Point cloud encoding method, point cloud decoding method, and device

Abstract

The present application relates to the field of computers, and discloses a point cloud encoding method, a point cloud decoding method, and a device. The point cloud encoding method comprises: when the current node to be encoded in a transform tree corresponding to point cloud to be encoded allows for intra-frame prediction, acquiring initial prediction weights for a first child node from neighboring nodes, and acquiring reference attribute values of the neighboring nodes and the first child node, the first child node being any child node of the current node to be encoded; on the basis of the reference attribute values of the neighboring nodes and the first child node, determining similarity information between the neighboring nodes and the first child node, and correcting the initial prediction weights on the basis of the similarity information to obtain final prediction weights for the first child node from the neighboring nodes; and predicting the first child node on the basis of the final prediction weight for the first child node from each neighboring node.

Description

Point cloud encoding methods, point cloud decoding methods and devices Cross-references to related applications This application claims priority to Chinese Patent Application No. 202510044069.2, filed on January 10, 2025, entitled “Point Cloud Encoding Method, Point Cloud Decoding Method and Apparatus”, the entire contents of which are incorporated herein by reference. Technical Field This application belongs to the field of computer technology, specifically relating to a point cloud encoding method, a point cloud decoding method, and a device. Background Technology Point clouds are a representation of three-dimensional objects or scenes, consisting of a set of randomly distributed discrete points in space that express the spatial structure and surface properties of the object or scene. To accurately reflect spatial information, a considerable number of discrete points are required. To reduce the bandwidth consumed during point cloud data storage and transmission, point cloud data needs to be encoded and compressed. Point cloud data typically consists of geometric information describing the location, such as three-dimensional coordinates (x, y, z), and attribute information for that location, such as color (R, G, B) or reflectivity. A common point cloud compression technique is geometry-based point cloud compression (GPCC). When compressing point clouds using GPCC, the encoding of geometric information and attribute information is performed separately. First, the geometric information is encoded, and then the attribute information is encoded using the reconstructed geometric information. There are two methods for encoding attribute information: one is level-of-detail prediction and boosting transform, and the other is region-adaptive hierarchical transform (RAHT). Currently, in the process of encoding the attribute information of intra-frame nodes using RAHT, prediction is mainly based on the spatial correlation between nodes, which has limited improvement in coding efficiency and needs to be improved. Summary of the Invention This application provides a point cloud encoding method, a point cloud decoding method, and a device, which can further improve encoding efficiency. Firstly, a point cloud encoding method is provided, which includes: If intra-frame prediction is allowed for the current node to be encoded in the transform tree corresponding to the point cloud to be encoded, the initial prediction weights of the neighboring nodes for the first child node are obtained, and the reference attribute values ​​of the neighboring nodes and the reference attribute values ​​of the first child node are obtained, wherein the first child node is any child node of the current node to be encoded. Based on the reference attribute values ​​of the neighboring nodes and the reference attribute values ​​of the first child node, the similarity information between the neighboring nodes and the first child node is determined, and the initial prediction weight is corrected based on the similarity information to obtain the final prediction weight of the neighboring nodes for the first child node. The similarity information is used to characterize the degree of similarity. The first child node is predicted based on the final prediction weights of each of the neighboring nodes for the first child node. Secondly, a point cloud decoding method is provided, which includes: If intra-frame prediction is allowed for the current node to be decoded in the transformation tree corresponding to the point cloud to be decoded, the initial prediction weights of the neighboring nodes for the first child node are obtained, and the reference attribute values ​​of the neighboring nodes and the reference attribute values ​​of the first child node are obtained, wherein the first child node is any child node of the current node to be decoded. Based on the reference attribute values ​​of the neighboring nodes and the reference attribute values ​​of the first child node, the similarity information between the neighboring nodes and the first child node is determined, and the initial prediction weight is corrected based on the similarity information to obtain the final prediction weight of the neighboring nodes for the first child node. The similarity information is used to characterize the degree of similarity. The first child node is predicted based on the final prediction weights of each of the neighboring nodes for the first child node. Thirdly, a point cloud encoding device is provided, the device comprising: The acquisition module is used to acquire the initial prediction weights of the neighboring nodes to the first child node in the transform tree corresponding to the point cloud to be encoded, provided that intra-frame prediction is allowed for the current node to be encoded; and to acquire the reference attribute values ​​of the neighboring nodes and the reference attribute values ​​of the first child node, wherein the first child node is any child node of the current node to be encoded. The weight correction module determines the similarity information between the neighbor node and the first child node based on the reference attribute values ​​of the neighbor node and the reference attribute values ​​of the first child node, and corrects the initial prediction weight based on the similarity information to obtain the final prediction weight of the neighbor node for the first child node. The similarity information is used to characterize the degree of similarity. The intra-frame prediction module is used to predict the first child node based on the final prediction weights of each of the neighboring nodes for the first child node. Fourthly, a point cloud decoding device is provided, the device comprising: The acquisition module is used to acquire the initial prediction weights of the neighboring nodes to the first child node in the transform tree corresponding to the point cloud to be decoded, provided that intra-frame prediction is allowed for the current node to be decoded; and to acquire the reference attribute values ​​of the neighboring nodes and the reference attribute values ​​of the first child node, wherein the first child node is any child node of the current node to be decoded. The weight correction module is used to determine the similarity information between the neighbor node and the first child node based on the reference attribute values ​​of the neighbor node and the reference attribute values ​​of the first child node, and to correct the initial prediction weight based on the similarity information to obtain the final prediction weight of the neighbor node for the first child node. The similarity information is used to characterize the degree of similarity. The intra-frame prediction module is used to predict the first child node based on the final prediction weights of each of the neighboring nodes for the first child node. Fifthly, a point cloud encoding apparatus is provided, the apparatus being configured to perform the steps of the method described in the first aspect. In a sixth aspect, a point cloud decoding apparatus is provided, the apparatus being configured to perform the steps of the method described in the second aspect. In a seventh aspect, a terminal is provided, the terminal including a processor and a memory, the memory storing a program or instructions executable on the processor, the program or instructions, when executed by the processor, implementing the steps of the method as described in the first aspect. Eighthly, a terminal is provided, the terminal including a processor and a memory, the memory storing a program or instructions executable on the processor, the program or instructions, when executed by the processor, implementing the steps of the method as described in the second aspect. A ninth aspect provides a terminal, including a processor and a communication interface, wherein the