to CUs.

[0081] In an MTT partitioning structure, blocks may be partitioned using a quadtree (QT) partition, a binary tree (BT) partition, and one or more types of triple tree (TT) (also called ternary' tree (TT)) partitions. A triple or ternary' tree partition is a partition where a block is split into three sub-blocks. In some examples, a triple or ternary tree partition divides a block into three sub-blocks without dividing the original block through the center. The partitioning types in MTT (e.g., QT, BT, and TT), may be symmetrical or asymmetrical.

[0082] When operating according to the AV1 codec, video encoder 200 and video decoder 300 may be configured to code video data in blocks. In AVI, the largest coding block that can be processed is called a superblock. In AV1, a superblock can be either 128x128 luma samples or 64x64 luma samples. However, in successor video coding formats (e.g., AV2), a superblock may be defined by different (e.g., larger) luma sample sizes. In some examples, a superblock is the top level of a block quadtree. Video encoder 200 may further partition a superblock into smaller coding blocks. Video encoder 200 may partition a superblock and other coding blocks into smaller blocks using square or non-square partitioning. Non-square blocks may include N / 2xN, NxN / 2, N / 4xN, and NxN / 4 blocks. Video encoder 200 and video decoder 300 may perform separate prediction and transform processes on each of the coding blocks.

[0083] AV1 also defines a tile of video data. A tile is a rectangular array of superblocks that may' be coded independently' of other tiles. That is, video encoder 200 and video decoder 300 may encode and decode, respectively, coding blocks within a tile without using video data from other tiles. However, video encoder 200 and video decoder 300 may perform filtering across tile boundaries. Tiles may be uniform or non-uniform in size. Tile-based coding may enable parallel processing and / or multi-threading for encoder and decoder implementations.

[0084] In some examples, video encoder 200 and video decoder 300 may use a single QTBT or MTT structure to represent each of the luminance and chrominance components, while in other examples, video encoder 200 and 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 respective chrominance components).1616-600W001

[0085] Video encoder 200 and video decoder 300 may be configured to use quadtree partitioning, QTBT partitioning, MTT partitioning, superblock partitioning, or other partitioning structures.

[0086] In some examples, a CTU includes a coding tree block (CTB) of luma samples, two corresponding CTBs of chroma samples of a picture that has three sample arrays, or a CTB of samples of a monochrome picture or a picture that is coded using three separate color planes and syntax structures used to code the samples. A CTB may be an NxN block of samples for some value of N such that the division of a component into CTBs is a partitioning. A component is an array or single sample from one of the three arrays (luma and two chroma) that compose a picture in 4:2:0, 4:2:2, or 4:4:4 color format or the array or a single sample of the array that compose a picture in monochrome format. In some examples, a coding block is an MxN block of samples for some values of M and N such that a division of a CTB into coding blocks is a partitioning.

[0087] The blocks (e.g., CTUs or CUs) may be grouped in various ways in a picture. As one example, a brick may refer to a rectangular region of CTU rows within a particular tile in a picture. A tile may be a rectangular region of CTUs within a particular tile column and a particular tile row in a 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 syntax elements (e.g., such as in a picture parameter set). A tile row refers to a rectangular region of CTUs having a height specified by syntax elements (e.g., such as in a picture parameter set) and a width equal to the width of the picture,

[0088] In some examples, a tile may be partitioned into multiple bricks, each of which may include one or more CTU rows within the tile. A tile that is not partitioned into multiple bricks may also be referred to as a brick. However, a brick that is a true subset of a tile may not be referred to as a tile. The bricks in a picture may also be arranged in a slice. A slice may be an integer number of bricks of a picture that may be exclusi vely contained in a single network abstraction layer (N AL) unit. In some examples, a slice includes either a number of complete tiles or only a consecutive sequence of complete bricks of one tile.

[0089] This disclosure may use “NxN” and “N by N” interchangeably to refer to the sample dimensions of a block (such as a CU or other video block) in terms of vertical and horizontal dimensions, e.g., 16x16 samples or 16 by 16 samples. In general, a 16x16 CU will have 16 samples in a vertical direction (y = 16) and 16 samples in a 1616-600W001horizontal direction (x = 16). Likewise, an NxN CU generally has N samples in a vertical direction and N samples in a horizontal direction, where N represents a nonnegative integer value. The samples in a CU may be arranged in rows and columns. Moreover, CUs need not necessarily have the same number of samples in the horizontal direction as in the vertical direction. For example, CUs may include NxM samples, where M is not necessarily equal to N.

[0090] Video encoder 200 encodes video data for CUs representing prediction and / or residual information, and other information. The prediction information indicates how the CU is to be predicted in order to form a prediction block for the CU. The residual information generally represents sample-by-sample differences between samples of the CU prior to encoding and the prediction block.

[0091] 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 the CU from data of a previously coded picture, whereas intra-prediction generally refers to predicting the CU from previously coded data of the same picture. To perform inter-prediction, video encoder 200 may generate the prediction block using one or more motion vectors. Video encoder 200 may generally perform a motion search to identify a reference block that closely matches the CU, e.g,, in terms of differences between the CU and the reference block. Video encoder 200 may calculate a difference metric using a 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 a reference block closely matches the current CU. In some examples, video encoder 200 may predict the current CU using uni-directional prediction or bi-directional prediction.

[0092] Some examples ofVVC also provide an affine motion compensation mode, which may be considered an inter-prediction mode. In affine motion compensation mode, video encoder 200 may determine two or more motion vectors that represent non- translational motion, such as zoom in or out, rotation, perspective motion, or other irregular motion types.

[0093] To perform intra-prediction, video encoder 200 may select an intra-prediction mode to generate the prediction block. Some examples ofVVC provide sixty-seven intra-prediction modes, including various directional modes, as well as planar mode and DC mode. In general, video encoder 200 selects an intra-prediction mode that describes neighboring samples to a current block (e.g., a block of a CU) from which to predict 1616-600W001samples of the current block. Such samples may generally be above, above and to the left, or to the left of the current block in the same picture as the c urrent block, assuming video encoder 200 codes CTUs and CUs in raster scan order (left to right, top to bottom).

[0094] Video encoder 200 encodes data representing the prediction mode for a current block. For example, for inter-prediction modes, video encoder 200 may encode data representing which of the various available inter-prediction modes is used, as well as motion information for the corresponding mode. For uni-directional or bi-directional inter-prediction, for example, video encoder 200 may encode motion vectors using advanced motion vector prediction (AMVP) or merge mode. Video encoder 200 may use similar modes to encode motion vectors for affine motion compensation mode.

[0095] AVI includes two general techniques for encoding and decoding a coding block 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 AVI, when predicting blocks of a 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 video data. For most intra prediction modes, video encoder 200 encodes blocks of a current frame based on the difference between sample values in the current block and predicted values generated from reference samples in the same frame. Video encoder 200 determines predicted values generated from the reference samples based on the intra prediction mode.

[0096] Following prediction, such as intra-prediction or inter-prediction of a block, video encoder 200 may calculate residual data for the block. The residual data, such as a residual block, represents sample by sample differences between the block and a prediction block for the block, formed using the corresponding prediction mode. Video encoder 200 may apply one or more transforms to the residual block, to produce transformed data in a transform domain instead of 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 residual video data.Additionally, video encoder 200 may apply a secondary transform following the first transform, such as a mode-dependent non-separable secondary transform (MDNSST), a signal dependent transform, a Karhunen-Loeve transform (KLT), or the like. Video encoder 200 produces transform coefficients following application of the one or more transforms.1616-600W001

[0097] As noted above, following any transforms to produce transform coefficients, video encoder 200 may perform quantization of the transform coefficients.Quantization generally refers to a process in which transform coefficients are quantized to possibly reduce the amount of data used to represent the transform coefficients, providing further compression. By performing the quantization process, video encoder 200 may reduce the bit depth associated with some or all of the transform coefficients. For example, video encoder 200 may round an n-bit value down 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 bitwise right-shift of the value to be quantized.

[0098] Following quantization, video encoder 200 may scan the transform coefficients, producing a one-dimensional vector from the two-dimensional matrix including tire quantized transform coefficients. The scan may be designed to place higher energy (and therefore lower frequency) transform coefficients at the front of the vector and to place lower energy (and therefore higher frequency) transform coefficients at the back of the vector. In some examples, video encoder 200 may utilize a predefined scan order to scan the quantized transform coefficients to produce a serialized vector, and then entropy encode the quantized transform coefficients of the vector. In other examples, video encoder 200 may perform an adaptive scan. After scanning the quantized transform coefficients to form the one-dimensional vector, video encoder 200 may entropy encode the one-dimensional vector e.g., according to context-adaptive binary arithmetic coding (CABAC). Video encoder 200 may also entropy encode values for syntax elements describing metadata associated with the encoded video data for use by video decoder 300 in decoding the video data.

[0099] To perform CABAC, video encoder 200 may assign a context within a context model to a symbol to be transmitted. The context may relate to, for example, whether neighboring values of the symbol are zero-valued or not. The probability determination may be based on a context assigned to the symbol.

[0100] Video encoder 200 may further generate syntax data, such as block -based syntax data, picture-based syntax data, and sequence -based syntax data, to video decoder 300, e.g., in a picture header, a block header, a slice header, or other syntax data, such as a sequence parameter set (SPS), picture parameter set (PPS), or video parameter set (VPS). Video decoder 300 may likewise decode such syntax data to determine how to decode corresponding video data.1616-600W001

[0101] In this manner, video encoder 200 may generate a bitstream including encoded video data, e.g., syntax elements describing partitioning of a picture into blocks (e.g., CUs) and prediction and / or residual information for the blocks. Ultimately, video decoder 300 may receive the bitstream and decode the encoded video data,

[0102] In general, video decoder 300 performs a reciprocal process to that performed by video encoder 200 to decode the encoded video data of the bitstream. For example, video decoder 300 may decode values for syntax elements of the bitstream using CABAC in a manner substantially similar to, albeit reciprocal to, the CABAC encoding process of video encoder 200. The syntax elements may define partitioning information for partitioning of a picture into CTUs, and partitioning of each CTU according to a corresponding partition structure, such as a QTBT structure, to define CUs of the CTU. The syntax elements may further define prediction and residual information for blocks (e.g., CUs) of video data.

[0103] The residual information may be represented by, for example, quantized transform coefficients. Video decoder 300 may inverse quantize and inverse transform the quantized transform coefficients of a block to reproduce a residual block for the block. Video decoder 300 uses a signaled prediction mode (intra- or inter-prediction) and related prediction information (e.g., motion information for inter-prediction) to form a prediction block for the block. Video decoder 300 may then combine the prediction block and the residual block (on a sample-by-sample basis) to reproduce the original block. Video decoder 300 may perform additional processing, such as performing a deblocking process to reduce visual artifacts along boundaries of the block.

[0104] Any of the video encoding or video decoding processes described above may be performed using a neural network (NN). Additionally or alternatively, a neural network may be trained to efficiently compress video data without necessarily separately performing prediction and residual coding. Studies have shown that embedding neural networks into the hybrid video coding framework of video encoder 200 and video decoder 300 can improve compression efficiency. Neural networks may be used for intra prediction and inter prediction to improve the prediction efficiency. NN-based inloop filtering and / or post-filtering have also performed well in heuristic testing.

[0105] For example, video encoder 200 and video decoder may use one or more NN- based filters for existing filters, such as deblocking filters, sample adaptive offset (SAO), and / or adaptive loop filtering (ALF). NN-based filters can also be applied exclusively, where NN-based filters are designed to replace all of the existing1616-600W001filters. Additionally or alternatively, NN-based filters may be designed to supplement, enhance, or replace any or all of the other filters.

[0106] In some examples, an NN-based filter may be a convolutional neural network (CNN)-based filter with multiple layers. An NN-based filtering process may take reconstructed samples as inputs, and may add the intermediate outputs back to the inputs to refine the input samples. The NN-based filter may use all color components (e.g., Y, U, and V, or Y, Cb, and Cr) as inputs 172 to exploit cross-component correlations. Different color components may share tire same filters (including network structure and model parameters) or each component may have its own specific filters.

[0107] The filtering process can also be generalized as follows:= R(i,j) + NN_filter_residual_output(R') Here, R(i, j) represents a reconstructed sample at position (i, j) in the picture, R’(i, j) represents the filtered version of the reconstructed sample, andNN_filter_residual_output(R) represents the intermediate samples discussed above that are calculated by the NN filter. The model structure and model parameters of NN-based filter(s) can be pre-defined and be stored at video encoder 200 and video decoder 300. The filters can also be signaled in the bitstream.

[0108] In some examples, an NN-based filter may include a series of feature extraction layers, followed by an output convolution. The feature extraction layers may include a 3x3 convolution (conv) layer followed by a parametric rectified linear unit (PReLU) layer. The convolutional layer applies a convolution operation to the input data, which involves a filter or kernel processing the input data (e.g., the reconstruction samples) in a sliding window fashion and computing dot products at each position. Tire convolution operation essentially captures local patterns within the input data. For example, in the context of image processing, these patterns could be edges, textures, or other visual features. The filter or kernel is a small matrix of weights that gets updated during the training process. By sliding this filter across the input data (or feature map from a previous layer) and computing the dot product at each position, the convolutional layer creates a feature map that encodes spatial hierarchies and patterns detected in the input. The output of a convolutional layer is a set of feature maps, each corresponding to one filter, capturing different aspects of the input data. This layer helps the neural network to learn increasingly complex and abstract features as the data passes through deeper layers of the network.1616-600W001

[0109] The PReLU layer is an activation function used in neural networks, and is a variant of the ReLU (Rectified Linear Unit) activation function. As described above, the convolution layer outputs feature maps, each corresponding to one filter, representing detected features in the input. Following the convolution layer, the PReLU layer applies the PReLU activation function to each element of the feature maps produced by the convolution layer. For positive values, the PReLU layer acts like a standard ReLU, passing the value through. For negative values, instead of setting them to zero (e.g., as ReLU does), the PReLU layer allows a small, linear, negative output. This keeps neurons of the N active and maintains the gradient flow, which can be beneficial for learning in deep networks.

