Abstract: An inter-frame prediction method, a device, and a readable storage medium are provided. The method includes: determining a plurality of motion vector prediction candidates from a motion vector prediction candidate list of a target codec unit according to a template; performing offset processing on the plurality of motion vector prediction candidates to obtain a plurality of offset motion vector prediction candidates; determining a target motion vector prediction candidate according to the plurality of offset motion vector prediction candidates; and using the target motion vector prediction candidate as a search start point to obtain a target motion vector prediction through template matching, where the plurality of motion vector prediction candidates are a plurality of motion vector prediction candidates whose error between a current template and a reference template in the motion vector prediction candidate list is ranked in a first preset position.

Abstract: A loop filtering method and a terminal. The loop filtering method in the embodiments of this application includes: obtaining, by an electronic device, semantic block information of a picture to be filtered; determining, based on the semantic block information, at least one semantic block included in the picture to be filtered; and performing, for each semantic block by using a filter corresponding to the semantic block, loop filtering processing on a first reconstruction value corresponding to the semantic block, to obtain a second reconstruction value corresponding to the semantic block.

Abstract: An inter-frame prediction method and a terminal and pertains to the field of video encoding and decoding technologies. The inter-frame prediction method according to embodiments of this application includes: obtaining first motion information of a first block and second motion information of a second block; determining first and second prediction values corresponding to each pixel point in a first pixel region associated with the first block; and determining a target prediction value corresponding to each pixel point in a second pixel region of the first block based on target information; where the first block is a block to be encoded, and the second block is an encoded block; or the first block is a block to be decoded, and the second block is a decoded block.

Abstract: An image prediction method and device are disclosed. The method includes: obtaining a first reference unit of an image unit, where respective predicted images are obtained for the image unit and the first reference unit by using a same affine model; obtaining motion information of basic motion compensation units at two or more preset positions of the first reference unit; and obtaining motion information of a basic motion compensation unit of the image unit according to the motion information. In this way, motion information of the first reference unit using a same affine motion prediction model is reused, and a more accurate motion vector of a current image unit is obtained, improving prediction accuracy and maintaining encoding and decoding complexity, thereby improving encoding and decoding performance.

Abstract: A point cloud encoding and decoding method and apparatus, and pertains to the field of point cloud processing technologies. The point cloud encoding method in embodiments of this application includes: An encoder determines, based on attribute information of target point cloud, a target attribute information distribution characteristic value corresponding to a target bit rate point, where the target bit rate point is a bit rate point corresponding to an attribute quantization step of the attribute information; the encoder determines a target run-length encoding manner based on the target attribute information distribution characteristic value; and the encoder performs encoding processing on the attribute information based on the target run-length encoding manner.

Abstract: A point cloud encoding/decoding processing method and apparatus, an encoding device, and a decoding device. The point cloud encoding processing method according to the embodiments of this application includes: performing, by an encoding end, synchronous multi-branch tree division on a current frame to be encoded and an encoded frame; and performing, by the encoding end based on a target value corresponding to a first node in the encoded frame, inter prediction on an occupancy status of a second node in the frame to be encoded, where the occupancy status includes: occupied or unoccupied, where a position of the first node in the encoded frame is the same as a position of the second node in the frame to be encoded.

Abstract: This application pertains to the field of encoding and decoding technologies, and discloses a point cloud encoding method and apparatus, a point cloud decoding method and apparatus, and an electronic device. The point cloud encoding method in embodiments of this application includes: obtaining, by an encoder, a to-be-encoded node in a point cloud sequence and m encoded reference nodes in the point cloud sequence, where m is a positive integer; determining, by the encoder, a context of a to-be-encoded subnode based on occupancy information of the m reference nodes and a position of the to-be-encoded subnode in the to-be-encoded node, where the to-be-encoded subnode is any subnode obtained by partitioning the to-be-encoded node based on a structure tree; and performing, by the encoder, entropy encoding on the to-be-encoded subnode based on the context to generate a target bit stream.

Abstract: A point cloud coding method, a point cloud decoding method and a related device. The point cloud coding method in an example of the present disclosure includes: determining a distribution characteristic value, where the distribution characteristic value is configured to denote distribution characteristic information of to-be-coded attribute information; determining, based on the distribution characteristic value, a target order corresponding to the to-be-coded attribute information, where the target order is an order of exponential-Golomb coding; and entropy-coding the to-be-coded attribute information based on an exponential-Golomb coding algorithm of the target order, and obtaining an attribute code stream.

Abstract: This application discloses a point cloud attribute coding method, a point cloud decoding method, and a terminal. The point cloud attribute coding method according to an embodiment of this application includes: obtaining, by an encoder side, a to-be-coded point cloud, where the to-be-coded point cloud includes at least one type of attribute information; determining, by the encoder side, a target ordering manner of the to-be-coded point cloud based on an initial order corresponding to the to-be-coded point cloud or based on the initial order and density corresponding to the to-be-coded point cloud; and coding, by the encoder side, the to-be-coded point cloud according to the target ordering manner.

Abstract: This application discloses an attribute quantization and dequantization method and apparatus, and a device. The attribute quantization method in embodiments of this application includes: A first device determines a quantization correction factor based on target information, where the target information includes at least one of geometric information or attribute information of a target point cloud, and the quantization correction factor is used to perform correction on a target quantization step size. The first device performs correction on the target quantization step size based on the quantization correction factor, to obtain a corrected quantization step size, and performs quantization processing on the target point cloud based on the corrected quantization step size.