processor is configured to, when intra-frame prediction is allowed for the current node to be encoded in the transform tree corresponding to the point cloud to be encoded, obtain initial prediction weights of neighboring nodes for a first child node, and obtain reference attribute values ​​of the neighboring nodes and the first child node, wherein the first child node is any child node of the current node to be encoded; determine similarity information between the neighboring nodes and the first child node based on the reference attribute values ​​of the neighboring nodes and the first child node, and correct the initial prediction weights based on the similarity information to obtain the final prediction weights of the neighboring nodes for the first child node, wherein the similarity information is used to characterize the degree of similarity; and predict the first child node based on the final prediction weights of each of the neighboring nodes for the first child node. In a tenth aspect, a terminal is provided, including a processor and a communication interface, wherein the processor is configured to, when intra-frame prediction is allowed for the current node to be decoded in the transform tree corresponding to the point cloud to be decoded, obtain initial prediction weights of neighboring nodes for a first child node, and obtain reference attribute values ​​of the neighboring nodes and the first child node, wherein the first child node is any child node of the current node to be decoded; determine similarity information between the neighboring nodes and the first child node based on the reference attribute values ​​of the neighboring nodes and the first child node, and correct the initial prediction weights based on the similarity information to obtain the final prediction weights of the neighboring nodes for the first child node, wherein the similarity information is used to characterize the degree of similarity; and predict the first child node based on the final prediction weights of each neighboring node for the first child node. Eleventhly, a readable storage medium is provided, on which a program or instructions are stored, which, when executed by a processor, implement the steps of the method described in the first aspect, or implement the steps of the method described in the second aspect. In a twelfth aspect, a wireless communication system is provided, comprising: an encoding end device and a decoding end device, wherein the encoding end device is configured to perform the steps of the method described in the first aspect, and the decoding end device is configured to perform the steps of the method described in the second aspect. In a thirteenth aspect, a chip is provided, the chip including a processor and a communication interface coupled to the processor, the processor being configured to run programs or instructions to implement the method as described in the first aspect, or to implement the method as described in the second aspect. In a fourteenth aspect, a computer program / program product is provided, which is stored in a storage medium and is executed by at least one processor to implement the steps of the two methods as described in the first aspect. In this embodiment, if intra-frame prediction is allowed for the current node to be encoded in the transform tree corresponding to the point cloud to be encoded, the initial prediction weights of neighboring nodes for the first child node are obtained, along with the reference attribute values ​​of the neighboring nodes and the first child node, where the first child node is any child node of the current node to be encoded. Based on the reference attribute values ​​of the neighboring nodes and the first child node, similarity information between the neighboring nodes and the first child node is determined, and the initial prediction weights are corrected based on the similarity information to obtain the final prediction weights of the neighboring nodes for the first child node. The first child node is then predicted based on these final prediction weights. Since the correlation between the reference attribute values ​​of the neighboring nodes and the current child node is utilized when determining the prediction weights of the neighboring nodes of the current child node to be encoded (the first child node), the obtained prediction weights are more accurate, thus further improving compression efficiency. Attached Figure Description Figure 1 is a schematic diagram of the encoder framework of GPCC in related technologies. Figure 2 is a schematic diagram of the GPCC decoder framework in related technologies. Figure 3 is a schematic diagram of the process for determining whether the current child node to be encoded should undergo intra-frame prediction in related technologies. Figure 4 is a schematic diagram of the node relationships in the transformation tree corresponding to the point cloud in the related technology. Figure 5 is a schematic diagram of the binary RAHT decomposition of a node block in a related technology. Figure 6 is a flowchart illustrating a point cloud encoding method proposed in an embodiment of this application. Figure 7 is a flowchart illustrating a point cloud decoding method proposed in an embodiment of this application. Figure 8A is a schematic diagram illustrating the effect of a point cloud encoding method proposed in an embodiment of this application. Figure 8B is a schematic diagram of the effect of a point cloud encoding method proposed in an embodiment of this application. Figure 8C is a schematic diagram of the effect of a point cloud encoding method proposed in an embodiment of this application. Figure 8D is a schematic diagram of the effect of a point cloud encoding method proposed in an embodiment of this application. Figure 9 is a schematic diagram of the structure of a point cloud encoding device proposed in an embodiment of this application. Figure 10 is a schematic diagram of the structure of a point cloud decoding device proposed in an embodiment of this application. Figure 11 is a schematic diagram of the structure of a communication device proposed in an embodiment of this application. Figure 12 is a schematic diagram of the structure of a terminal proposed in an embodiment of this application. Detailed Implementation The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application. The terms "first," "second," etc., used in this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such terms can be used interchangeably where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first" and "second" are generally of the same class, not limited in number; for example, the first object can be one or more. Furthermore, "or" in this application indicates at least one of the connected objects. For example, the scope of protection for "A or B" covers at least three scenarios: Scenario 1: including A but not B; Scenario 2: including B but not A; Scenario 3: including both A and B. In addition, the terms "A and / or B," "at least one of A and B," and "at least one of A or B" also cover at least the above three scenarios. The character " / " generally indicates that the preceding and following objects are in an "or" relationship. The term "instruction" in this application can be either a direct instruction (or explicit instruction) or an indirect instruction (or implicit instruction). A direct instruction can be understood as one in which the sender explicitly informs the receiver of specific information, the operation to be performed, or the requested result, etc., in the instruction sent. An indirect instruction can be understood as one in which the receiver determines the corresponding information based on the instruction sent by the sender, or makes a judgment and determines the operation to be performed or the requested result, etc., based on the judgment result. The encoding / decoding end corresponding to the encoding / decoding method in this application embodiment can be a terminal, which can also be called a terminal device or user equipment (UE). The terminal can be a mobile phone, tablet personal computer, laptop computer or notebook computer, personal digital assistant (PDA), handheld computer, netbook, ultra-mobile personal computer (UMPC), mobile internet device (MID), augmented reality (AR) / virtual reality (VR) device, robot, wearable device or vehicle-mounted device (VUE), pedestrian terminal (PUE), etc. Wearable devices include: smartwatches, bracelets, headphones, glasses, etc. It should be noted that the specific type of terminal is not limited in this application embodiment. For ease of understanding, the relevant technologies involved in the embodiments of this application will be described below. As shown in Figure 1, in Geometry-based Point Cloud Compression (GPCC), the geometric information and attribute information of the point cloud are encoded separately. First, the geometric information is encoded, and then the attribute information is encoded using the reconstructed geometric information. The geometric information