[0110] When NN-based filtering is applied in video coding, the whole video signal (pixel data) may be split into multiple processing units (e.g., 2D blocks), and each processing unit can be processed separately or be combined with other information associated with this block of pixels. For example, a processing unit may be a frame, a slice / tile, a CTU, or any pre-defined or signaled shapes and sizes. Typically, NN-based filtering is performed on reconstructed blocks of video data. Here, reconstructed blocks and samples may refer to both decoded blocks produced by video decoder 300, as well as blocks reconstructed in a reconstruction loop of video encoder 200.

[0111] To further improve the performance of NN-based filtering, different types of input data can be processed jointly to produce the filtered output. Input data may include, but is not limited to, reconstruction pixels / samples, prediction pixels / samples, pixels / samples after the loop filter(s), partitioning structure information, deblocking parameters (e.g., boundary strength (BS)), quantization parameter (QP) values, slice or picture types, or a filter's applicability or coding modes map. Input data can be provided at different granularities. Luma reconstruction and prediction samples may be provided at the original resolution, whereas chroma samples may be provided at lower resolution, e.g. for 4:2:0 representation, or can be up-sampled to the Luma resolution to achieve per-pixel representation. Similarly, QP, BS, partitioning, or coding mode information can be provided at lower resolution, including cases with a single value per frame, slice or processing block (e.g. QP). In other examples, QP, BS, partitioning, or coding mode information can be expanded (e.g., replicated) to achieve per-pixel / sample representation.

[0112] To further improve the performance of NN-based filtering, multi-mode solutions can be used. For example, for each processing unit, video encoder 200 may select a 1616-600W001mode from a set of modes based on rate-distortion optimization and signal the selected mode in the bit-stream. The different modes may include different NN models, different values that may be used as the input information of the NN models, etc. In one example, video encoder 200 and video decoder 300 may use an NN-based filtering solution with multiple modes based on a single NN model by using different QP values as input to the NN model for different modes.

[0113] This disclosure may generally refer to “signaling” certain information, such as syntax elements. The term “signaling” may generally refer to tire communication of values for syntax elements and / or other data used to decode encoded video data. That is, video encoder 200 may signal values for syntax elements in the bitstream. In general, signaling refers to generating a value in the bitstream. As noted above, source device 102 may transport the bitstream to destination device 116 substantially in real time, or not in real time, such as might occur when storing syntax elements to storage device 112 for later retrieval by destination device 116.

[0114] As described above, video encoder 200 may perform a transform and video decoder 300 may perform an inverse transform. In some examples, video encoder 200 and video decoder 300 may utilize non-separable transforms. As one example, video encoder 200 and video decoder 300 may apply a separable transform (e.g,, DCT-II) as a primary transform and a non-separable transform as a second transform, such as a low- frequency non-separable transform (LFNST). In some examples, video encoder 200 and video decoder 300 may apply a non-separable transform as the primary transform and not use a secondary transform. Examples of the non-separable primary transform (NSPT) are described in more detail below.

[0115] In both LFNST and NSPT, there may be a plurality of transform kernels (e.g., different ways in which the transform is performed). Video encoder 200 and video decoder 300 may utilize various example techniques to determine which transform kernel to use.

[0116] In accordance with examples described in this disclosure, for blocks of video data, such as those having a size for which non-separable primary’ transform (NSPT) is applied, the selection of a transform kernel may be based on the coding mode of the block, in some examples, video encoder 200 and video decoder 300 may store information indicative of which coding modes are associated with which category of a plurality of categories, and each category is associated with a respective set of transform kernels. For instance, a first category may be associated with modes such as template- 1616-600W001based intramode derivation (TIMD) mode, decoder side intramode derivation (DIMD) mode, extrapolation intra prediction (EIP) mode, matrix intra prediction (MIP) mode, and spatial geometric partitioning mode (SGPM). Another category may be associated with intra template matching prediction (IntraTMP) mode or inter-coding mode.

[0117] Tire techniques of this disclosure provide for improved transform kernel selection. Video encoder 200 and video decoder 300 may determine that the current block is coded in a first mode associated with a first category of the plurality of categories. The video coder may also determine an angular mode for the current block, the angular mode being different than the first mode. For example, the first mode may be a non-directional mode (like MIP), and the angular mode may be a primary intramode derived using a histogram of gradient (HoG) process. Tire angular mode may be a secondary' intramode derived using HoG process in other examples.

[0118] Video encoder 200 and video decoder 300 may then determine a transform kernel for NSPT for the current block based on the angular mode and transform kernels associated with a second category, the second category' being different than the first category. For example, the second category may be associated with angular modes and a neural network (IntraNN) mode.

[0119] Unlike standard intra-prediction modes (such as Planar, DC, or Angular) that use linear extrapolation to predict pixel values from neighboring samples, IntraNN mode may employ architectures like Fully Connected Networks (FCN) or Convolutional Neural N etworks (CNN) to analyze the complex non-linear spatial correlations in boundary pixels. By feeding reconstructed neighboring pixels into a pre-trained network, the IntraNN mode generates a predicted block that can better represent complex textures and edges that traditional modes fail to capture, significantly reducing the residual error and bitrate, albeit at the cost of higher computational complexity for both video encoder 200 and video decoder 300.

[0120] This determination of the transform kernel from the second category may be conditional. In one example, video decoder 300 performs this selection by parsing information indicating whether a primary transform set or a secondary transform set is to be used for transform kernel selection. The transform kernel is determined based on the angular mode and the transform kernels associated with the second category based on the information indicating that the secondary transform set is to be used. In other examples, the angular mode comprises a first angular mode, and the video coder may also determine a second angular mode for the current block. The video coder may 1616-600W001determine a difference value between the first angular mode and the second angular mode, and the selection of the transform kernel from the first category or the second category may be based on whether the difference value is less than a threshold or greater than a threshold.

[0121] Tire following describes Secondary Transformation: Low-Frequency Non- Separable Transform (LFNST) Extension with Large Kernel. The LFNST design in VVC may be extended as follows:® The number of LFNST sets (5) and candidates (C) are extended to, S'=35 and C.and the LFNST set (lfnstTrSetIdx) for a given intramode (predModeIntra) is derived according to the following formula:o For predModeIntra < 2, lfnstTrSetIdx is equal to 2o lfnstTrSetIdx ==:predModeIntra, for predModeIntra in [0,34] o lfnstTrSetIdx = 68 - predModeIntra, for predModeIntra in [35,66]• Three different kernels, LFNST4, LFNST8, and LFNST16, are defined to indicate LFNST kernel sets, which are applied to 4xN / Nx4 (N>4), 8xN / Nx8 (N>8), and MxN (M, N>16), respectively.

[0122] The kernel dimensions are specified by:(LFNST4, LFNST8*, LFNST16*) - (16x16, 32x64, 32x96)

[0123] The forward LFNST is applied to the top-left low frequency region, which is called Region-Of-Interest (ROI). When LFNST is applied, primary-transformed coefficients that exist in the region other than ROI are zeroed out, which is not changed from the VVC standard.

[0124] FIG. 6 is a conceptual diagram illustrating a region of interest (ROI) 600 for LFNST16. The ROI includes six 4x4 sub-blocks, which are consecutive in scan order. Since the number of input samples is 96, the transform matrix for forward LFNST 16 can be Rx96. R is chosen to be 32 in this contribution, 32 coefficients (two 4x4 sub¬ blocks) are generated from forward LFNST16 accordingly, which are placed following coefficien t scan order. That is, the 96 coefficien ts of the six 4x4 sub-blocks are multiplied by the 32x96 matrix to output the 32 coefficients.

[0125] FIG. 7 is a conceptual diagram illustrating a ROI 700 for LFNST8. In other words, a ROI 700 for LFNST8 is shown in FIG. 7. Tire forward LFNST8 matrix can be Rx64 and R is chosen to be 32. The generated coefficients are located in the same manner as with LFNST16.1616-600W001

[0126] FIG. 8 is a table 800 illustrating a mapping of intra prediction modes to a low- frequency non-separable transform (LFNST) set index. The mapping from intra prediction modes to these LFNST kernel sets is shown in the table of FIG. 8. For example, for the intra prediction mode -14, an LFNST kernel set with LFNST set index 2 is used.

[0127] A similar mapping concept may be applied for non-separable primary transform (NSPT), but where the mapping is dependent on a category associated with the coding mode of the block. For example, video encoder 200 or video decoder 300 may maintain multiple mapping tables or one common mapping table. In the example of multiple mapping tables, one for each category, each table maps a derived angular mode index to a specific set of transform kernels (also called transform kernel set, transform set, or kernel set). In the example of a common mapping table, the set of transform kernels associated with each angular mode index may be different for different categories.

[0128] For instance, angular mode X in the common mapping table may map to index Y for a set of transform kernels. For one category, index Y may map to a particular set of transform kernels, but for another category, index Y may map to another particular set of transform kernels. These sets of transform kernels may include newly designed, unique kernels, or may share or overlap with kernels from other categories. For instance, the total number of transform sets (e.g., set of transform kernels) may be 35, with 3 candidate kernels per set, but the specific kernel associated with a given set index (e.g., index 8) may be different in a first category versus a second category. This structure provides a larger total pool of available transform kernels without necessarily signaling a larger number of transform sets.

[0129] For blocks using matrix intra prediction (MIP) or intra template matching prediction (IntraTMP) prediction, the LFNST set index is derived as follows. Decoder side intramode derivation (DIMD) is used to derive the intra prediction mode of the current block based on the MIP or IntraTMP predicted samples. For MIP, this is done before upsampling. Specifically, a horizontal gradient and a vertical gradient are calculated for each predicted sample to build a histogram of gradients (HoG), as shown in FIG. 9.

[0130] In other words, for each sample of the prediction block, a horizontal gradient value is calculated that indicates a rate of change at the sample of sample values in the horizontal direction and a vertical gradient value is calculated that indicates a rate of change at the sample of sample values in the vertical direction. A directional intramode 1616-600W001(e.g., angular mode) for the sample is determined based on the horizontal gradient value and the vertical gradient value is then determined. The HoG represents counts of directional intra prediction modes for the samples in the prediction block. The intra prediction mode with the largest histogram amplitude values is then used to determine the LFNST transform set and an LFNST Transpose flag. For example, video encoder 200 and video decoder 300 may use transposed kernels to generate predictions for a block of dimensions WxH with directional intra prediction mode 18 (horizontal mode) and a HxW block with directional intra prediction mode 50 (vertical mode) because of symmetry of these directional intra prediction modes. So, kernels for up to intra prediction mode 34 are defined, and a block of WxH with mode m (with m > 34) will use the transposed kernel of HxW with mode (68 - m). The transposition of the kernel may be indicated by an LFNST transpose flag (indicating kernel transpose is needed).

[0131] FIG. 9 is a conceptual diagram illustrating example Mapping from Intra Prediction (MIP) prediction samples to build a histogram of gradient (HoG), In the example of FIG. 9, video encoder 200 or video decoder 300 may first perform matrix¬ vector multiplication using matrix-vector multiplication unit 900 to generate an initial block 902. Video encoder 200 or video decoder 300 may then perform MIP prediction upsampling using MIP prediction upsampling unit 904 to generate an upsampled block 906. Additionally, video encoder 200 or video decoder 300 may perform DIMD derivation using DIMD derivation unit 908 to generate a HoG 910.

[0132] Non-separable primary transform (NSPT) for intra coding is now discussed, The separable DCT-II plus LFNST transform combinations are replaced with NSPT for the block shapes 4x4, 4x8, 8x4 and 8x8, 4x16, 16x4, 8x16 and 16x8. The affected block sizes are summarized in FIG. 10.

[0133] FIG. 10 is a conceptual diagram illustrating an overview of non-separable primary transforms (NSPTs) among existing LFNSTs. In other words, FIG. 10 shows an example of sets of kernels 1000 for NSPT. Sets of kernels 1000 include three groups. A first group corresponds to transform block sizes of 4xN and Nx4, where N is greater than or equal to 4 (e.g., 4x4, 4x8, 8x4, 4x16, and 16x4). A second group corresponds to transform block sizes of 8xN and Nx8, where N is greater than or equal to 8 (e.g., 8x8, 8x16, and 16x8). A third group corresponds to transform block sizes of MxN, where M and N are greater than or equal to 16.

[0134] In FIG. 10, a kernel 1002 (NSPT 4x4) corresponds to transform block size 4x4, a kernel 1004 (NSPT 4x8) corresponds to transform block size 4x8, a kernel 1006 (NSPT 1616-600W0018x4) corresponds to transform block size 8x4, a kernel 1008 (NSPT 4x16) corresponds to transform block size 4x16, a kernel 1010 (NSPT 16x4) corresponds to transform block size 16x4, and a kernel 1012 (DCT-II+LFNST4, i.e., DCT-II followed by LFNST- 4) corresponds to other transform block sizes of 4xN / Nx4 with N greater than or equal to 4.

[0135] A kernel 1014 (NSPT 8x8) corresponds to transform block size 8x8, a kernel 1016 (NSPT 8x16) corresponds to transform block size 8x16, a kernel 1018 (NSPT 16x8) corresponds to transform block size 16x8, and a kernel 1020 (DCT-II+LFNST8) corresponds to other transform block sizes of 8xN / Nx8 with N greater than or equal to 8. A kernel 1022 corresponds to block sizes MxN where M and N are greater than or equal to 16.

[0136] In one example, the NSPTs include 35 sets (e.g., kernel sets or transform kernel sets) and 3 candidates (similar to the current LFNST). The kernels (e.g., transform kernels) of NSPTs have the following shapes:- NSPT4x4: 16x16- NSPT4x8 / NSPT8x4: 32x20- NSPT8x8: 64x32- NSPT4x16 / NSPT16x4: 64x24- NSPT8x16 / NSPT16x8: 128x40- NSPT4x32 / NSPT32x4: 128x20- NSPT8x32 / NSPT32x8: 256x24

[0137] Therefore, 12, 32, 40 and 88 coefficients are zeroed-out using NSPT4x8 / NSPT8x4, NSPT8x8, NSPT4x16 / NSPT16x4 and NSPT8x16 / NSPT16x8 respectively. For NSPT4x32 / NSPT32x4 and NSPT8x32 / NSPT32x8, remaining 108 and 232 positions in each transform block are zeroed-out, respectively,

[0138] LFNST and NSPT for inter coding are now discussed. For an inter coded block, an intra prediction mode is first derived according to the inter prediction block with a DIMD-like process applied to the prediction. Then the derived intra prediction mode is used to select an LFNST / NSPT transform set and the transform is processed with the selected LFNST / NSPT kernel, like in the intra coding process.