Abstract: An encoding control method includes obtaining a current to-be-processed node in a target queue, where the target queue includes nodes whose corresponding space blocks are occupied in a constructed tree structure; obtaining a first mode identifier of the current to-be-processed node; and in a case that the first mode identifier is a first identifier, if nodes that have been processed continuously meet a first condition, performing multiway tree occupancy encoding on the current to-be-processed node; where the first condition includes that: a mode identifier of the node that has been processed continuously is the first identifier and a number of nodes that have been processed continuously is less than a first threshold.

Abstract: An intra-frame prediction method and a terminal. The intra-frame prediction method includes: determining a prediction sample corresponding to a prediction block based on an angle prediction mode corresponding to the prediction block; and modifying the prediction sample by using a position dependent intra prediction combination PDPC, to generate a modified prediction sample, where the modifying the prediction sample by using a position dependent intra prediction combination includes: obtaining texture information of a reference sample set corresponding to a first angle prediction mode, the first angle prediction mode being an angle prediction mode corresponding to the prediction block, and the reference sample set including at least one reference sample; determining a second angle prediction mode according to the texture information; and modifying the prediction sample according to a target reference sample corresponding to the angle prediction mode.

Abstract: A point cloud attribute coding method, a point cloud attribute decoding method, and a terminal are provided. The point cloud attribute coding method includes: obtaining a to-be-coded point cloud; partitioning the to-be-coded point cloud into at least one to-be-coded point cloud block based on the number of to-be-coded points included in the to-be-coded point cloud, volume information of a bounding box corresponding to the to-be-coded point cloud, and an order of each to-be-coded point; for each to-be-coded point cloud block including at least two to-be-coded points, partitioning the to-be-coded point cloud block into N to-be-coded point cloud groups based on the number of to-be-coded points included in the to-be-coded point cloud block and a preset maximum transform order; performing transform coding and quantization processing on first residual information corresponding to a first target to-be-coded point cloud group, to obtain a quantized first transform coefficient.

Abstract: An image prediction method and device are disclosed. The method includes: obtaining a first reference unit of an image unit, where respective predicted images are obtained for the image unit and the first reference unit by using a same affine model; obtaining motion information of basic motion compensation units at two or more preset positions in the first reference unit; and obtaining motion information of a basic motion compensation unit of the image unit according to the motion information. In this way, motion information of the first reference unit using a same affine motion prediction model is reused, and a more accurate motion vector of a current image unit is obtained, improving prediction accuracy and maintaining encoding and decoding complexity, thereby improving encoding and decoding performance.

Abstract: An encoding control method includes obtaining a current to-be-processed node in a target queue, where the target queue includes nodes whose corresponding space blocks are occupied in a constructed tree structure; obtaining a first mode identifier of the current to-be-processed node; and in a case that the first mode identifier is a first identifier, if nodes that have been processed continuously meet a first condition, performing multiway tree occupancy encoding on the current to-be-processed node; where the first condition includes that: a mode identifier of the node that has been processed continuously is the first identifier and a number of nodes that have been processed continuously is less than a first threshold.

Abstract: The present disclosure provides a reference image encoding method, including: encoding sequence, to obtain a reconstructed image of a first and second reference images; determining whether a local area in the second reference image is suitable for updating first reference image; determining whether the first reference image is updated; when the first reference image is not updated, and it is determined the second reference image is suitable for updating the first reference image, replacing a pixel value of a corresponding area or a related area in the first reference image with a pixel value of the local area in the second reference image; and compiling, into the bitstream.

Abstract: A point cloud attribute information encoding method includes: obtaining first information; determining, based on second information associated with the first information, whether to perform DCT transform on K to-be-coded points; in a case that it is determined to perform DCT transform on the K to-be-coded points, performing DCT transform on the K to-be-coded points to obtain a transform coefficient of the K to-be-coded points; and quantizing the transform coefficient of the K to-be-coded points and performing entropy coding based on a quantized transform coefficient, to generate a binary bit stream. The first information includes the K to-be-coded points and the second information includes attribute prediction information for the K to-be-coded points; or the first information includes N coded points prior to the K to-be-coded points and the second information includes attribute reconstruction information for the N coded points.

Abstract: A point cloud encoding processing method includes obtaining a current to-be-encoded node in a target queue, where the target queue includes an occupied node in a space block corresponding to a constructed tree structure based on first geometric information, the first geometric information is obtained by preprocessing based on geometric information of a to-be-encoded N-th frame point cloud, and N is an integer greater than 1; and determining an encoding mode of the current to-be-encoded node based on whether the current to-be-encoded node meets an encoding condition of an isolated point. The encoding condition of the isolated point includes an occupancy status of a target node meets a preset condition. The target node is a node that is in encoded nodes and that is associated with the current to-be-encoded node.

Abstract: This application discloses an entropy encoding method and apparatus, and an entropy decoding method and apparatus. The entropy encoding method in the embodiments of this application includes: obtaining, by an entropy encoding apparatus, sparsity/density information of a to-be-encoded target point cloud; determining, based on the sparsity/density information, a type of an occupancy code context model used to perform entropy encoding on the target point cloud; and performing entropy encoding on the target point cloud based on the type of the occupancy code context model.

Abstract: This application discloses a point cloud encoding processing method and apparatus and a point cloud decoding processing method and apparatus. The point cloud encoding processing method of embodiments of this application includes: determining sparsity level information of a target point cloud; and in a case that the sparsity level information indicates the target point cloud being a sparse point cloud and a node to be encoded corresponding to the target point cloud satisfies an isolated point encoding condition, enabling an isolated point encoding mode.