encoding process may include: performing coordinate transformation on the geometric information so that the entire point cloud is contained within a bounding box; then performing coordinate quantization (voxelation), where quantization mainly serves a scaling function. Since quantization rounds the geometric coordinates, some points have the same geometric information, which are called duplicate points. Whether to remove duplicate points is determined based on parameters. The two steps of quantization and removal of duplicate points are also called the voxelization process. Next, the bounding box is partitioned into a multi-way tree, such as an octree, quadtree, or binary tree. In the multi-way tree-based geometric information encoding framework, the bounding box is divided into eight equal sub-cubes. The non-empty sub-cubes are further partitioned until a unit cube with leaf nodes of 1x1x1 is obtained. The number of points in the leaf nodes is encoded to generate a binary code stream. After geometric encoding is completed, the geometric information is reconstructed for subsequent recoloring. Attribute encoding mainly targets color and reflectance information. The attribute encoding process may include: first, determining whether to perform color space conversion based on parameters; if so, converting the color information from the Red Green Blue (RGB) color space to the Luminance Color (YUV) color space. Then, recoloring the geometrically reconstructed point cloud using the original point cloud, so that the unencoded attribute information corresponds to the reconstructed geometric information. In color information encoding, after sorting the point cloud using Morton codes, the nearest neighbors of the point to be predicted are searched using geometric spatial relationships. The reconstructed attribute values ​​of the found neighbors are used to predict the predicted attribute value of the point to be predicted. Then, the original attribute value and the predicted attribute value are differencing to obtain the prediction residual. Finally, the prediction residual is quantized and encoded to generate a binary code stream. As shown in Figure 2, in GPCC, the geometric bit stream (also known as the geometric code stream) is first decoded to obtain the reconstructed geometric information. Then, a transform tree is constructed based on the reconstructed geometric information to assist in the decoding of the attribute bit stream and obtain the reconstructed attribute information. Finally, the reconstructed geometric information and the reconstructed attribute information are superimposed to output the point cloud. There are two encoding methods for attribute information: prediction and boosting transformation based on level of detail, and region adaptive hierarchical transformation (RAHT). RAHT, an attribute coding method, can be divided into intra-frame and inter-frame RAHT based on its prediction method. Intra-frame prediction uses only the encoded neighboring nodes of the current frame to predict the current node, while inter-frame prediction uses nodes from a reference frame to predict the current node. The embodiments in this application are related to intra-frame prediction; therefore, the following mainly introduces the intra-frame RAHT technology in related technologies. Intra-frame RAHT (Region Adaptive Transformation) is a technique based on upsampling prediction, performed layer by layer from top to bottom. In each layer, RAHT is applied to each cell node containing 2×2×2 child nodes. When encoding a cell node with 8 child nodes in the current layer, the attributes of the placeholder child node are weighted and predicted using its parent node (upper layer), coplanar neighbor parent node (upper layer), collinear neighbor parent node (upper layer), and already encoded coplanar and collinear neighbor child nodes (same layer). The predicted attribute value is then subtracted from the original attribute value (true attribute value), and the RAHT transformation is performed to obtain the AC coefficient residual. This AC coefficient residual is then quantized and encoded. However, if the number of child nodes to be encoded, the number of their grandparent nodes, and the number of their parent nodes do not meet certain conditions, intra-frame prediction is not enabled, and the RAHT transformation is directly performed on the original attribute value of the current node to be encoded, followed by quantization and encoding of the transformation coefficients. The following is the specific process of intra-frame RAHT technology: The first step is to construct the transformation tree. Starting from the bottom layer, an octree structure is built from the bottom up. During the construction of the transformation tree, corresponding Morton code information, attribute information, and weight information need to be generated for the merged nodes. The second step is a top-down parse tree. Starting from the root node, upsampling prediction and Region Adaptive Hierarchical Transformation (RAHT) are performed layer by layer from top to bottom. If the current node to be encoded is the root node, then no upsampling prediction is performed. Instead, the original attribute information of the node is directly transformed by RAHT to obtain the DC coefficient and AC coefficient. If the current node to be encoded is not the root node, it is necessary to determine whether its eight child nodes should be predicted. The determination process is shown in Figure 3. 1. Determine if the number of placeholder child nodes NumValidc is equal to 1: If the number of placeholder child nodes (non-empty) of the current node to be encoded is 1, then set the number of its neighboring parent nodes NumValidP to val (e.g., 19), do not perform prediction, directly perform RAHT transformation on the original attribute information of the current node to be encoded, and then quantize and entropy encode the obtained AC coefficients.

[0073] Otherwise, continue to step 2. 2. Determine if the number of neighboring grandparent nodes NumValidGP is greater than or equal to TH1 (e.g., 2): If the number of neighboring grandparent nodes (including grandparent nodes) of the current child node to be encoded is less than 2, then no prediction is performed. The original attribute information of the current node is directly subjected to RAHT transformation, and then the resulting AC coefficients are quantized and entropy encoded. Otherwise, proceed to step 3. 3. Neighbor Search. The search scope includes: the parent node of the current child node to be encoded (1 node), the coplanar neighbors of the parent node of the current child node to be encoded (6 nodes), the collinear neighbors of the parent node of the current child node to be encoded (12 nodes), the coplanar neighbors of the current child node to be encoded (6 nodes), and the collinear neighbors of the current child node to be encoded (12 nodes). The above neighbors are searched sequentially. If a neighbor exists, its corresponding index information is recorded, along with the number of neighbors of the parent node (including the parent node itself). Then proceed to step 4. 4. Determine if the number of neighboring parent nodes is greater than or equal to TH2 (e.g., 6): If the number of neighboring parent nodes of the current child node to be encoded is less than TH2, then no weighted prediction is performed. Instead, the original attribute information of the current node to be encoded is directly subjected to RAHT transformation, and then the resulting AC coefficients are quantized and entropy encoded. Otherwise, weighted prediction is performed. The third step is weighted prediction. The nearest neighbors found in the neighbor search are used to perform weighted prediction on each child node of the node to be encoded. The prediction weight of the parent node can be set to 9, the prediction weight of the neighboring child node that is coplanar with the current child node to be encoded is 5, the prediction weight of the neighboring child node that is collinear with the current child node to be encoded is 2, the prediction weight of the neighboring parent node that is coplanar with the current child node to be encoded is 3, and the prediction weight of the neighboring parent node that is collinear with the current child node to be encoded is 1. In this context, the parent node can be directly used to predict each child node of the current block to be encoded. For other neighbor nodes (neighbor parent nodes and already encoded neighbor child nodes), a step is needed to determine whether they can be used to predict the child nodes of the current block. The steps for this determination are as follows: 1. As shown in Figure 4, for the neighboring parent nodes with indices 1-7, since their child nodes have not yet been encoded, the prediction can only be made using the neighboring parent nodes. First, check if the neighboring parent node at this position exists. If it does not exist, continue to check the next neighboring parent node. If such a neighboring parent node exists, it is necessary to determine whether it can be used for prediction, and to filter neighboring parent nodes. The specific process is as follows: 1) Two prediction thresholds are set based on the parent node's attribute values ​​to further filter nearest neighbors, eliminating unreasonable neighbor parent nodes to improve prediction accuracy. Let these two thresholds be limitLow and limitHigh, and let the parent node's attribute value be attrPar, then: limitLow = attrpar × 2 limitHigh = attrpar × 25 Suppose the attribute value of the current neighbor's parent node is attrNei, and perform the following judgment on it: limitLow<attrnei×10<limitHigh