[0139] The signaling of inter LFNST / NSPT index is different from that of intra LFNST / NSPT index. As noted above, the NSPTs may include 35 kernel sets and 3 candidates. Each of the candidates is an individual kernel (e.g., individual transform kernel). The LFNST / NSPT index indicates which candidate is used. 0 indicates no 1616-600W001LFNST / NSPT, and {1,2,3} indicates which LFNST / NSPT candidate is to be used. So, this is not a transform set index but just to indicate which candidate in LFNST / NSPT set. The intra LFNST / NSPT index binarization employs two context coded bins for each symbol, while truncated unary code with different context models for inter LFNST / NSPT index coding is used. LFNST / NSPT index signaling is prohibited in the case of sub-biock transform (SBT), of which index value is inferred as 0.

[0140] Encoding fast algorithm like inter multiple transform selection (MTS) is applied to reduce the encoding time. The last non-zero position is checked for signaling of LFNST / NSPT index using luma component only. For low-delay coding configuration intra-LFNST / NSPT is enabled for I-slice, and inter-LFNST / NSPT is enabled with the faster encoding option.

[0141] The following describes intra prediction types (e.g., intramodes or intra prediction modes) in the enhanced compression model (ECM). One example mode is the directional intra prediction (e.g., angular mode) in WC. To capture the arbitrary edge directions presented in natural video, the number of directional intra modes in VTM5 is extended from 33, as used in HEVC, to 65. The new directional modes in VVC are depicted in FIG. 11, and the planar and DC modes remain the same. These denser directional intra prediction modes apply for all block sizes and for both luma and chroma intra predictions. FIG. 11 is a conceptual diagram illustrating regular intra prediction modes 1100 in VVC Test Model (VTM) 7.0.

[0142] Conventional angular intra prediction directions are defined from 45 degrees to - 135 degrees in clockwise direction, which corresponds to mode 2 to mode 66 in FIG, 11. To provide better prediction for non-square blocks, in VVC, the angles beyond 45 to -135 degrees are considered, which are shown in FIG. 11 for modes [67, 80], and mode [-1, -14], For blocks with width (W) greater than height (H) modes [67, 80] are considered, and for blocks with width (W) less than height (H) modes [-1, -14] are considered. These directional intra prediction modes can be either used in combination with multiple reference lines (MRL), or with an intra-sub partition mode (ISP). The details can be found in G. J, Sullivan, J.-R. Ohm, W.-J. Han and T, Wiegand, “Overview of the High Efficiency Video Coding (HEVC) Standard,” IEEE Transactions on Circuits and Systems for Video Technology, vol. 22, no. 12, Dec. 2012.

[0143] The matrix weighted intra prediction (MIP) method is an intra prediction technique that is part of VVC. FIG. 12 is a conceptual diagram illustrating an example matrix intra prediction process. To predict the samples of a rectangular block of width 1616-600W001W and height H, MIP takes one line of H reconstructed neighboring boundary samples left of the block 1200 and one line of reconstructed neighboring boundary samples above the block 1202 as input. If a reconstructed sample is unavailable, the reconstructed sample may be generated in the same way as conventional intra prediction. The generation of the prediction signal is based on the following three steps, which are averaging (1204), matrix vector multiplication (1206) and linear interpolation (1208).

[0144] There are three different size indices (idx) used for an MIP process. An index is defined as follows. For idx= 0, 1 and 2, 16, 12 and 6 matrices are defined, which also define the number of modes for that given idx. Additionally, each mode is allowed to be transposed, where the samples from left and above are swapped before performing matrix-vector multiplication. So, additionally a transpose flag is signaled (along with the mode signaling) when a CU is coded with MIP.

[0145] Decoder side intramode derivation and fused intra prediction (DIMD) is now discussed. In Abdoli et al., “Non-CE3: Decoder-side Intra Mode Derivation with Prediction Fusion Using Planar,” Video Experts Team (JVET) of ITU-T SG 16 WP 3 and ISO / IEC JTC 1 / SC 29 / WG 11, 22nd Meeting, by teleconference, 20-28 Apr. 2021, document JVET-O0449, intra prediction is performed based on decoder derived intra modes (using already decoded neighboring reconstructed samples) and fusing it with planar predicted samples. Two angular modes are selected from a Histogram of Gradient (HoG) computed from the neighboring pixels of current block. Once the two angular modes are selected, their predictors are computed using conventional angular intra prediction, and the final predictor of the block is generated. The weight of the planar mode is kept at 21 / 64 ( -=1 / 3) and the rest of 43 / 64 is distributed to two angular modes proportionally based on the corresponding amplitudes in the HoG. HoG is computed by sliding a 3x3 window on left and above neighbors reconstructed samples.

[0146] FIG. 13 is a conceptual diagram of HoG computation. In the example of FIG. 13, a template 1300 is located left and above a current block 1302. Template 1300 exists in a reconstructed area 1304 of a picture. An unavailable area 1306 exists below and to the right of current block 1302. If current block 1302 is a 4x4 block, template 1300 comprises an above reference sample set 1308 and a left reference sample set 1310, each comprising 3x3 samples. Video encoder 200 or video decoder 300 may generate a histogram of gradient 1312 based on above reference sample set 1308 and left reference sample set1616-600W0011310, If current block 1302 has block dimensions other than 4x4, aHoG 1314 is computed by sliding 3x3 window 1316 on left and above neighbors reconstructed samples.

[0147] FIG. 14 is a conceptual diagram of weight determination and final predictor generation. In the example of FIG. 14, an Mi directional intramode and an Mi directional intramode are determined using HoG 1400. Weights coi, CO2, and OJ.; are determined based on the amplitudes of Mi and Mi as shown in equations 1402. Additionally, prediction blocks 1404, 1406, and 1408 are determined for a current block 1410 using the Mi directional intra mode, the Mi directional intra prediction mode, and a planar intra mode. Prediction blocks 1404, 1406, and 1408 are then multiplied by the corresponding weights Wi, CO2, and cos and the results are summed to generate a predictor block 1412.

[0148] Template-based intramode derivation with fusion (TIMD) is now discussed. In Wang et al., “EE2-related: Template -based intramode derivation using most probable modes (MPMs),” JVET-V0098, another decoder-side intramode derivation method is proposed as a template-based intramode derivation. The idea for TIMD is shown in FIG.15. FIG. 15 is a conceptual diagram illustrating an example template and reference samples used in template-based intra model derivation with fusion (TIMD). In the example of FIG. 15, a picture 1500 includes a current CU 1502. Two template regions 1504 A, 1504B are chosen (above the current CU 1 02 and left of the current CU 1502) and a reference of template 1506 is chosen correspondingly. For each mode in the MPM list, prediction is generated for the template region and sum of absolute transformed distances (SATD) cost is computed on the template region between the prediction and the reconstruction samples. The mode with the lowest cost is chosen as the mode for TIMD, Also, the number of angular intra modes (including wide angle modes) are extended (doubled) compared to VVC, i.e., the angles are twice densely arranged.

[0149] Furthermore, in Cao et al., “EE2 -related: Fusion for template-based intramode derivation,'’ Video Experts Team (JVET) of ITU-T SG 16 WP 3 and ISO / IECJTC 1 / SC 29 / WG 11, 23rd Meeting, by teleconference, 7-16 July 2021, JVET-W0123, fusion for TIMD is proposed. That is, instead of selecting the only one mode with the smallest SATD cost, this contribution proposes to choose the first two modes with the smallest SATD costs for the intra modes derived using TIMD method and then fuse them with the weights, and such weighted intra prediction is used to code the current CU. The costs of the two selected modes are compared with a threshold, in the test the cost factor of 2 is applied as follows:costMode2 < 2*costModel.1616-600W001

[0150] If this condition is true, the fusion is applied, otherwise only Model is used. Weights of the modes are computed from their SATD costs as follows:weight 1 = costMode2 / (costModel+ costMode2) weight2 = 1 - weight!

[0151] Furthermore, in Andrivon et al., “EE2-1.20: TIMD fusion with non-angular predictor,” Joint Video Experts Team (J VET) of ITU-T SG 16 WP 3 and ISO / IEC JTC 1 / SC 29 / WG 11, 33rd Meeting, by teleconference, 17-26 January 2024, JVET- AG0092, the third fusion candidate is introduced, which is a non-angular intra prediction mode. The conditions below are checked to determine whether the non- angular intra prediction mode is used in fusion:- the non-angular intra prediction mode is different from the two selected intra prediction modes.costMode3 < 1,5*costMode 1, where the costMode3 is the SATD cost of the non- angular intra prediction mode and costModel is the SATD cost of the first intra prediction mode.

[0152] If both of the conditions are true, three intra prediction modes are used to generate the prediction. The weights of each intra prediction mode are computed from an SATD cost, as follows:weight_i = costMode_i / sumSATD = Σ costMode_j

[0153] Spatial geometric partitioning mode (SGPM) is now discussed. SGPM is an intramode that resembles the inter coding tool of geometric partitioning mode (GPM), where the two prediction parts are generated from an intra predicted process, in this mode, a candidate list is built with each entry containing one partition split and two intra prediction modes. 26 partition modes and 9 of intra prediction modes are used to form the combinations. The length of the candidate list is set equal to 16. The selected candidate index is signalled.

[0154] FIG. 16 is a conceptual diagram illustrating an example of spatial geometric partitioning mode. In FIG. 16, samples of a prediction unit 1600 include a first prediction part 1602 and a second prediction part 1604. Samples in the first prediction part 1602 are generated using a first intra prediction process. Samples in the second prediction part 1604 are generated using a second intra prediction process. Samples in a transition zone 1606 are a blend of samples generated using the first intra prediction process and samples generated using the second intra prediction process.1616-600W001

[0155] In the current ECM design, NSPT and LFNST kernels are selected based on block size and derived intramode obtained by intramode derivation process for transform set selection for a given signaled intramode, The derivation process includes either mapping the signaled intra direction to one of 35 transform selection intramodes, or the intramode is derived from DIMD process applied to block neighboring template (DIMD, TIMD), or the prediction block (MIP, extrapolation intra prediction (EIP), IntraTMP, SGPM, intraNN). The DIMD process using Sobel operator and histogram of gradients provide 2 highest occurring intramode directions in the DIMD region that results in a primary and a secondary intramode for transform selection.

[0156] That is, in the current ECM design, the selection of a transform kernel for a current block of video data may depend on two paths. In a first path, if the current block is coded using a conventional directional intra prediction mode (e.g., an angular mode), that signaled mode may be mapped directly to one of 35 transform set indexes (e.g., of FIG, 8). In a second path, if the current block is coded using a non-directional mode (such as MIP, TIMD, EIP, IntraTMP, or SGPM), a separate intramode derivation process is performed for the purpose of transform selection.

[0157] This derivation process, such as a decoder side intramode derivation (DIMD) process, is used for determining an angular mode for the current block. This process may be applied to a prediction block of the current block associated with the non- directional mode, or to neighboring reconstructed samples. This process, which may compute a histogram of gradient (HoG), identifies one or more dominant directions. For instance, the DIMD process may output the two highest occurring intramode directions, which are referred to as a primary intramode (e.g., the most dominant gradient, also called a first angular mode) and a secondary intramode (e.g., the second most dominant gradient, also called a second angular mode). These derived angular modes are then used to select a transform kernel.

[0158] NSPT or LFNST selection is determined by the block size. LFNST transforms cover 35 intramode classes (defined by intramode derivation process for transform set selection), where in each class there are 3 transform kernels. The transform kernel in a class is signaled. NSPT is similar to LFNST transforms but they are further split into 3 categories in ECM-15. Category 1 covers intra modes (TIMD, DIMD, EIP, MIP, SGPM), Category 2 covers IntraTMP and Inter case NSPTs and Category 0 covers the conventional intra modes and intraNN mode. Each category of NSPT follows the LFNST transform set formation.1616-600W001

[0159] For example, a video coder, such as video decoder 300, may be configured to maintain or store information indicative of which coding modes are associated with which category of a plurality of categories. This information maps the signaled coding mode of a block to a specific category. For instance, a first category’ (e.g., Category' 1) is associated with specific non-directional intra modes, such as a template-based intramode derivation (TIMD) mode, decoder side intramode derivation (DIMD) mode, extrapolation intra prediction (EIP) mode, matrix intra prediction (MIP) mode, and spatial geometric partitioning mode (SGPM). A second category (e.g., Category 2) is associated with an intra template matching prediction (IntraTMP) mode or inter-coding mode. A third category (e.g., Category 0) is associated with conventional directional (angular) modes and a neural network (IntraNN) mode. Stated another way, each category is associated with a respective set of transform kernels. When a block is coded, the coding mode (e.g., signaled or derived) of the block determines the category.

[0160] For prediction blocks that do not have clear directionality, more transform kernel choices can benefit. For this, a secondary intramode for transform selection is introduced for TIMD, DIMD, EIP, MIP, SGPM, and IntraTMP blocks, resulting in alternative transform kernels to choose from. Alternatively, for NSPT blocks, since there are multiple categories (currently 3) in ECM-15.0, instead of using a secondary intramode for secondary set of transforms, additional transform categories for NSPT can use the primary intramode for transform selection only. The indication for whether to use primary or secondary transform sets is indicated by a signaled flag.

[0161] For instance, for blocks coded with non-directional modes (such as those in Category 1 and Category 2), relying on a single derived intramode and a single category may be insufficient. In accordance with examples described in this disclosure, to improve coding efficiency, the example techniques introduce more diversity to the available transform kernels, such as by making a secondary, or alternative, set of transforms available.

[0162] This disclosure describes various methods that define an improved non- separable transform kernel choice. Tire elements of the described techniques may be used independently or in any combination.

[0163] As a first example related to NSPT category selection for secondary transform set, for transform sizes where NSPT is applied, for category' 1 intramodes (TIMD, DIMD, EIP, MIP, SGPM), the primary intramode derived for transform selection is used for category 1 transform kernels. When secondary transform set usage is signaled, the 1616-600W001primary' intramode derived for transform selection is applied to category' 0 transform kernels (conventional intramodes and PDP modes). Similarly for category 2 intramodes (IntraTMP and Inter cases), when secondary transform set usage is signaled, the primary' intramode derived for transform selection is applied to category’ 0 transform kernels (conventional intramodes and PDP modes).

[0164] In this example of NSPT category selection, a video coder (e.g., video decoder 300) may implement a category switching technique. Video decoder 300 may determine that a current block has size for which non-separable primary transform (NSPT) is applied. Video decoder 300 may also determine that the current block is coded in a first mode associated with a first category of the plurality of categories. For instance, the first mode may be a MIP mode, which is associated with Category 1, or an IntraTMP mode, which is associated with Category 2. More generally, the first category (e.g., Category 1) may be associated with template-based intramode derivation (TIMD) mode, decoder side intramode derivation (DIMD) mode, extrapolation intra prediction (EIP) mode, matrix intra prediction (MIP) mode, and spatial geometric partitioning mode (SGPM). In this example, the first mode is one of the TIMD mode, DIMD mode, EIP mode, MIP mode, or SGPM. As another example, the first category (e.g., Category 2) may be associated with intra template matching prediction (IntraTMP) mode or inter-coding mode.

[0165] Video decoder 300 may determine an angular mode for the current block, such as the primary intramode derived from the DIMD process. This derived angular mode is different than the first mode (e.g., the primary' intramode is an angular mode, yvhile the first mode is a non-directional mode). If the signaled information indicates that the secondary transform set is to be used, video decoder 300 may determine a transform kernel for NSPT for the current block based on the angular mode and transform kernels associated yvith a second category; the second category' being different than the first category.

[0166] For instance, if the first category is Category 1 or Category 2, and the secondary transform set is signaled, video decoder 300 may apply the derived angular mode (the primary intramode) to the transform kernel sets associated with the second category; which is Category 0 and Category 0 does not include the coding mode in which the current block was coded. Video decoder 300 may use the transform kernel for performing, based on the transform kernel, an inverse transform on a transform coefficient block to generate a residual block associated with the current block and 1616-600W001reconstruct the current block based on the residual block. A video encoder (e.g., video encoder 200) performs a reciprocal process, including performing, based on the transform kernel, a transform on a residual block to generate a transform coefficient block and signaling information based on the transform coefficient block.

[0167] This category switching technique provides several advantages. Again, category' switching, in this example, means that the current block is coded in a coding mode associated with a first category' (e.g., Category’ 1), but the transform kernel for the current block is selected based on a set of transform kernels associated with a second category (e.g., Category 0), where the second category is not associated with the coding mode used to code the current block.

[0168] For instance, a device for encoding or decoding video data may include one or more memories configured to store information indicative of which coding modes are associated with which categories of a plurality of categories, and each category’ is associated with a respective set of transform kernels. Processing circuitry of the device may determine that a current block has size for which non-separable primary / transform (NSPT) is applied, and determine that the current block is coded in a first mode associated with a first category of the plurality' of categories. The processing circuitry’ may also determine an angular mode for the current block, the angular mode being different than the first mode.

[0169] By determining a transform kernel for NSPT for the current block based on the angular mode and transform kernels associated with a second category, the second category being different than the first category’, the techniques increase the diversity of available transform kernels. This improved flexibility is particularly beneficial for blocks coded with non-directional intra-prediction modes (e.g., those in the first category ) that lack clear directionality. This allows the block to use a transform kernel from an alternative “bucket” of kernels, increasing the likelihood of finding an efficient transform, which improves coding efficiency.

[0170] This advantage is achieved efficiently, as the same derived angular mode (e.g., the primary- intramode) is used to access the transform kernels of both the first category' and the second category. This avoids the need to derive or signal additional information, minimizing impact on signaling overhead. A device for encoding video data may perform, based on the transform kernel, a transform on a residual block to generate a transform coefficient block associated with the current block, and signal information based on the transform coefficient block. A device for decoding video data may 1616-600W001perform, based on the transform kernel, an inverse transform on a transform coefficient block to generate a residual block associated with the current block, and reconstruct the current block based on the residual block.

[0171] In one or more examples, whether the example category switching technique is used may be based on signaling from video encoder 200 of whether a primary transform set or a secondary transform set is to be used. In such examples, video decoder 300 may parse information indicating whether a primary transform set or a secondary’ transform set is to be used for transform kernel selection. Video decoder 300 may determine the transform kernel for NSPT for the current block based on the angular mode and the transform kernels associated with the second category' based on the information indicating that the secondary’ transform set is to be used. For example, video decoder 300 may input the angular mode into the mapping table for the second category', and determine a set of transform kernels, from which video decoder 300 determines the transform kernel.

[0172] Otherwise, in this example, if video decoder 300 parses information indicating that the primary transform set is to be used for transform kernel selection, then video decoder 300 may determine the transform kernel for NSPT for the current block based on the angular mode and the transform kernels associated with the first category (e.g., the category that is associated with the coding mode of the current block). For example, video decoder 300 may input the angular mode into the mapping table for the first category, and determine a set of transform kernels, from which video decoder 300 determines the transform kernel.

[0173] The following is examples for alternative NSPT category selection for secondary transform set. For transform sizes where NSP T is applied, for category 1 intramodes (TIMD, DIMD, EIP, MIP, SGPM) and category 2 intramodes (intraTMP and inter modes), the primary and secondary' intramode for transform selection is derived. The difference between the primary and secondary intramodes is compared to a predetermined threshold.

[0174] As further examples, in an alternative technique, for blocks coded in the first category' (e.g., Category' 1 or 2) and applicable NSPT sizes, the derived angular mode comprises a first angular mode (e.g., the primary intramode). Video decoder 300 may be configured to determine a second angular mode for the current block (e.g., the secondary intramode). For instance, video decoder 300 may apply the DIMD process described above to determine two angular modes: the first angular mode (e.g., primary 1616-600W001intramode) and the second angular mode (e.g., secondary intramode). Video decoder 300 may determine a difference value between the first angular mode and the second angular mode, for example by calculating an absolute difference. Video decoder 300 may compare this difference value to a predetermined threshold. The result of this comparison is used to select the transform kernel, as in the following cases.

[0175] In case 1, if the absolute difference is less than a threshold, then for secondary transform set, alternative category described above in NSPT category selection for secondary transform set is used (i.e. Category' 0) with the primary derived transform. Otherwise, the secondary derived intramode is used to select the transform set from the intramode category' (TIMD, DIMD, EIP, MIP, SGPM) or (intraTMP and inter modes) category’ intra block type it belongs to,

[0176] For example, in this example (Case 1), tire selection process depends on the comparison (e.g., comparison of the difference value of the difference between the first and second angular modes to a threshold). In response to the difference value being less than a threshold (i.e., the primary and secondary intramodes are similar), video decoder 300 may determine the transform kernel using the category switching technique (e.g., the NSPT category' selection). This includes determining the transform kernel for NSPT for the current block based on the angular mode (the primary intramode (e.g., the first angular mode)) and transform kernels associated with the second category' (e.g., Category 0). Alternatively, in response to the difference value being greater than or equal to the threshold (i.e., the primary' and secondary intramodes are diverse), video decoder 300 selects the transform set from the original category' (e.g., Category’ 1 or 2) using the secondary intramode (e.g., the second angular mode).

[0177] In case 2, if the absolute difference is less than a threshold, then for secondary transform set, alternative NSPT category' described above in NSPT category selection for secondary transform set is used (i.e. Category' 0) with the secondary' derived transform. Otherwise secondary derived intramode is used to select the transform set from the intramode category (TIMD, DIMD, EIP, MIP, SGPM) or (intra TMP and inter modes) category' intra block type it belongs to.

[0178] As further examples, in another example (Case 2), if tire absolute difference is less than the threshold (e.g., the difference value of the difference between the first and second angular modes is less than the threshold), video decoder 300 may determine the transform kernel from the alternative NSPT category' (e.g., Category' 0) using the secondary derived transform (i.e., the secondary intramode, also called second angular 1616-600W001mode in this example). Otherwise (i.e., if the absolute difference is greater than or equal to the threshold), video decoder 300 may select the transform kernel from the original intramode category' (e.g., Category 1 or 2) using the secondary' intramode.

[0179] In case 3, if the absolute difference is greater than a threshold, then for secondary transform set, alternative category' described above in NSPT category' selection for secondary transform set is used (i.e. Category 0) with the primary' derived transform. Otherwise secondary’ derived intramode is used to select the transform set from the intramode category' (TIMD, DIMD, EIP, MIP, SGPM) or (intraTMP and inter modes) category intra block type it belongs to.

[0180] For instance, in another example (Case 3), if the absolute difference is greater than the threshold (e.g., the difference value of the difference between the first and second angular modes is greater than the threshold), video decoder 300 may determine the transform kernel from tire alternative category' (e.g., Category 0) using the primary derived transform (i.e., the primary' intramode, also called first angular mode in this example). Otherwise (i.e., if the absolute difference is less than or equal to the threshold), video decoder 300 may select (e.g., determine) the transform kernel from the original intramode category' (e.g., Category 1 or 2) using the secondary' intramode (also called second angular mode in this example),

[0181] In case 4, if the absolute difference is greater than a threshold, then for secondary transform set, the alternative NSPT category, described above in NSPT category' selection for secondary transform set, is used (i.e. Category' 0) with the secondary’ derived transform. Otherwise secondary’ derived intramode is used to select the transform set from the intramode category (TIMD, DIMD, EIP, MIP, SGPM) or (intraTMP and inter modes) category intra block type it belongs to.

[0182] In this example (Case 4), the selection process again depends on the comparison. In response to the difference value (e.g., of the difference between the first and second angular modes) being greater than a threshold (i.e., the primary and secondary intramodes are diverse), video decoder 300 may determine the transform kernel using the category’ switching technique. This includes determining the transform kernel for NSPT for the current block based on the angular mode (which in this case is the primary' intramode) and transform kernels associated with the second category (e.g., Category 0). Alternatively, in response to the difference value being less than or equal to the threshold (i.e., the modes are similar), video decoder 300 may select the transform set from the original category (e.g., Category 1 or 2) using the secondary intramode.1616-600W001

[0183] The motivation is to introduce diversity to the transform sets for a given intra mode, in all above 4 cases, the alternative NSPT category could be the other available option also, e.g. if the current mode belongs to category 2, the alternative can be category’ 0 as described above or can be category 1. This alternative category' assignment is predetermined and can be signaled at slice / picture / PPS / SPS level, similarly for an intramode belonging to category 1, the alternative can be category 0 or 2

[0184] This threshold-based determination provides a specific technical advantage in reducing redundancy during the rate-distortion search. When the first and second derived angular modes are numerically close (e.g., the difference value is below the threshold), using both modes to select kernels from the same category’ often results in testing substantially similar transform characteristics. By switching to a different category (e.g., the second category) specifically when the modes are similar, the video coder better ensures that the tested transform kernels remain sufficiently distinct. This better maximizes the diversity of the candidate pool and improves the probability of finding a kernel that efficiently compacts the residual energy, even when the directional derivation yields clustered results.

[0185] That is, the example techniques are not limited to Category' 0, In some examples, if the mode for the current block belongs to Category' 2, the alternative category' may be Category 0 or Category 1. Similarly, if the mode for the current block belongs to Category’ 1, the alternative category’ may be Category 0 or Category 2. This alternative category assignment may be predetermined (e.g., fixed in a coding standard) or signaled at a high level, such as in a slice header, picture parameter set (PPS), or sequence parameter set (SPS).

[0186] This method can be extended to intramodes that belong to category’ 0 (conventional intra modes, PDP). In cases where secondary transforms can be used (e.g. intraNN), the alternative sets can come from category 1 or 2 transform sets. The indication for using primary set or secondary’ set transforms is signaled at CU level for block sizes and intramode where secondary transform sets can be used.

[0187] Stated another way, the category switching technique may also be applied to blocks associated with Category 0. For example, if a block is coded using an intraNN mode (which is in Category' 0), and an indication to use a secondary transform set is signaled, the alternative transform kernel sets may be selected from Category' 1 or1616-600W001Category 2. The indication for using primary set or secondary set transforms is signaled at CU level for the applicable block sizes and intramodes.

[0188] The following relates to M SPT / LFNST kernel set selection. For some intra prediction modes additional kernel sets can be tested, and the decision is signaled.

[0189] In one example for the special intra prediction modes TIMD, DIMD, SGPM, EIP, MIP, IntraTMP, intra block copy (IBC), NN-Intra (also called IntraNN), and PDP, the second kernel set is tested. The dominant intra direction is derived using DIMD-like process, based on the derived primary intramode the transform index is computed. The transforms with the same transform indexes from two different kernel sets are tested during the RD process and the index of the kernel set is signaled in the bitstream. The signaling depends on the availability of non-separable transform sets for the given transform block size.

[0190] As further examples, in some techniques for NSPT or LFNST kernel set selection, a dominant intra direction (e.g., a primary intramode) is derived using a DIMD-like process for a block coded with a special intra prediction mode (e.g., TIMD, DIMD, MIP, etc.). Video encoder 200 and video decoder 300 may each compute a transform index based on this derived primary intramode. Video encoder 200 may then test two different transforms. These two transforms are associated with the same transform index, but are selected from two different kernel sets (e.g., two different categories). Video encoder 200 may use a rate-distortion (RD) process to determine which of the two kernel sets provides better coding efficiency. Video encoder 200 may signal an index of the selected kernel set (e.g., an index of 0 or 1) in the bitstream,

[0191] In one example for the special intra prediction modes TIMD, DIMD, SGPM, EIP, MIP, IntraTMP, IBC, NN-Intra, and PDP, the second kernel and the third set are tested like the previous example. The final index is signaled using unary binarization and separate context for each of the bins.