[0088] If this condition is not met, the current neighboring parent node cannot be used to predict the child nodes of the current block to be encoded; if this condition is met, the following judgment is performed. 2) Determine whether the current neighboring parent node satisfies the condition that it is coplanar and collinear with each of the current child nodes to be encoded. If this condition is not met, the current neighboring nodes cannot be used to perform weighted prediction on the current child nodes to be encoded; if this condition is met, the current neighboring nodes are used to perform weighted prediction on the current child nodes to be encoded. 2. For neighboring parent nodes with indices 8-19, since their child nodes have already been encoded, if there are neighboring child nodes at the same position, they can be used to replace their respective neighboring parent nodes for prediction, resulting in better prediction performance. First, check if the neighboring parent node at this position exists. If it does not exist, continue to check the next neighboring parent node. If such a neighboring parent node exists, it is necessary to determine whether it can be used for prediction, and to filter neighboring parent nodes. The specific process is as follows: 1) Two prediction thresholds are set based on the attribute values ​​of the parent node to further filter nearest neighbors, eliminating unreasonable neighbor parent nodes to improve prediction accuracy. Let the attribute value of the current neighbor parent node be attrNei, and the following judgments be made on it: limitLow<attrnei×10<limitHigh

[0095] If this condition is not met, the current neighboring parent node cannot be used to predict the child nodes of the current block to be encoded; if this condition is met, the following judgment is made: 2) Determine if there are any child nodes on the same floor at the same position of the current parent node's neighbors: For coplanar neighbor parent nodes, if one of their child nodes is also coplanar with the current child node to be predicted, then use that coplanar neighbor child node instead of the coplanar neighbor parent node for weighted prediction of the child node to be predicted; for collinear neighbor parent nodes, if one of their child nodes is also collinear with the current child node to be predicted, then use that collinear neighbor child node instead of the collinear neighbor parent node for weighted prediction. If no such child node exists, continue using the current neighbor parent node for weighted prediction. 3. Finally, each child node of the current node to be encoded uses its neighboring nodes that meet the conditions as a set of reference points to perform weighted prediction and obtain the attribute prediction value. The fourth step is the RAHT transformation, and the process is as follows: If no prediction is performed on the current block, only the original attribute values ​​need to be transformed using RAHT; if intra-frame prediction is performed on the current block, then the intra-frame prediction residuals of the attributes need to be transformed using RAHT. Before performing the RAHT transformation, the original attribute values ​​and the predicted attribute values ​​(if prediction is performed) need to be normalized first, and then the RAHT transformation is performed on the processed values. First, the original attribute values ​​are normalized. Let the original attribute value of the current child node be A. i The weight is w i (The size is the number of points contained in the current child node), then If prediction is used, let the weighted predicted value of the current child node's attributes be A. pi ,but The predicted value of the processed attribute A′ pi and the original attribute value A′ i After subtraction (subtraction is performed if prediction is enabled, otherwise it is not), RAHT transformation is performed. For a 2×2×2 node, the transformation is performed along three directions, with four transformations in each direction. The specific transformation flowchart is shown in Figure 5: 1) Perform the transformation along the first direction to obtain the low-frequency L node and the high-frequency H node; 2) Transform the L and H nodes along the second direction to obtain the low-frequency LL node and the high-frequency LH, HL, and HH nodes; 3) Transform the LL, LH, HL, and HH nodes along the third direction to obtain the low-frequency LLL node and the high-frequency LLH, LHL, LHH, HLL, HLH, HHL, and HHH nodes. Where LLL is the DC coefficient, and LLH, LHL, LHH, HLL, HLH, HHL, and HHH are the AC coefficients. The DC and AC coefficients of the root node are quantized and entropy encoded in the order of (LLL)0, LLH(4), LHL(2), HLL(1), LHH(6), HLH(5), HHL(3), HHH(7). For the remaining nodes in the remaining layers, only the AC coefficients are quantized and entropy encoded in the order of LLH(4), LHL(2), HLL(1), LHH(6), HLH(5), HHL(3), HHH(7). Due to the sparsity of point clouds, each 2×2×2 node block typically occupies fewer than 8 child nodes. Therefore, not all AC coefficients will exist, and non-existent AC coefficients will not be encoded. Specifically, when performing a two-point transformation, assume the input attribute values ​​are T. 01 T11 The transformed coefficients T1, It is obtained from the following transformation formula: Where a and b are calculated from the weights of the two current transformed child nodes (the weight is the number of points contained in the child node – Question 9, how to understand the number of points contained in a child node?). If the current node is the root node, no prediction is performed. Instead, the DC and AC coefficients of the original attribute values ​​are quantized and entropy encoded. If the current node is not the root node and no prediction is performed, the AC coefficients of the original attribute values ​​are quantized and entropy encoded after transformation. If the current node is not the root node and prediction is performed, then the residuals of the AC coefficients are quantized and entropy encoded. The coefficients are decoded and dequantized at the decoding end. Then, an inverse RAHT transform is performed. The inverse RAHT transform is the reverse process of the RAHT transform, and the formula for the inverse transform is as follows: Among them, T1, T represents the transformation coefficients. 01 T 11 The calculation methods for a and b are the same during the reconstruction of attribute values. If the current node is the root node, then perform the inverse transformation directly to obtain the attribute reconstruction values ​​of each child node. If the current node is not the root node and no prediction is performed, the DC coefficients inherit the attribute reconstruction values ​​of the parent node, and then perform an inverse Raht transformation with the reconstructed AC coefficients to obtain the attribute reconstruction values ​​of each child node of the current node. If the current node is not the root node and prediction is performed, the DC coefficients inherit the attribute reconstruction values ​​of the parent node and are subtracted from the attribute values ​​of the parent node predicted within the frame to obtain the DC coefficient residuals. Then, they are inversely transformed together with the reconstructed AC coefficient residuals to obtain the attribute prediction residuals of each child node of the current node. Finally, the attribute prediction values ​​are added to obtain the attribute reconstruction values ​​of each child node. In the GPCC attribute encoding process of related technologies, only the correlation between the reconstructed temporal and spatial domains of the current attribute is utilized to improve compression efficiency when encoding the current attribute. When point cloud data contains more than one type of attribute information, there is a certain correlation between different types of attribute information, which is currently not utilized, so the encoding efficiency needs to be further improved. The point cloud encoding and decoding method proposed in this application introduces reference attribute information to correct the prediction weights in RAHT intra-frame prediction, thereby correcting the prediction attribute values, which can improve prediction accuracy and thus improve the encoding and decoding