[0192] Stated another way, this technique may be extended to test three or more kernel sets. For example, a transform associated with the derived transform index may be tested from a first kernel set (e.g,. Category’ 0), a second kernel set (e.g.. Category’ 1), and a third kernel set (e.g., Category' 2). The final index indicating which of the three sets is selected (e.g., an index of 0, 1, or 2) is then signaled, for example, using unary binarization with separate contexts for each bin.

[0193] The following describes NSPT / LFNST kernel set selection and transform index selection. In one example for the special intra prediction modes TIMD, DIMD, SGPM, 1616-600W001EIP, MIP, IntraTMP, IBC, NN-Intra, and PDP, the priman' and secondary' intramode are derived using a DIMD-like process for transform index selection. For transform blocks smaller than a predefined threshold the transform index derived by the primary' intramode is tested from two kernel sets, for the blocks larger than a predefined threshold the two transform indexes derived by primary' and secondary' intramode are tested from the same kernel set. The index of selection is signaled. For the blocks smaller than the threshold the indexes 0 or 1 indicate from which kernel set the transform should be taken, for the blocks larger than threshold the indexes 0 or 1 indicate whether the transform index is derived using dominant or second intra direction.

[0194] As further examples, in another technique for kernel set and transform index selection, video encoder 200 and / or video decoder 300 may derive both a primary intramode and a secondary intramode. For transform blocks smaller than a predefined threshold, video encoder 200 may test two transforms: both transforms are based on the transform index derived from the primary intramode, but each is taken from one of two different kernel sets. Video encoder 200 may signal an index (e.g., 0 or 1) that indicates from which kernel set the transform is selected.

[0195] Stated another way, for transform blocks larger than or equal to the predefined threshold, video encoder 200 may test two different transforms: a first transform is based on the transform index derived from the primary intramode, and a second transform is based on the transform index derived from the secondary intramode. In this case, both transforms are selected from the same kernel set. The signaled index (e.g., 0 or 1) indicates whether the transform index derived from the dominant (primary) or second (secondary) intra direction is to be used.

[0196] In one example for the special intra prediction modes TIMD, DIMD, SGPM, EIP, MIP, IntraTMP, IBC, NN-Intra, and PDP, two transform indexes are derived by primary and secondary intra modes and transforms from two different kernel sets are tested. That is, video encoder 200 and video decoder 300 may' derive two transform indexes using the primary' and secondary' intramodes. Video encoder 200 may test transforms from two different kernel sets. For instance, video encoder 200 may test a first transform from a first kernel set using tire primary intramode, and a second transform from a second kernel set using the secondary' intramode. Video encoder 200 may also test four combinations: (primary intramode, set 0), (primary intramode, set 1),1616-600W001(secondary intramode, set 0), and (secondary intramode, set 1), and signal the best combination.

[0197] In one example for the special intra prediction modes TIMD, DIMD, SGPM, EIP, MIP, IntraTMP, IBC, NN-Intra, and PDP, two transform indexes are set using primary and secondary' intramode derived by a DIMD-like process. Based on the difference between the intra modes the decision is made whether to test transforms from different kernel sets or two different transform indexes from the same set. If the difference between the first and the second intramode is smaller than a predefined threshold, the transforms with the index derived from dominant intra direction from two different kernel sets will be tested, and the decision is signaled.

[0198] Stated another way, in yet another example, video encoder 200 and / or video decoder 300 may derive both the primary' and secondary intramodes and calculate a difference between these two intramodes. This difference is used to determine which set of transforms to test.

[0199] As further examples, if the difference between the first (primary) and the second (secondary) intramode is smaller than a predefined threshold (i.e., the modes are similar), video encoder 200 may test transforms from two different kernel sets, both using the transform index derived from the dominant (primary') intra direction. Video encoder 200 may signal a decision (e.g., an index of 0 or 1) to indicate which kernel set is chosen. Alternatively, if the difference between the intra modes is greater than or equal to the threshold (i.e., the modes are diverse), video encoder 200 may test two different transform indexes (one from the primary' intramode, one from the secondary intramode) from the same kernel set. Video encoder 200 may signal a decision to indicate the intramode for which the transform index is chosen.

[0200] FIG. 2 is a block diagram illustrating an example video encoder 200 that may perform the techniques of this disclosure. FIG. 2 is provided for purposes of explanation and should not be considered limiting of the techniques as broadly exemplified and described in this disclosure. For purposes of explanation, this disclosure describes video encoder 200 according to the techniques of VVC and HEVC. However, the techniques of this disclosure may be performed by video encoding devices that are configured to other video coding standards and video coding formats, such as AVI and successors to the AVI video coding format.

[0201] In the example of FIG. 2, video encoder 200 includes video data memory 230, mode selection unit 202, residual generation unit 204, transform processing unit 206, 1616-600W001quantization unit 208. inverse quantization unit 210, inverse transform processing unit 212, reconstruction unit 214, filter unit 216, decoded picture buffer (DPB) 218, and entropy encoding unit 220. Any or all of video data memory 230, mode selection unit 202, residual generation unit 204, transform processing unit 206, quantization unit 208, inverse quantization unit 210, inverse transform processing unit 212, reconstruction unit 214, filter unit 216, DPB 218, and entropy encoding unit 220 may be implemented in one or more processors or in processing circuitry'. For instance, the units of video encoder 200 may be implemented as one or more circuits or logic elements as part of hardware circuitry, or as part of a processor, ASIC, or FPGA. Moreover, video encoder 200 may include additional or alternative processors or processing circuitry' to perform these and other functions.

[0202] Video data memory' 230 is an example of a memory system that may store video data to be encoded by the components of video encoder 200. Video encoder 200 may receive the video data stored in video data memory 230 from, for example, video source 104 (FIG. 1). DPB 218 is an example of a memory- system that may act as a reference picture memory that stores reference video data for use in prediction of subsequent video data by video encoder 200. Video data memory' 230 and DPB 218 may each be formed by any of a variety of one or more memory' devices or memory- units, such as dynamic random access memory (DRAM), including synchronous DRAM (SDRAM), magnetoresistive RAM (MRAM), resistive RAM (RRAM), or other types of memory devices. Video data memory' 230 and DPB 218 may be provided by the same memorydevice or separate memory devices. In various examples, video data memory 230 may be on-chip with other components of video encoder 200, as illustrated, or off-chip relative to those components.

[0203] In this disclosure, reference to video data memory- 230 should not be interpreted as being limited to memory internal to video encoder 200, unless specifically described as such, or memory external to video encoder 200, unless specifically described as such. Rather, reference to video data memory- 230 should be understood as reference memory that stores video data that video encoder 200 receives for encoding (e.g., video data for a current block that is to be encoded). Memory 106 of FIG. 1 may also provide temporary storage of outputs from the various units of video encoder 200.

[0204] The various units of FIG. 2 are illustrated to assist with understanding the operations performed by video encoder 200. The units may be implemented as fixed-function circuits, programmable circuits, or a combination thereof. Fixed-function 1616-600W001circuits refer to circuits that provide particular functionality, and are preset on the operations that can be performed. Programmable circuits refer to circuits that can be programmed to perform various tasks, and provide flexible functionality in the operations that can be performed. For instance, programmable circuits may execute software or firmware that cause the programmable circuits to operate in the manner defined by instructions of the software or firmware. Fixed-function circuits may execute software instructions (e.g., to receive parameters or output parameters), but the types of operations that the fixed-function circuits perform are generally immutable. In some examples, one or more of the units may be distinct circuit blocks (fixed-function or programmable), and in some examples, one or more of the units may be integrated circuits.

[0205] Video encoder 200 may include arithmetic logic units (ALUs), elementary function units (EFUs), digital circuits, analog circuits, and / or programmable cores, formed from programmable circuits. In examples where the operations of video encoder 200 are performed using software executed by the programmable circuits, memory 106 (FIG. 1) may store the instructions (e.g., object code) of the software that video encoder 200 receives and executes, or another memory within video encoder 200 (not shown) may store such instructions.

[0206] Video data memory 230 is configured to store received video data. Video encoder 200 may retrieve a picture of the video data from video data memory 230 and provide the video data to residual generation unit 204 and mode selection unit 202. Video data in video data memory’ 230 may be raw video data that is to be encoded.

[0207] Mode selection unit 202 includes a motion estimation unit 222, a motion compensation unit 224, and an intra-prediction unit 226. Mode selection unit 202 may include additional functional units to perform video prediction in accordance with other prediction modes. As examples, mode selection unit 202 may include a palette unit, an intra-block copy unit (which may be part of motion estimation unit 222 and / or motion compensation unit 224), an affine unit, a linear model (LM) unit, or the like.

[0208] Mode selection unit 202 generally coordinates multiple encoding passes to test combinations of encoding parameters and resulting rate-distortion values for such combinations. The encoding parameters may include partitioning of CTUs into CUs, prediction modes for the CUs, transform types for residual data of the CUs, quantization parameters for residual data of the CUs, and so on. Mode selection unit 202 may1616-600W001ultimately select the combination of encoding parameters having rate -distortion values that are better than the other tested combinations.

[0209] Video encoder 200 may partition a picture retrieved from video data memory 230 into a series of CTUs, and encapsulate one or more CTUs within a slice. Mode selection unit 202 may partition a CTU of the picture in accordance with a tree structure, such as the MTT structure, QTBT structure, superblock structure, or the quad-tree structure described above. As described above, video encoder 200 may form one or more CUs from partitioning a CTU according to the tree structure. Such a CU may also be referred to generally as a "video block” or “block.”

[0210] In general, mode selection unit 202 also controls the components thereof (e.g., motion estimation unit 222, motion compensation unit 224, and intra-prediction unit 226) to generate a prediction block for a current block (e.g., a current CU, or in HEVC, the overlapping portion of a PU and a TU). For inter-prediction of a current block, motion estimation unit 222 may perform a motion search to identify one or more closely matching reference blocks in one or more reference pictures (e.g., one or more previously coded pictures stored in DPB 218). In particular, motion estimation unit 222 may calculate a value representative of how similar a potential reference block is to the current block, e.g., according to sum of absolute difference (SAD), sum of squared differences (SSD), mean absolute difference (MAD), mean squared differences (MSD), or the like. Motion estimation unit 222 may generally perform these calculations using sample-by-sample differences between the current block and the reference block being considered. Motion estimation unit 222 may identify a reference block having a lowest value resulting from these calculations, indicating a reference block that most closely matches the current block.

[0211] Motion estimation unit 222 may form one or more motion vectors (MVs) that defines the positions of the reference blocks in the reference pictures relative to the position of the current block in a current picture. Motion estimation unit 222 may then provide the motion vectors to motion compensation unit 224. For example, for unidirectional inter-prediction, motion estimation unit 222 may provide a single motion vector, whereas for bi-directional inter-prediction, motion estimation unit 222 may provide two motion vectors. Motion compensation unit 224 may then generate a prediction block using the motion vectors. For example, motion compensation unit 224 may retrieve data of the reference block using the motion vector. As another example, if the motion vector has fractional sample precision, motion compensation unit 224 may 1616-600W001interpolate values for the prediction block according to one or more interpolation filters. Moreover, for bi-directional inter-prediction, motion compensation unit 224 may retrieve data for two reference blocks identified by respective motion vectors and combine the retrieved data, e.g., through sample-by-sample averaging or weighted averaging.

[0212] When operating according to the AV1 video coding format, motion estimation unit 222 and motion compensation unit 224 may be configured to encode coding blocks of video data (e.g., both luma and chroma coding blocks) using translational motion compensation, affine motion compensation, overlapped block motion compensation (OBMC), and / or compound inter-intra prediction.

[0213] As another example, for intra-prediction, or intra-prediction coding, intraprediction unit 226 may generate the prediction block from samples neighboring the current block. For example, for directional modes, intra-prediction unit 226 may generally mathematically combine values of neighboring samples and populate these calculated values in the defined direction across the current block to produce the prediction block. As another example, for DC mode, intra-prediction unit 226 may calculate an average of the neighboring samples to the current block and generate the prediction block to include this resulting average for each sample of the prediction block.

[0214] When operating according to the AV1 video coding format, intra-prediction unit 226 may be configured to encode coding blocks of video data (e.g., both luma 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. Mode selection unit 202 may include additional functional units to perform video prediction in accordance with other prediction modes.

[0215] Mode selection unit 202 provides the prediction block to residual generation unit 204. Residual generation unit 204 receives a raw, unencoded version of tire current block from video data memory 230 and the prediction block from mode selection unit 202. Residual generation unit 204 calculates sample-by-sample differences between the current block and the prediction block. The resulting sample-by-sample differences define a residual block for the current block. In some examples, residual generation unit 204 may also determine differences between sample values in the residual block to generate a residual block using residual differential pulse code modulation (RDPCM).1616-600W001In some examples, residual generation unit 204 may be formed using one or more subtractor circuits that perform binary subtraction.

[0216] In examples where mode selection unit 202 partitions CUs into PUs, each PU may be associated with a luma prediction unit and corresponding chroma prediction units. Video encoder 200 and video decoder 300 may support PUs having various sizes. As indicated 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 a luma prediction unit of the PU. Assuming that the size of a particular CU is 2Nx2N, video encoder 200 may support PU sizes of 2Nx2N or NxN for intra prediction, and symmetric PU sizes of 2Nx2N, 2NxN, Nx2N, NxN, or similar for inter prediction. Video encoder 200 and video decoder 300 may also support asymmetric partitioning for PU sizes of 2NxnU, 2NxnD, nLx2N, and nRx2N for inter prediction.

[0217] In examples where mode selection unit 202 does not further partition a CU into PUs, each CU may be associated with a luma coding block and corresponding chroma coding blocks. As above, the size of a CU may refer to the size of the luma coding block of the CU. The video encoder 200 and video decoder 300 may support CU sizes of 2Nx2N, 2NxN, orNx2N.

[0218] For other video coding techniques such as an intra-block copy mode coding, an affine-mode coding, and linear model (LM) mode coding, as some examples, mode selection unit 202, via respective units associated with the coding techniques, generates a prediction block for the current block being encoded. In some examples, such as palette mode coding, mode selection unit 202 may not generate a prediction block, and instead generate syntax elements that indicate the manner in which to reconstruct the block based on a selected palette. In such modes, mode selection unit 202 may provide these syntax elements to entropy encoding unit 220 to be encoded.

[0219] As described above, residual generation unit 204 receives the video data for the current block and the corresponding prediction block. Residual generation unit 204 then generates a residual block for the current block. To generate the residual block, residual generation unit 204 calculates sample-by-sample differences between the prediction block and the current block.

[0220] Transform processing unit 206 applies one or more transforms to the residual block to generate a block of transform coefficients (referred to h erein as a “transform coefficient block”). Transform processing unit 206 may apply various transforms to a residual block to form the transform coefficient block. For example, transform1616-600W001processing unit 206 may apply a discrete cosine transform (DCT), a directional transform, a Karhunen-Loeve transform (KLT), or a conceptually similar transform to a residual block. In some examples, transform processing unit 206 may perform multiple transforms to a residual block, e.g,, a primary transform and a secondary’ transform, such as a rotational transform. In some examples, transform processing unit 206 does not apply transforms to a residual block.

[0221] When operating according to AV1, transform processing unit 206 may apply one or more transforms to the residual block to generate a block of transform coefficients (referred to herein as a “transform coefficient block”). Transform processing unit 206 may apply various transforms to a residual block to form the transform coefficient block. For example, transform processing unit 206 may apply a horizontal / vertical transform combination that may include a discrete cosine transform (DCT), an asymmetric discrete sine transform (ADST), a flipped ADST (e.g., an ADST in reverse order), and an identity transform (IDTX). When using an identity transform, the transform is skipped in one of the vertical or horizontal directions. In some examples, transform processing may be skipped.

[0222] Quantization unit 208 may quantize the transform coefficients in a transform coefficient block, to produce a quantized transform coefficient block. Quantization unit 208 may quantize transform coefficients of a transform coefficient block according to a quantization parameter (QP) value associated with the current block. Video encoder 200 (e.g., via mode selection unit 202) may adjust the degree of quantization applied to the transform coefficient blocks associated with the current block by adjusting the QP value associated with the CU. Quantization may introduce loss of information, and thus, quantized transform coefficients may have lower precision than the original transform coefficients produced by transform processing unit 206,

[0223] Inverse quantization unit 210 and inverse transform processing unit 212 may apply inverse quantization and inverse transforms to a quantized transform coefficient block, respectively, to reconstruct a residual block from the transform coefficient block. Reconstruction unit 214 may produce a reconstructed block corresponding to the current block (albeit potentially w ith some degree of distortion) based on the reconstructed residual block and a prediction block generated by mode selection unit 202. For example, reconstruction unit 214 may add samples of the reconstructed residual block to corresponding samples from the prediction block generated by mode selection unit 202 to produce the reconstructed block.1616-600W001

[0224] Filter unit 216 may perform one or more filter operations on reconstructed blocks. For example, filter unit 216 may perform deblocking operations to reduce blockiness artifacts along edges of CUs. Operations of filter unit 216 may be skipped, in some examples,

[0225] When operating according to AV1, filter unit 216 may perform one or more filter operations on reconstructed blocks. For example, filter unit 216 may perform deblocking operations to reduce blockiness artifacts along edges of CUs. In other examples, filter unit 216 may apply a constrained directional enhancement filter (CDEF), which may be applied after deblocking, and may include the application of non-separable, non-linear, low-pass directional filters based on estimated edge directions. Filter unit 216 may also include a loop restoration filter, which is applied after CDEF, and may include a separable symmetric normalized Wiener filter or a dual self-guided filter.

[0226] Video encoder 200 stores reconstructed blocks in DPB 218. For instance, in examples -where operations of filter unit 216 are not performed, reconstruction unit 214 may store reconstructed blocks to DPB 218. In examples where operations of filter unit 216 are performed, filter unit 216 may store the filtered reconstructed blocks to DPB 218. Motion estimation unit 222 and motion compensation unit 224 may retrieve a reference picture from DPB 218, formed from the reconstructed (and potentially filtered) blocks, to inter-predict blocks of subsequently encoded pictures. In addition, intra-prediction unit 226 may use reconstructed blocks in DPB 218 of a current picture to intra-predict other blocks in the current picture.

[0227] In general, entropy encoding unit 220 may entropy encode syntax elements received from other functional components of video encoder 200. For example, entropy encoding unit 220 may entropy encode quantized transform coefficient blocks from quantization unit 208. As another example, 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 mode selection unit 202. Entropy encoding unit 220 may perform one or more entropy encoding operations on the syntax elements, which are another example of video data, to generate entropy-encoded data. For example, entropy encoding unit 220 may perform a context-adaptive variable length coding (CAVLC) operation, a CABAC operation, a variable-to-variable (V2V) length coding operation, a syntax-based context-adaptive binary' arithmetic coding (SBAC) operation, a Probability Interval Partitioning Entropy (PIPE) coding operation, an 1616-600W001Exponential-Golomb encoding operation, or another type of entropy encoding operation on the data. In some examples, entropy encoding unit 220 may operate in bypass mode where syntax elements are not entropy encoded.

[0228] Video encoder 200 may output a bitstream that includes the entropy encoded syntax elements needed to reconstruct blocks of a slice or picture. In particular, entropy encoding unit 220 may output the bitstream.

[0229] In accordance with AVI, entropy encoding unit 220 may be configured as a symbol-to-symbol adaptive multi-symbol arithmetic coder. A syntax element in AVI includes an alphabet of N elements, and a context (e.g., probability model) includes a set of N probabilities. Entropy encoding unit 220 may store the probabilities as n-bit (e.g., 15-bit) cumulative distribution functions (CDFs). Entropy encoding unit 220 may perform recursive scaling, with an update factor based on the alphabet size, to update the contexts.

[0230] The operations described above are described with respect to a block. Such description should be understood as being operations for a luma coding block and / or chroma coding blocks. As described above, in some examples, the luma coding block and chroma coding blocks are luma and chroma components of a CU. In some examples, the luma coding block and the chroma coding blocks are luma and chroma components of a PU.

[0231] In some examples, operations performed with respect to a luma coding block need not be repeated for the chroma coding blocks. As one example, operations to identify a motion vector (MV) and reference picture for a luma coding block need not be repeated for identifying a MV and reference picture for tire chroma blocks. Rather, the MV for the luma coding block may be scaled to determine the MV for the chroma blocks, and the reference picture may be the same. As another example, the intraprediction process may be the same for the luma coding block and the chroma coding blocks.

[0232] 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 in circuitry' and configured to perform any techniques of this disclosure for transform selection. For example, video encoder 200 or a device that includes video encoder 200 may’ include one or more memories (e.g., any of the example memories described in this disclosure) and processing circuitry coupled to the one or more memories. The one or more memories may be configured to store 1616-600W001information indicative of which coding modes are associated with which categories of a plurality of categories, and each category is associated with a respective set of transform kernels. The processing circuitry may be configured to determine that a current block has size for which non-separable primary transform (NSPT) is applied, determine that the current block is coded in a first mode associated with a first category of the plurality of categories, determine an angular mode for the current block, the angular mode being different than the first mode, determine a transform kernel for NSPT for the current block based on the angular mode and transform kernels associated with a second category, the second category being different than the first category, perform, based on the transform kernel, a transform on a residual block to generate a transform coefficient block associated with the current block, and signal information based on the transform coefficient block.

[0233] FIG. 3 is a block diagram illustrating an example video decoder 300 that may perform the techniques of this disclosure, FIG. 3 is provided for purposes of explanation and is not limiting on the techniques as broadly exemplified and described in this disclosure. For purposes of explanation, this disclosure describes video decoder 300 according to the techniques of VVC and HEVC. However, the techniques of this disclosure may be performed by video coding devices that are configured to other video coding standards.

[0234] In the example of FIG. 3, video decoder 300 includes coded picture buffer (CPB) memory 320, entropy decoding unit 302, prediction processing unit 304, inverse quantization unit 306, inverse transform processing unit 308, reconstruction unit 310, filter unit 312, and DPB 314. Any or all of CPB memory 320, entropy decoding unit 302, prediction processing unit 304, inverse quantization unit 306, inverse transform processing unit 308, reconstruction unit 310, filter unit 312, and DPB 314 may be implemented in one or more processors or in processing circuitry. For instance, the units of video decoder 300 may be implemented as one or more circuits or logic elements as part of hardware circuitry, or as part of a processor, ASIC, or FPGA.Moreover, video decoder 300 may include additional or alternative processors or processing circuitry to perform these and other functions.

[0235] Prediction processing unit 304 includes motion compensation unit 316 and intraprediction unit 318. Prediction processing unit 304 may include additional units to perform prediction in accordance with other prediction modes. As examples, prediction processing unit 304 may include a palette unit, an intra-block copy unit (which may 1616-600W001form part of motion compensation unit 316). an affine unit, a linear model (LM) unit, or the like. In other examples, video decoder 300 may include more, fewer, or different functional components.

[0236] When operating according to AV1, motion compensation unit 316 may be configured to decode coding blocks of video data (e.g., both luma and chroma coding blocks) using translational motion compensation, affine motion compensation, OBMC, and / or compound inter-intra prediction, as described above. Intra-prediction unit 318 may be configured to decode coding blocks of video data (e.g., both luma and chroma coding blocks) using directional intra prediction, non-directional intra prediction, recursive filter intra prediction, CFL, IBC, and / or color palette mode, as described above,

[0237] CPB memory 320 is an example of a memory system that may store video data, such as an encoded video bitstream, to be decoded by the components of video decoder 300. The video data stored in CPB memory 320 may be obtained, for example, from computer-readable medium 110 (FIG. 1). CPB memory 320 may include a CPB that stores encoded video data (e.g., syntax elements) from an encoded video bitstream. Also, CPB memory 320 may store video data other than syntax elements of a coded picture, such as temporary' data representing outputs from the various units of video decoder 300. DPB 314 is an example of a memory system that generally stores decoded pictures, which video decoder 300 may output and / or use as reference video data when decoding subsequent data or pictures of the encoded video bitstream. CPB memory' 320 and DPB 314 may each be formed by any of a variety of memory devices or memory units, such as DRAM, including SDRAM, MRAM, RRAM, or other types of memory devices. CPB memory 320 and DPB 314 may be provided by the same memory device or separate memory devices. In various examples, CPB memory 320 may be on-chip with other components of video decoder 300, or off-chip relative to those components.

[0238] Additionally or alternatively, in some examples, video decoder 300 may retrieve coded video data from memory 120 (FIG. 1). That is, memory 120 may store data as discussed above with CPB memory 320. Likewise, memory 120 may store instructions to be executed by video decoder 300, when some or all of the functionality of video decoder 300 is implemented in software to be executed by processing circuitry of video decoder 300.1616-600W001

[0239] The various units shown in FIG. 3 are illustrated to assist with understanding the operations performed by video decoder 300. The units may be implemented as fixed- function circuits, programmable circuits, or a combination thereof. Similar to FIG. 2, fixed-function circuits refer to circuits that provide particular functionality, and are preset on the operations that can be performed. Programmable circuits refer to circuits that can be programmed to perform various tasks, and provide flexible functionality in the operations that can be performed. For instance, programmable circuits may execute software or firmware that cause the programmable circuits to operate in the manner defined by instructions of the software or firmware. Fixed-function circuits may execute software instructions (e.g., to receive parameters or output parameters), but the types of operations that the fixed-function circuits perform are generally immutable. In some examples, one or more of the units may be distinct circuit blocks (fixed-function or programmable), and in some examples, one or more of the units may be integrated circuits.

[0240] Video decoder 300 may include ALUs, EFUs, digital circuits, analog circuits, and / or programmable cores formed from programmable circuits. In examples where the operations of video decoder 300 are performed by software executing on the programmable circuits, on-chip or off-chip memory may store instructions (e.g., object code) of the software that video decoder 300 receives and executes.

[0241] Entropy decoding unit 302 may receive encoded video data from the CPB and entropy decode the video data to reproduce syntax elements. Prediction processing unit 304, inverse quantization unit 306, inverse transform processing unit 308, reconstruction unit 310, and filter unit 312 may generate decoded video data based on the syntax elements extracted from the bitstream.

[0242] In general, video decoder 300 reconstructs a picture on a block-by-block basis. Video decoder 300 may perform a reconstruction operation on each block individually (where the block currently being reconstructed, i.e., decoded, may be referred to as a “current block”).

[0243] Entropy decoding unit 302 may entropy decode syntax elements defining quantized transform coefficients of a quantized transform coefficient block, as well as transform information, such as a quantization parameter (QP) and / or transform mode indication(s). Inverse quantization unit 306 may use the QP associated with the quantized transform coefficient block to determine a degree of quantization and, likewise, a degree of inverse quantization for inverse quantization unit 306 to apply. 1616-600W001Inverse quantization unit 306 may, for example, perform a bitwise left-shift operation to inverse quantize the quantized transform coefficients. Inverse quantization unit 306 may thereby form a transform coefficient block including transform coefficients.

[0244] After inverse quantization unit 306 forms the transform coefficient block, inverse transform processing unit 308 may apply one or more inverse transforms to the transform coefficient block to generate a residual block associated with the current block. For example, inverse transform processing unit 308 may apply an inverse DCT, an inverse integer transform, an inverse Karhunen-Loeve transform (KLT), an inverse rotational transform, an inverse directional transform, or another inverse transform to the transform coefficient block.

[0245] Furthermore, prediction processing unit 304 generates a prediction block according to prediction information syntax elements that were entropy decoded by entropy decoding unit 302. For example, if the prediction information syntax elements indicate that the current block is inter-predicted, motion compensation unit 316 may generate the prediction block. In this case, the prediction information syntax elements may indicate a reference picture in DPB 314 from which to retrieve a reference block, as well as a motion vector identifying a location of the reference block in the reference picture relative to the location of the current block in the current picture. Motion compensation unit 316 may generally perform the inter-prediction process in a manner that is substantially similar to that described with respect to motion compensation unit 224 (FIG. 2).