efficiency of point clouds. In related technologies, each type of attribute information is encoded individually. For each attribute type, a transformation tree structure is constructed, and the encoding is performed layer by layer and node by node from the root node level to the leaf node level. For each level, after encoding and reconstructing a node, at most eight reconstructed attribute values ​​(its child nodes) are obtained. After encoding and reconstructing all nodes in the current level, the reconstructed attribute values ​​of all child nodes in the current level are obtained. Then, the nodes in the next level are encoded and reconstructed. After all nodes in the next level are encoded and reconstructed, the reconstructed attribute values ​​of all child nodes in the next level are obtained. This process continues until the leaf node level is reached. In this embodiment, however, a reference attribute reconstruction value (hereinafter referred to as the reference attribute value) is introduced for each node in each level. In some embodiments, the reference attribute value of a node is the reconstructed attribute value obtained when encoding the reference attribute of the node, which is then stored as the reference attribute value of the node. It can be understood that when the attribute to be encoded is the first attribute, the reference attribute is one or more of the remaining attributes besides the first attribute. Accordingly, in the following embodiments, the reference attribute value of a neighboring node can be the reference attribute value of the first attribute of the neighboring node, the reference attribute value of the first child node can be the reference attribute value of the first attribute of the first child node, the original attribute value of the first child node can be the original attribute value of the first attribute of the first child node, and the intra-frame predicted attribute value of the first child node can be the intra-frame predicted attribute value of the first attribute of the first child node, wherein the first child node can be any child node of the node currently to be encoded. The point cloud encoding method and point cloud decoding method provided in this application will be described in detail below with reference to the accompanying drawings and through some embodiments and application scenarios. As shown in Figure 6, a point cloud encoding method proposed in one embodiment of this application may include: Step 601: If intra-frame prediction is allowed for the current node to be encoded in the transform tree corresponding to the point cloud to be encoded, obtain the initial prediction weights of the neighboring nodes for the first child node, and obtain the reference attribute values ​​of the neighboring nodes and the reference attribute values ​​of the first child node, wherein the first child node is any child node of the current node to be encoded. Optionally, before step 601, a transformation tree corresponding to the point cloud to be encoded can be generated based on the geometric information of the point cloud to be encoded. Specifically, the point cloud is reordered based on its geometric coordinates, and a P-layer transformation tree is constructed from the reordered point cloud data. A bottom-up construction method can be used, with the bottom layer containing all nodes and the top layer being the root node layer, containing only one node. Optionally, during the construction of the transformation tree, corresponding Morton code information, attribute information, and weight information are generated for the node after the current attribute merging. In some embodiments, step 601 occurs in the second step of the RAHT intra-prediction process described above, so it can be determined in the manner of the second step whether the current node to be encoded is allowed to perform intra-prediction. In some embodiments, the initial prediction weights of the neighbor nodes to the first child node obtained in step 601 may be prediction weights obtained in related technologies based on the temporal and spatial correlations of the reconstructed neighbor nodes. In some embodiments, assuming the attribute to be encoded is the first attribute, the reference attribute value of the neighboring node obtained in step 601 can be the reference attribute value of the first attribute of the neighboring node, and the reference attribute value of the first child node can be the reference attribute value of the first attribute of the first child node. The reference attribute can be one or more attributes other than the first attribute. Step 602: Based on the reference attribute values ​​of the neighboring nodes and the reference attribute values ​​of the first child node, determine the similarity information between the neighboring nodes and the first child node, and correct the initial prediction weights based on the similarity information to obtain the final prediction weights of the neighboring nodes for the first child node. The similarity information is used to characterize the degree of similarity. In some embodiments, a reference attribute of a node may include at least one channel. For example, if the reference attribute of a node is color, it may include three channels: R, G, and B; if the reference attribute of a node is reflectivity, it may include only one channel. In some embodiments, the similarity information between the neighbor node and the first child node may include: the error between the reference attribute values ​​of the neighbor node and the first child node, wherein determining the similarity information between the neighbor node and the first child node based on the reference attribute values ​​of the neighbor node and the first child node may include: determining the error between the neighbor node and the first child node in at least one channel based on the reference attribute values ​​of the neighbor node and the first child node in at least one channel, wherein one channel corresponds to one error, the larger the error, the smaller the similarity, and the smaller the error, the greater the similarity. It is understood that the error between the reference attribute values ​​of the neighboring node and the first child node can be used to quickly and accurately characterize the similarity between the two, so that the correlation performance between the two can better influence the final prediction weight of the neighboring node for the first child node, thereby improving the accuracy of the final prediction weight and thus improving the coding efficiency. In one example, the reference attribute value of the neighbor node can be set to `neibor_ref`. When the reference attribute value of the first child node is `curChild_ref`, the initial prediction weight of the neighbor node for the first child node is `weight_org`, and the attribute value of the neighbor node is `neibor_cur`. The step of determining the error between the neighbor node and the first child node in the at least one channel based on the reference attribute values ​​of the neighbor node and the first child node in the at least one channel includes: Based on the first formula, the reference attribute values ​​of the neighbor node and the first child node under the at least one channel, determine the error between the neighbor node and the first child node under the at least one channel; The first formula is: error=abs(neibor_ref-curChild_ref) Here, error represents the error; the larger the error, the worse the similarity, and vice versa. In some embodiments, correcting the initial prediction weights based on the similarity information to obtain the final prediction weights of the neighbor nodes for the first child node may include: A first scaling ratio is determined for the initial prediction weights based on the similarity information, wherein the higher the similarity, the larger the first scaling ratio, and the lower the similarity, the smaller the first scaling ratio. The initial prediction weights are corrected according to the first scaling ratio to obtain the final prediction weights of the neighbor nodes for the first child node. Optionally, the reference attributes of each node include at least one channel, wherein determining the first scaling factor for the initial prediction weights based on the similarity information may include: 1) Obtain N pre-set thresholds: th1, th2, th3, ..., thN, where the N thresholds are N positive integers in ascending order; 2) Based on the second formula, the pre-set N thresholds, and the reference attribute values ​​of the first child node under the at least one channel, determine the N reference attribute thresholds corresponding to each of the at least one channel; Wherein, the N thresholds are N positive integers that increase sequentially, and the second formula is: refAttrThi=curChild_ref×thi Where refAttrThi represents the threshold of the i-th reference attribute, curChild_ref represents the reference attribute value of the first child node, and thi represents the i-th threshold