[0246] As another example, if the prediction information syntax elements indicate that the current block is intra-predicted, intra-prediction unit 318 may generate the prediction block according to an intra-prediction mode indicated by the prediction information syntax elements. Again, intra-prediction unit 318 may generally perform the intra-prediction process in a manner that is substantially similar to that described with respect to intra-prediction unit 226 (FIG. 2). Intra-prediction unit 318 may retrieve data of neighboring samples to the current block from DPB 314.

[0247] Reconstruction unit 310 may reconstruct the current block using the prediction block and the residual block. For example, reconstruction unit 310 may add samples of the residual block to corresponding samples of the prediction block to reconstruct the current block.

[0248] Filter unit 312 may perform one or more filter operations on reconstructed blocks. For example, filter unit 312 may perform deblocking operations to reduce 1616-600W001blockiness artifacts along edges of the reconstructed blocks. Operations of filter unit 312 are not necessarily performed in all examples.

[0249] Video decoder 300 may store the reconstructed blocks in DPB 314. For instance, in examples where operations of filter unit 312 are not performed, reconstruction unit 310 may store reconstructed blocks to DPB 314. In examples where operations of filter unit 312 are performed, filter unit 312 may store the filtered reconstructed blocks to DPB 314. As discussed above, DPB 314 may provide reference information, such as samples of a current picture for intra-prediction and previously decoded pictures for subsequent motion compensation, to prediction processing unit 304. Moreover, video decoder 300 may output decoded pictures (e.g., decoded video) from DPB 314 for subsequent presentation on a display device, such as display device 118 of FIG. 1.

[0250] In this manner, video decoder 300 represents an example of a video decoding device including a memory’ configured to store video data, and one or more processing units implemented in circuitry' and configured to perform any techniques of this disclosure for transform selection. For example, video decoder 300 or a device that includes video decoder 300 may include one or more memories (e.g., any of the example memories described in this disclosure) and processing circuitry coupled to the one or more memories. The processing circuitry' may be configured to determine that a current block has size for which non-separable primary transform (NSPT) is applied, determine that the current block is coded in a first mode associated with a first category of the plurality' of categories, determine an angular mode for the current block, the angular mode being different than the first mode, determine a transform kernel for NSPT for the current block based on the angular mode and transform kernels associated with a second category', the second category' being different than the first category', perform, based on the transform kernel, an inverse transform on a transform coefficient block to generate a residual block associated with the current block, and reconstruct the current block based on the residual block.

[0251] FIG, 4 is a flowchart illustrating an example method for encoding a current block in accordance with the techniques of this disclosure. The current block may be or include a current CU. Although described with respect to video encoder 200 (FIGS. 1 and 2), it should be understood that other devices may be configured to perform a method similar to that of FIG. 4.1616-600W001

[0252] In this example, video encoder 200 initially predicts the current block (400). For example, video encoder 200 may form a prediction block for the current block. Video encoder 200 may then calculate a residual block for the current block (402). To calculate the residual block, video encoder 200 may calculate a difference between the original, unencoded block and the prediction block for the current block. Video encoder 200 may then transform the residual block and quantize transform coefficients of the residual block (404). Next, video encoder 200 may scan the quantized transform coefficients of the residual block (406). During the scan, or following the scan, video encoder 200 may entropy encode the transform coefficients (408). For example, video encoder 200 may encode the transform coefficients using CAVLC or CAB AC. Video encoder 200 may then output the entropy encoded data of the block (410),

[0253] FIG. 5 is a flowchart illustrating an example method for decoding a current block of video data in accordance with the techniques of this disclosure. The current block may be or 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 perform a method similar to that of FIG. 5.

[0254] Video decoder 300 may receive entropy encoded data for the current block, such as entropy encoded prediction information and entropy encoded data for transform coefficients of a residual block corresponding to the current block (500). Video decoder 300 may entropy decode the entropy encoded data to determine prediction information for the current block and to reproduce transform coefficients of the residual block (502). Video decoder 300 may predict the current block (504), e.g,, using an intra- or interprediction mode as indicated by the prediction information for the current block, to calculate a prediction block for the current block. Video decoder 300 may then inverse scan the reproduced transform coefficients (506), to create a block of quantized transform coefficients. Video decoder 300 may then inverse quantize the transform coefficients and apply an inverse transform to the transform coefficients to produce a residual block (508). Video decoder 300 may ultimately decode the current block by combining the prediction block and the residual block (510).

[0255] FIG. 17 is a flowchart illustrating an example operation of a video decoder, in accordance with one or more techniques of this disclosure. The operation may be performed by a device for decoding video data, such as the destination device 116 or the video decoder 300, which includes processing circuitry and one or more memories.1616-600W001

[0256] Initially, the one or more memories of the video decoder 300 are configured to store information indicative of which coding modes are associated with which category of a plurality of categories, and each category is associated with a respective set of transform kernels (1700). This stored mapping may be utilized by the processing circuitry to perform example techniques like tire category-switching techniques. For example, the first category' is associated with template -based intramode derivation (TIMD) mode, decoder side intramode derivation (DIMD) mode, extrapolation intra prediction (EIP) mode, matrix intra prediction (MIP) mode, and spatial geometric partitioning mode (SGPM), and wherein the first mode is one of the TIMD mode, DIMD mode, EIP mode, MIP mode, or SGPM. In another example, the first category is associated with intra template matching prediction (IntraTMP) mode or inter-coding mode. The second category; which serves as an alternative “bucket’ of kernels, is associated with angular modes and a neural network (IntraNN) mode.

[0257] The processing circuitry may be configured to determine that a current block has size for which non-separable primary transform (NSPT) is applied (1702). The processing circuitry may be also configured to determine that the current block is coded in a first mode associated with a first category of the plurality of categories (1704). For example, the first mode may be the MIP mode, which is a non-directional mode associated with the first category / (e.g., Category' 1).

[0258] Tlie processing circuitry may be configured to determine an angular mode for the current block, the angular mode being different than the first mode (1706). This step may' be advantageous because the first mode (e.g., MIP) is non-directional, but the derived angular mode provides a basis for transform selection. In some examples, to determine the angular mode, the processing circuitry' may be configured to determine the angular mode based on at least one of: parsing information indicating the angular mode or histogram of gradient (HoG) computed from neighboring samples of the current block (e.g., DIMD-like process). This HoG-based derivation allows video decoder 300 to find a dominant direction even in non-directional blocks.

[0259] The processing circuitry may be configured to determine a transform kernel for NSPT for the current block based on the angular mode and transform kernels associated with a second category, the second category being different than the first category' (1708). Uris “category' switching” provides an advantage by increasing the diversity of available transform kernels. Instead of being limited to the kernels associated with the block's own first category, video decoder 300 can select a more efficient kernel from the 1616-600W001second category's “bucket” using the same derived angular mode. This improves coding efficiency, especially for blocks lacking clear directionality, without the signaling overhead of deriving a new mode.

[0260] The determination to use the second category may be signaled. For example, the processing circuitry may be configured to parse information indicating whether a primary transform set or a secondary transform set is to be used for transform kernel selection. In this case, to determine the transform kernel, the processing circuitry may be configured to determine the transform kernel for NSPT for the current block based on the angular mode and the transform kernels associated with the second category based on the information indicating that the secondary transform set is to be used.

[0261] In other examples, the selection is conditional. The angular mode comprises a first angular mode (e.g., a primary intramode from the HoG). The processing circuitry' may be further configured to determine a second angular mode for the current block (e.g., a secondary’ intramode), and determine a difference value between the first angular mode and the second angular mode. Tire selection logic then depends on this difference. For instance, to determine the transform kernel, the processing circuitry may be configured to determine the transform kernel for NSPT for the current block based on the angular mode and transform kernels associated with the second category in response to the difference value being less than a threshold. In another example, to determine the transform kernel, the processing circuitry may be configured to determine the transform kernel for NSPT for the current block based on the angular mode and transform kernels associated with the second category in response to the difference value being greater than a threshold. This conditional logic provides a more nuanced selection, using the alternative category' only when the derived modes suggest a clear benefit.

[0262] After the transform kernel is determined, the processing circuitry’ may be configured to perform, based on the transform kernel, an inverse transform on a transform coefficient block to generate a residual block associated with the current block (1710). The processing circuitry may be configured to reconstruct the current block based on the residual block (1712), for example, by adding the residual block to a corresponding prediction block.

[0263] FIG. 18 is a flowchart illustrating an example operation of a video encoder, in accordance with one or more techniques of this disclosure. The operation may be performed by a device for encoding video data, such as the source device 102 or the video encoder 200, yvhich includes processing circuitry and one or more memories. 1616-600W001

[0264] The one or more memories are configured to store information indicative of which coding modes are associated with which categories of a plurality of categories, and each category is associated with a respective set of transform kernels (1800). This stored mapping allows video encoder 200 to organize and access different sets of kernels. For example, the first category is associated with template-based intramode derivation (TIMD) mode, decoder side intramode derivation (DIMD) mode, extrapolation intra prediction (EIP) mode, matrix intra prediction (MIP) mode, and spatial geometric partitioning mode (SGPM), or associated with intra template matching prediction (IntraTMP) mode or inter-coding mode, and wherein the second category is associated with angular modes and a neural network (IntraNN) mode.

[0265] The processing circuitry’ may be configured to determine that a current block has size for which non-separable primary transform (NSPT) is applied (1802). Tire processing circuitry may be also configured to determine that the current block is coded in a first mode associated with a first category’ of the plurality of categories (1804). For instance, the processing circuitry' determines the block is coded using a non -directional mode like MIP, which belongs to the first category.

[0266] The processing circuitry may be configured to determine an angular mode for the current block, the angular mode being different than the first mode (1806). This provides the advantage of deriving a dominant direction (the angular mode), for example using a histogram of gradient process, which can be used for transform selection even when the block's coding mode (the first mode) is non-directional.

[0267] The processing circuitry may be configured to determine a transform kernel for NSPT for the current block based on the angular mode and transform kernels associated with a second category', the second category’ being different than the first category (1808). Tills “category’ switching” technique may be advantageous, allowing video encoder 200 to test and select a transform kernel from an alternative “bucket” of kernels. Video encoder 200 may perform a rate -distortion analysis to determine if selecting a kernel from the second category provides better compression efficiency than selecting from the first category', thereby improving the overall coding performance.

[0268] Once the optimal transform kernel is determined, the processing circuitry may be configured to perform, based on the transform kernel, a transform on a residual block to generate a transform coefficient block associated with the current block (1810). The residual block represents the difference between the original block and a prediction block.1616-600W001

[0269] The processing circuitry may be configured to signal information based on the transform coefficient block (1812). This signaling includes the quantized transform coefficients and any necessary' syntax elements (e.g., a flag or index) to inform video decoder 300 which transform kernel or category was selected, enabling video decoder 300 to perform the correct inverse operations.

[0270] The following numbered clauses illustrate one or more clauses of the devices and techniques described in this disclosure.

[0271] Clause 1 A. A method of coding video data, the method comprising: determining a non-separable transform kernel according to any combination of techniques of this disclosure.

[0272] Clause 2A. Tire method of Clause 1A, wherein the non-separable transform kernel is part of a non-separable primary transform or a low -frequency non-separable transform.

[0273] Clause 3A, The method of any of Clauses 1 A-2A, wherein coding comprises decoding.

[0274] Clause 4A. Tire method of any of Clauses 1A-3A, wherein coding comprises encoding.

[0275] Clause 5A. A device for coding video data, the device comprising one or more means for performing the method of any of Clauses 1 A-4A.

[0276] Clause 6A. The device of Clause 5A, wherein the one or more means comprise one or more processors implemented in circuitry.

[0277] Clause 7A, The device of any of Clauses 5 A and 6A, further compri sing a memory to store the video data.

[0278] Clause 8A. lire device of any of Clauses 5A-7A, further comprising a display configured to display decoded video data.

[0279] Clause 9A. The device of any of Clauses 5A-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.

[0280] Clause 10A. The device of any of Clauses 5A-9A, wherein the device comprises a video decoder.

[0281] Clause 11A. The device of any of Clauses 5 A- 10A, wherein the device comprises a video encoder.1616-600W001

[0282] Clause 12A. A computer-readable storage medium having stored thereon instructions that, when executed, cause one or more processors to perform the method of any of Clauses 1A-4A.

[0283] Clause IB. A method of decoding video data, the method comprising storing information indicative of which coding modes are associated with which category of a plurality of categories, and each category is associated with a respective set of transform kernels, determining that a current block has size for which non-separable primary transform (NSPT) is applied, determining that the current block is coded in a first mode associated with a first category of the plurality of categories, determining an angular mode for the current block, the angular mode being different than the first mode, determining a transform kernel for NSPT for the current block based on the angular mode and transform kernels associated with a second category, the second category being different than tire first category, performing, based on the transform kernel, an inverse transform on a transform coefficient block to generate a residual block associated with the current block, and reconstructing the current block based on the residual block.

[0284] Clause 2B. The method of Clause IB, further comprising parsing information indicating whether a primary transform set or a secondary transform set is to be used for transform kernel selection, and wherein determining tire transform kernel comprises determining the transform kernel for NSPT for the current block based on the angular mode and the transform kernels associated with the second category' based on the information indicating that the secondary’ transform set is to be used.

[0285] Clause 3B. The method of any of Clauses 1B-2B, wherein the angular mode comprises a first angular mode, the method further comprising determining a second angular mode for the current block, and determining a difference value between the first angular mode and the second angular mode.

[0286] Clause 4B. The method of Clause 3B, wherein determining the transform kernel comprises determining the transform kernel for NSPT for the current block based on the angular mode and transform kernels associated with the second category' in response to the difference value being less than a threshold.