among the N pre-set thresholds, i = 1, 2, ..., N. 3) Based on the N reference attribute thresholds corresponding to each of the at least one channel, determine the M reference attribute threshold intervals corresponding to each of the at least one channel, where M is a positive integer and M can be less than, equal to or greater than N. For example, the following M reference attribute threshold intervals can be determined: [0, refAttrTh1], [refAttrTh1, refAttrTh2], ..., [refAttrThN-1, refAttrThN], [refAttrThN, ∞]. It is easy to see that at this time, M = N+1. In this configuration, a reference attribute threshold interval within a single channel corresponds to a scaling factor. Optionally, M reference attribute threshold intervals within a single channel correspond to M scaling factors, and these M scaling factors can be the same or different. For example, the correspondence between reference attribute threshold intervals and scaling factors can be: [0,refAttrTh1]:scale=8 [refAttrTh1,refAttrTh2]:scale=7 … [refAttrThN-1,refAttrThN]:scale=6 [refAttrThN,∞]:scale=1 It is understandable that by setting a reasonable reference attribute threshold range and the corresponding scaling ratio, the first scaling ratio for the initial prediction weight can be accurately determined, resulting in an accurate final prediction weight, thereby improving the accuracy of the final prediction weight and thus improving coding efficiency. Optionally, determining the first scaling ratio for the initial prediction weights based on the similarity information may further include: 4) Multiply the error in the at least one channel by a first factor, wherein the first factor is used to ensure the accuracy of the error. For example, multiply the error by 10, that is: error = 10 × error It should be noted that the purpose of the above four steps is to simulate the process of calculating the relative error between the neighbor node and the first child node: abs(neibor_ref-curChild_ref) / curChild_ref. This formula can calculate the relative error between the reference attribute of the neighbor node and the reference attribute of the first child node, reflecting the similarity between the two, and making the error as integer as possible to facilitate calculation, thereby further improving coding efficiency. Optionally, determining the first scaling ratio for the initial prediction weights based on the similarity information may further include: From the M reference attribute threshold intervals corresponding to each of the at least one channel, determine the first reference attribute threshold interval where the error under the at least one channel is located, and obtain at least one first reference attribute threshold interval, wherein the larger the error, the lower the similarity, and the smaller the error, the higher the similarity; for example, assuming that the reference attribute is color and includes three channels, three first reference attribute threshold intervals can be obtained; A first scaling factor for the initial prediction weights is determined based on the threshold range of the at least one first reference attribute. The step of determining the first scaling ratio for the initial prediction weights based on the at least one first reference attribute threshold range may include: At least one scaling ratio for the initial prediction weight is determined based on the at least one first reference attribute threshold interval, wherein one reference attribute threshold interval corresponds to one scaling ratio. For example, assuming the reference attribute is color and includes three channels, three scaling ratios can be obtained. A first scaling factor for the initial prediction weights is determined based on the at least one scaling factor. In some embodiments, determining the first scaling factor for the initial prediction weights based on the at least one scaling factor may include: determining the average of the at least one scaling factor as the first scaling factor for the initial prediction weights. For example, assuming the reference attribute is color, including three channels, resulting in three scaling factors, the average of these three scaling factors can be determined as the first scaling factor for the initial prediction weights. In some embodiments, determining the first scaling factor for the initial prediction weights based on the at least one scaling factor may include: determining the maximum value among the at least one scaling factor as the first scaling factor for the initial prediction weights. In some embodiments, the point cloud encoding method shown in FIG6 may further include: sending the first information in the attribute encoding bitstream to the decoding end device so that the decoding end device can perform decoding. The first information includes at least one of the following: N thresholds are used to determine the threshold intervals of the M reference attributes corresponding to each of the at least one channel; Each of the at least one channel corresponds to M reference attribute threshold intervals; The scaling ratio corresponding to each of the M reference attribute threshold intervals of the at least one channel. Each channel corresponds to M reference attribute threshold intervals, and each reference attribute threshold interval corresponds to a scaling ratio. In some embodiments, the first information may also be embedded in both the encoding and decoding devices. It is understood that embedding the first information in the decoding device can reduce the size of the attribute-encoded bitstream and eliminate the need for the decoding device to retrieve the first information from the attribute-encoded bitstream. In the first embodiment, the step of correcting the initial prediction weights according to the first scaling ratio to obtain the final prediction weights of the neighbor nodes for the first child node may include: Based on the first scaling ratio and the initial prediction weight, determine a correction value for the initial prediction weight; The initial influence weight is adjusted based on the correction value to obtain the final influence weight of the neighbor node on the first child node. Wherein, determining the correction value for the initial prediction weight based on the first scaling ratio and the initial prediction weight includes: Based on the third formula, the first scaling factor, and the initial prediction weights, a correction value for the initial prediction weights is determined, wherein the third formula is: δw=weight_org×scale>>precisebit-weight_org Where δw represents the correction value, weight_org represents the initial prediction weight, scale represents the first scaling ratio, and precisebit is the number of shift bits used to ensure the accuracy of scale. Based on this embodiment, the step of correcting the initial prediction weight according to the correction value to obtain the final prediction weight of the neighbor node for the first child node includes: summing the correction value and the initial prediction weight to obtain the final prediction weight of the neighbor node for the first child node, that is: weight_final = weight_org + δw Wherein, weight_final represents the final predicted weight of the neighbor node for the first child node. In the second embodiment, the step of correcting the initial prediction weights according to the first scaling ratio to obtain the final prediction weights of the neighbor nodes for the first child nodes includes: Based on the fourth formula, the first scaling factor, and the initial prediction weights, the final prediction weights of the neighbor nodes for the first child nodes are determined, wherein the fourth formula is: weight_final=weight_org×scale>>precisebit Wherein, weight_final represents the final predicted weight, weight_org represents the initial predicted weight, scale represents the first scaling ratio, and precisebit is the number of shift bits used to ensure the precision of scale. In the third and fourth formulas above, ">>" is a shift symbol used to ensure accuracy. It is understood that scaling and shifting the initial prediction weights according to the first scaling ratio can yield the final prediction weights more quickly. It is understandable that the above-mentioned correction of the initial prediction weight is performed on each neighbor node of the first child node. Finally, the intra-frame prediction attribute value is obtained by weighted averaging of the corrected final prediction weight and the attribute values ​​of the neighbor nodes for each first child node. Step 603: Predict the first child node based on the final prediction weights of each neighbor node for the first child node. In some embodiments, predicting the first child node based on the final prediction weights of each of the neighboring nodes for the first child node includes: The intra-frame prediction attribute value of the first child node is determined based on the attribute values ​​of each neighbor node and the final prediction weight of each