[0287] Clause 5B. The method of Clause 3B, wherein determining the transform kernel comprises determining the transform kernel for NSPT for the current block based on the angular mode and transform kernels associated with the second category' in response to the difference value being greater than a threshold.1616-600W001

[0288] Clause 6B. The method of any of Clauses 1B-5B, wherein determining the angular mode comprises determining the angular mode based on at least one of parsing information indicating the angular mode or histogram of gradient (HoG) computed from neighboring samples of the current block,

[0289] Clause 7B. The method of any of Clauses 1B-6B, wherein the first category is associated with template-based intramode derivation (TIMD) mode, decoder side intramode derivation (DIMD) mode, extrapolation intra prediction (EIP) mode, matrix intra prediction (MIP) mode, and spatial geometric partitioning mode (SGPM), and wherein the first mode is one of the TIMD mode, DIMD mode, EIP mode, MIP mode, or SGPM.

[0290] Clause 8B. The method of any of Clauses 1B-6B, wherein the first category is associated with intra template matching prediction (IntraTMP) mode or inter-coding mode.

[0291] Clause 9B. Tire method of any of Clauses 1B-8B, wherein the second category’ is associated with angular modes and a neural network (IntraNN) mode.

[0292] Clause 10B. A device for decoding video data, the device comprising one or more memories configured to store information indicative of which coding modes are associated with which category’ of a plurality.' of categories, and each category is associated with a respective set of transform kernels, and processing circuitry' coupled to the one or more memories, the processing circuitry being configured to determine that a current block has size for which non-separable primary transform (NSPT) is applied, determine that the current block is coded in a first mode associated with a first category’ of the plurality of categories, determine an angular mode for the current block, the angular mode being different than the first mode, determine a transform kernel for NSPT for the current block based on the angular mode and transform kernels associated with a second category; the second category' being different than the first category, perform, based on the transform kernel, an inverse transform on a transform coefficient block to generate a residual block associated with the current block, and reconstruct the current block based on the residual block.

[0293] Clause 11B. The device of Clause 10B, wherein the processing circuitry' is configured to parse information indicating whether a primary transform set or a secondary transform set is to be used for transform kernel selection, and wherein to determine the transform kernel, the processing circuitry is configured to determine the transform kernel for NSPT for the current block based on the angular mode and the 1616-600W001transform kernels associated with the second category' based on the information indicating that the secondary transform set is to be used.

[0294] Clause 12B. The device of any of Clauses 1 OB-1 IB, wherein the angular mode comprises a first angular mode, and wherein the processing circuitry' is further configured to determine a second angular mode for the current block, and determine a difference value between the first angular mode and the second angular mode.

[0295] Clause 13B. The device of Clause 12B, wherein to determine the transform kernel, the processing circuitry is configured to determine the transform kernel for NSPT for the current block based on the angular mode and transform kernels associated with the second category in response to the difference value being less than a threshold.

[0296] Clause 14B, The device of Clause 12B, wherein to determine the transform kernel, tire processing circuitry' is configured to determine the transform kernel for NSPT for the current block based on the angular mode and transform kernels associated with the second category' in response to the difference value being greater than a threshold.

[0297] Clause 15B. The device of any of Clauses 10B-14B, wherein to determine the angular mode, the processing circuitry is configured to determine the angular mode based on at least one of parsing information indicating the angular mode or histogram of gradient (HoG) computed from neighboring samples of tire current block.

[0298] Clause 16B. The device of any of Clauses 10B-15B, wherein the first category is associated with template-based intramode derivation (TIMD) mode, decoder side intramode derivation (DIMD) mode, extrapolation intra prediction (EIP) mode, matrix intra prediction (MIP) mode, and spatial geometric partitioning mode (SGPM), and wherein the first mode is one of the TIMD mode, DIMD mode, EIP mode, MIP mode, or SGPM,

[0299] Clause 17B. The device of any of Clauses 10B-15B, wherein the first category' is associated with intra template matching prediction (IntraTMP) mode or inter-coding mode.

[0300] Clause 18B. Tire device of any of Clauses 10B-17B, wherein the second category'- is associated with angular modes and a neural network (IntraNN) mode.

[0301] Clause 19B. A device for encoding video data, the device comprising one or more memories configured to store information indicative of w hich coding modes are associated with which categories of a plurality of categories, and each category' is associated with a respective set of transform kernels, and processing circuitry coupled to 1616-600W001the one or more memories, the processing circuitry being configured to determine that a current block has size for which non-separable primary transform (NSPT) is applied, determine that the current block is coded in a first mode associated with a first category of the plurality’ of categories, determine an angular mode for the current block, the angular mode being different than the first mode, determine a transform kernel for NSPT for the current block based on the angular mode and transform kernels associated with a second category, the second category’ being different than the first category, perform, based on the transform kernel, a transform on a residual block to generate a transform coefficient block associated with the current block, and signal information based on the transform coefficient block.

[0302] Clause 20B, The device of Clause 19B, wherein the first category is associated with template-based intramode derivation (TIMD) mode, decoder side intramode derivation (DIMD) mode, extrapolation intra prediction (EIP) mode, matrix intra prediction (MIP) mode, and spatial geometric partitioning mode (SGPM), or associated with intra template matching prediction (IntraTMP) mode or inter-coding mode, and wherein the second category is associated with angular modes and a neural network (IntraNN) mode.

[0303] It is to be recognized that depending on the example, certain acts or events of any of the techniques described herein can be performed in a different sequence, may be added, merged, or left out altogether (e.g., not all described acts or events are necessary for the practice of the techniques). Moreover, in certain examples, acts or events may be performed concurrently, e.g., through multi-threaded processing, interrupt processing, or multiple processors, rather than sequentially.

[0304] In one or more examples, tire functions described may be implemented in hardware, software, firmware, or any combination thereof. If implemented in software, the functions may be stored on or transmited over as one or more instructions or code on a computer-readable medium and executed by a hardware-based processing unit. Computer-readable media may include computer-readable storage media, which corresponds to a tangible medium such as data storage media, or communication media including any medium that facilitates transfer of a computer program from one place to another, e.g., according to a communication protocol. In this manner, computer- readable media generally may correspond to (1) tangible computer-readable storage media which is non-transitory or (2) a communication medium such as a signal or carrier wave. Data storage media may be any available media that can be accessed by 1616-600W001one or more computers or one or more processors to retrieve instructions, code and / or data structures for implementation of the techniques described in this disclosure. A computer program product may include a computer-readable medium.

[0305] By way of example, and not limitation, such computer-readable storage media may include one or more of 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 desired program code in the form of instructions or data structures and that can be accessed by a computer. Also, any connection is properly termed a computer-readable medium. For example, if instructions are transmitted from a website, server, or other remote source using a 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 medium. It should be understood, however, that computer-readable storage media and data storage media do not include connections, carrier waves, signals, or other transitory media, but are instead directed to non-transitory, tangible storage media. Disk and disc, as used herein, includes compact disc (CD), laser disc, optical disc, digital versatile disc (DVD), floppy disk and Blu-ray disc, where disks usually reproduce data magnetically, while discs reproduce data optically with lasers.Combinations of the above should also be included within the scope of computer- readable media.

[0306] Instructions may be executed by one or more processors, such as one or more DSPs, general purpose microprocessors, ASICs, FPGAs, or other equivalent integrated or discrete logic circuitry. Accordingly, the terms “processor” and “processing circuitry,” as used herein may refer to any of the foregoing structures or any other structure suitable for implementation of the techniques described herein. In addition, in some aspects, the functionality described herein may be provided within dedicated hardware and / or software modules configured for encoding and decoding, or incorporated in a combined codec. Also, the techniques could be fully implemented in one or more circuits or logic elements.

[0307] The techniques of this disclosure may be implemented in a wide variety of devices or apparatuses, including a wireless handset, an integrated circuit (IC) or a set of ICs (e.g., a chip set). Various components, modules, or units are described in this disclosure to emphasize functional aspects of devices configured to perform the 1616-600W001disclosed techniques, but do not necessarily require realization by different hardware units. Rather, as described above, various units may be combined in a codec hardware unit or provided by a collection of interoperative hardware units, including one or more processors as described above, in conjunction with suitable software and / or firmware.

[0308] Various examples have been described. These and other examples are within the scope of the following claims.1616-600W001

Claims

WHAT IS CLAIMED IS:

1. A method of decoding video data, the method comprising:storing information indicative of which coding modes are associated with which category of a plurality of categories, and each category is associated with a respective set of transform kernels;determining that a current block has size for which non-separable primary’ transform (NSPT) is applied;determining that the current block is coded in a first mode associated with a first category of the plurality of categories;determining an angular mode for the current block, the angular mode being different than the first mode;determining a transform kernel for NSPT for the current block based on the angular mode and transform kernels associated with a second category’, the second category'- being different than the first category';performing, based on the transform kernel, an inverse transform on a transform coefficient block to generate a residual block associated with the current block; and reconstructing the current block based on the residual block,2. Tire method of claim 1, further comprising parsing information indicating whether a primary’ transform set or a secondary transform set is to be used for transform kernel selection, and w’herein determining the transform kernel comprises determining the transform kernel for NSPT for the current block based on the angular mode and the transform kernels associated with the second category’ based on the information indicating that the secondary’ transform set is to be used.

3. The method of claim 1, wherein the angular mode comprises a first angular mode, the method further comprising:determining a second angular mode for the current block; anddetermining a difference value between the first angular mode and the second angular mode.

4. The method of claim 3, wherein determining the transform kernel comprises determining the transform kernel for NSPT for the current block based on the angular 1616-600W001mode and transform kernels associated with the second category' in response to the difference value being less than a threshold.

5. The method of claim 3, wherein determining the transform kernel comprises determining the transform kernel for NSPT for the current block based on the angular mode and transform kernels associated with the second category in response to the difference value being greater than a threshold.

6. The method of claim 1, wherein determining the angular mode comprises determining the angular mode based on at least one of: parsing information indicating the angular mode or histogram of gradient (HoG) computed from neighboring samples of the current block.

7. The method of claim 1, wherein the first category is associated with templatebased intramode derivation (TIMD) mode, decoder side intramode derivation (DIMD) mode, extrapolation intra prediction (EIP) mode, matrix intra prediction (MIP) mode, and spatial geometric partitioning mode (SGPM), and wherein the first mode is one of the TIMD mode, DIMD mode, EIP mode, MIP mode, or SGPM,8. The method of claim 1, wherein the first category is associated with intra template matching prediction (IntraTMP) mode or inter-coding mode.

9. The method of claim 1, wherein the second category is associated with angular modes and a neural network (IntraNN) mode.

10. A device for decoding video data, the device comprising:one or more memories configured to store information indicative of which coding modes are associated with which category of a plurality of categories, and each category is associated with a respective set of transform kernels; andprocessing circuitry coupled to the one or more memories, the processing circuitry being configured to:determine that a current block has size for which non-separable primary’ transform (NSPT) is applied;1616-600W001determine that the current block is coded in a first mode associated with a first category of the plurality of categories;determine an angular mode for the current block, the angular mode being different than the first mode;determine a transform kernel for NSPT for the current block based on the angular mode and transform kernels associated with a second category, the second category’ being different than the first category;perform, based on the transform kernel, an inverse transform on a transform coefficient block to generate a residual block associated with the current block; andreconstruct the current block based on the residual block.

11. The device of claim 10, wherein the processing circuitry is configured to parse information indicating whether a primary transform set or a secondary transform set is to be used for transform kernel selection, and wherein to determine the transform kernel, the processing circuitry is configured to determine the transform kernel for NSPT for the current block based on the angular mode and the transform kernels associated with the second category based on the information indicating that the secondary transform set is to be used.

12. The device of claim 10, wherein the angular mode comprises a first angular mode, and wherein the processing circuitry is further configured to:determine a second angular mode for the current block; anddetermine a difference value between the first angular mode and the second angular mode.

13. The device of claim 12, wherein to determine the transform kernel, the processing circuitry is configured to determine the transform kernel for NSPT for the current block based on the angular mode and transform kernels associated with the second category in response to the difference value being less than a threshold.

14. The device of claim 12, wherein to determine the transform kernel, the processing circuitry is configured to determine the transform kernel for NSPT for the1616-600W001current block based on the angular mode and transform kernels associated with the second category in response to the difference value being greater than a threshold.

15. The device of claim 10, wherein to determine the angular mode, the processing circuitry is configured to determine the angular mode based on at least one of: parsing information indicating the angular mode or histogram of gradient (HoG) computed from neighboring samples of the current block.

16. The device of claim 10, wherein the first category is associated with templatebased intramode derivation (TIMD) mode, decoder side intramode derivation (DIMD) mode, extrapolation intra prediction (EIP) mode, matrix intra prediction (MIP) mode, and spatial geometric partitioning mode (SGPM), and wherein the first mode is one of the TIMD mode, DIMD mode, EIP mode, MIP mode, or SGPM.

17. The device of claim 10, wherein the first category is associated with intra template matching prediction (IntraTMP) mode or inter-coding mode.

18. The device of claim 10, wherein the second category is associated with angular modes and a neural network (IntraNN) mode.

19. A device for encoding video data, the device comprising:one or more memories configured to store information indicative of which coding modes are associated with which categories of a plurality of categories, and each category is associated with a respective set of transform kernels; andprocessing circuitry coupled to the one or more memories, the processing circuitry being configured to:determine that a current block has size for which non-separable primary transform (NSPT) is applied;determine that the current block is coded in a first mode associated w ith a first category of the plurality of categories;determine an angular mode for the current block, the angular mode being different than the first mode;1616-600W001determine a transform kernel for NSPT for the current block based on the angular mode and transform kernels associated with a second category, the second category being different than the first category;perform, based on the transform kernel, a transform on a residual block to generate a transform coefficient block associated with the current block; and signal information based on the transform coefficient block.

20. The device of claim 19, wherein the first category is associated with template-based intramode derivation (TIMD) mode, decoder side intramode derivation (DIMD) mode, extrapolation intra prediction (EIP) mode, matrix intra prediction (MIP) mode, and spatial geometric partitioning mode (SGPM), or associated with intra template matching prediction (IntraTMP) mode or inter-coding mode, and wherein the second category is associated with angular modes and a neural network (IntraNN) mode.1616-600W001

Images