neighbor node for the first child node. Based on the original attribute value and intra-predicted attribute value of the first child node, determine the intra-predicted residual of the first child node; The intra-frame prediction residual of the first child node is encoded to obtain the attribute-coded bitstream. In some embodiments, the intra-frame prediction attribute value of the first child node is: the attribute value of each of the neighboring nodes and the weighted average of the final prediction weights of each of the neighboring nodes for the first child node. In some embodiments, the intra-prediction residual of the first child node is the difference between the original attribute value and the intra-prediction attribute value of the first child node. In some embodiments, encoding the intra-frame prediction residual of the first child node to obtain an attribute-coded bitstream may specifically include: The intra-frame prediction residual of the first child node is subjected to Region Adaptive Hierarchical Transform (RAHT) to obtain the residual transformation coefficients of the first child node. The residual transform coefficients of the first child node are encoded (e.g., quantization entropy encoding) to obtain the attribute-encoded bitstream. The process of performing Region Adaptive Hierarchical Transformation (RAHT) on the intra-frame prediction residual of the first sub-node to obtain the residual transformation coefficients of the first sub-node can be referred to the explanation of Figure 5 above. The encoding method for encoding the residual transform coefficients of the first child node can include, but is not limited to, exponential Golomb coding, arithmetic coding, etc. The position in the bitstream can be in the attribute parameter set (aps) or the slice parameter set (abh), without restriction. Optionally, in some embodiments, the first child node to be encoded satisfies one of the following: The first child node belongs to the nodes of the first A layers of the transformation tree; The first child node belongs to the node of the last B layer of the transformation tree; The first child node does not belong to the nodes of the first O layers of the transformation tree; The first child node does not belong to the nodes of the later P layer of the transformation tree; Among them, A, B, O, and P are greater than or equal to 0. Furthermore, the point cloud encoding method shown in Figure 6 may also include: sending the first parameter in the attribute encoding bitstream to the decoding device so that the decoding device can determine whether or not to perform cross-attribute prediction. The first parameter is used to indicate one of the following: The nodes of the first A layers of the transformation tree perform cross-attribute prediction; The nodes of the last B layer of the transformation tree perform cross-attribute prediction; The nodes in the first O layers of the transformation tree do not perform cross-attribute prediction; The nodes in the last P layers of the transformation tree do not perform cross-attribute prediction. In other words, the embodiments of this application can perform cross-attribute prediction on nodes of some layers, or not perform cross-attribute prediction on nodes of some layers, thus flexibly setting the layers to perform cross-attribute prediction. The point cloud encoding method proposed in the embodiment shown in Figure 6 introduces the correlation between the reference attribute values ​​of neighboring nodes and the child node to be encoded (the first child node) when determining the prediction weight of the child node of the current node to be encoded. That is, cross-attribute prediction is performed, which makes the obtained prediction weight more accurate and thus can further improve the compression efficiency. Figure 7 illustrates a point cloud decoding method corresponding to the point cloud encoding method shown in Figure 6, which will be explained below. As shown in Figure 7, a point cloud decoding method proposed in one embodiment of this application may include: Step 701: If intra-frame prediction is allowed for the current node to be decoded in the transform tree corresponding to the point cloud to be decoded, obtain the initial prediction weights of the neighboring nodes for the first child node, and obtain the reference attribute values ​​of the neighboring nodes and the reference attribute values ​​of the first child node, wherein the first child node is any child node of the current node to be decoded. Optionally, before step 701, a transformation tree corresponding to the point cloud to be decoded can be generated based on the reconstructed geometric information of the point cloud to be decoded. Specifically, the point cloud is reordered based on the reconstructed geometric coordinates of the point cloud to be decoded, and a P-layer transformation tree is constructed on the reordered point cloud data. A bottom-up construction method can be adopted, with the bottom layer containing all nodes and the top layer being the root node layer, containing only one node. Optionally, during the construction of the transformation tree, corresponding Morton code information and weight information are generated for the nodes after the current attribute merging. In some embodiments, it can be determined whether the current node to be decoded is allowed to perform intra-frame prediction using methods found in related technologies. In some embodiments, the initial prediction weights of the neighbor nodes to the first child node obtained in step 701 may be prediction weights obtained in related technologies based on the temporal and spatial correlations of the reconstructed neighbor nodes. In some embodiments, assuming the attribute to be decoded is the first attribute, the reference attribute value of the neighboring node obtained in step 701 can be the reference attribute value of the first attribute of the neighboring node, and the reference attribute value of the first child node can be the reference attribute value of the first attribute of the first child node. The reference attribute can be one or more attributes other than the first attribute. Step 702: Based on the reference attribute values ​​of the neighboring nodes and the reference attribute values ​​of the first child node, determine the similarity information between the neighboring nodes and the first child node, and correct the initial prediction weights based on the similarity information to obtain the final prediction weights of the neighboring nodes for the first child node. The similarity information is used to characterize the degree of similarity. In some embodiments, a reference attribute of a node may include at least one channel. For example, if the reference attribute of a node is color, it may include three channels: R, G, and B; if the reference attribute of a node is reflectivity, it may include only one channel. In some embodiments, the similarity information between the neighbor node and the first child node may include: the error between the reference attribute values ​​of the neighbor node and the first child node, wherein determining the similarity information between the neighbor node and the first child node based on the reference attribute values ​​of the neighbor node and the first child node may include: determining the error between the neighbor node and the first child node in at least one channel based on the reference attribute values ​​of the neighbor node and the first child node in at least one channel, wherein one channel corresponds to one error, and the error is inversely proportional to the similarity, that is, the larger the error, the smaller the similarity, and the smaller the error, the larger the similarity. It is understood that the error between the reference attribute values ​​of the neighboring node and the first child node can be used to quickly and accurately characterize the similarity between the two, so that the correlation performance between the two can better influence the final prediction weight of the neighboring node for the first child node, thereby improving the accuracy of the final prediction weight and thus improving the coding efficiency. In one example, the reference attribute value of the neighbor node can be set to `neibor_ref`. When the reference attribute value of the first child node is `curChild_ref`, the initial prediction weight of the neighbor node for the first child node is `weight_org`, and the attribute value of the neighbor node is `neibor_cur`. The step of determining the error between the neighbor node and the first child node in the at least one channel based on the reference attribute values ​​of the neighbor node and the first child node in the at least one channel includes: Based on the first formula, the reference attribute values ​​of the neighbor node and the first child node under the at least one channel, determine the error between the neighbor node and the first child node under the at least one channel; The first formula is: error=abs(neibor_ref-curChild_ref) Here, error represents the error; the larger the error, the worse the similarity, and vice versa. In some embodiments, correcting the initial prediction weights based on the similarity information to obtain the final prediction weights of the neighboring nodes for the first child node may include: A first scaling ratio is determined for the initial prediction weights based on the similarity information, wherein the higher the similarity, the larger the first scaling ratio, and the lower the similarity, the smaller the first scaling ratio. The initial prediction weights are corrected according to the first scaling ratio to obtain the final prediction weights of the neighbor nodes for the first child node. Optionally, the reference attributes of each node include at least one channel, wherein determining the first scaling factor for the initial prediction weights based on the similarity information may include: 1) Obtain N pre-set thresholds: th1, th2, th3, ..., thN, where the N thresholds are N positive integers in ascending order; 2) Based on the second formula, the pre-set N thresholds, and the reference attribute values ​​of the first child node under the at least one channel, determine the N reference attribute thresholds corresponding to each of the at least one channel; Wherein, the N thresholds are N positive integers that increase sequentially, and the second formula is: refAttrThi=curChild_ref×thi Where refAttrThi represents the threshold of the i-th reference attribute, curChild_ref represents the reference attribute value of the first child node, and thi represents the i-th threshold among the N pre-set thresholds, i = 1, 2, ..., N. 3) Based on the N reference attribute thresholds corresponding to each of the at least one channel, determine the M reference attribute threshold intervals corresponding to each of the at least one channel, where M is a positive integer and M can be less than, equal to or greater than N. For example, the following M reference attribute threshold intervals can be determined: [0, refAttrTh1], [refAttrTh1, refAttrTh2], ..., [refAttrThN-1, refAttrThN], [refAttrThN, ∞]. It is easy to see that at this time, M = N+1. In this configuration, a reference attribute threshold interval within a single channel corresponds to a scaling factor. Optionally, M reference attribute threshold intervals within a single channel correspond to M scaling factors, and these M scaling factors can be the same or different. For example, the correspondence between reference attribute threshold intervals and scaling factors can be: [0,refAttrTh1]:scale=8 [refAttrTh1,refAttrTh2]:scale=7 … [refAttrThN-1,refAttrThN]:scale=6 [refAttrThN,∞]:scale=1 It is understandable that by setting a reasonable reference attribute threshold range and the corresponding scaling ratio, the first scaling ratio for the initial prediction weight can be accurately determined, resulting in an accurate final prediction weight, thereby improving the accuracy of the final prediction weight and thus improving coding efficiency. Optionally, determining the first scaling ratio for the initial prediction weights based on the similarity information may further include: 4) Multiply the error in the at least one channel by a first factor, wherein the first factor is used to ensure the accuracy of the error. For example, multiply the error by 10, that is: error = 10 × error It should be noted that the purpose of the above four steps is to simulate the process of calculating the relative error between the neighbor node and the first child node: abs(neibor_ref-curChild_ref) / curChild_ref. This formula can calculate the relative error between the reference attribute of the neighbor node and the reference attribute of the first child node, reflecting the similarity between the two, and making the error as integer as possible to facilitate calculation, thereby further improving coding efficiency. Optionally, determining the first scaling ratio for the initial prediction weights based on the similarity information may further include: From the M reference attribute threshold intervals corresponding to each of the at least one channel, determine the first reference attribute threshold interval where the error under the at least one channel is located, and obtain at least one first reference attribute threshold interval, wherein the larger the error, the lower the similarity, and the smaller the error, the higher the similarity; for example, assuming that the reference attribute is color and includes three channels, three first reference attribute threshold intervals can be obtained; A first scaling factor for the initial prediction weights is determined based on the threshold range of the at least one first reference attribute. The step of determining the first scaling ratio for the initial prediction weights based on the at least one first reference attribute threshold range may include: At least one scaling ratio for the initial prediction weight is determined based on the at least one first reference attribute threshold interval, wherein one reference attribute threshold interval corresponds to one scaling ratio. For example, assuming the reference attribute is color and includes three channels, three scaling ratios can be obtained. A first scaling factor for the initial prediction weights is determined based on the at least one scaling factor. In some embodiments, determining the first scaling factor for the initial prediction weights based on the at least one scaling factor may include: determining the average of the at least one scaling factor as the first scaling factor for the initial prediction weights. For example, assuming the reference attribute is color, including three channels, resulting in three scaling factors, the average of these three scaling factors can be determined as the first scaling factor for the initial prediction weights. In some embodiments, determining the first scaling factor for the initial prediction weights based on the at least one scaling factor may include: determining the maximum value among the at least one scaling factor as the first scaling factor for the initial prediction weights. In some embodiments, the point cloud decoding method shown in FIG7 may further include: receiving first information carried in the attribute-encoded bitstream to facilitate decoding by the decoding end device. The first information includes at least one of the following: N thresholds are used to determine the threshold intervals of the M reference attributes corresponding to each of the at least one channel; Each of the at least one channel corresponds to M reference attribute threshold intervals; The scaling ratio corresponding to each of the M reference attribute threshold intervals of the at least one channel. Each channel corresponds to M reference attribute threshold intervals, and each reference attribute threshold interval corresponds to a scaling ratio. In some embodiments, the first information may also be embedded in both the encoding and decoding devices. It is understood that embedding the first information in the decoding device can reduce the size of the attribute-encoded bitstream and eliminate the need for the decoding device to retrieve the first information from the attribute-encoded bitstream. In the first embodiment, the step of correcting the initial prediction weights according to the first scaling ratio to obtain the final prediction weights of the neighbor nodes for the first child node may include: Based on the first scaling ratio and the initial prediction weight, determine a correction value for the initial prediction weight; The initial influence weight is adjusted based on the correction value to obtain the final influence weight of the neighbor node on the first child node. Wherein, determining the correction value for the initial prediction weight based on the first scaling ratio and the initial prediction weight includes: Based on the third formula, the first scaling factor, and the initial prediction weights, a correction value for the initial prediction weights is determined, wherein the third formula is: δw=weight_org×scale>>precisebit-weight_org Where δw represents the correction value, weight_org represents the initial prediction weight, scale represents the first scaling ratio, and precisebit is the number of shift bits used to ensure the accuracy of scale. Based on this embodiment, the step of correcting the initial prediction weight according to the correction value to obtain the final prediction weight of the neighbor node for the first child node includes: summing the correction value and the initial prediction weight to obtain the final prediction weight of the neighbor node for the first child node, that is: weight_final = weight_org + δw Wherein, weight_final represents the final predicted weight of the neighbor node for the first child node. In the second embodiment, the step of correcting the initial prediction weights according to the first scaling ratio to obtain the final prediction weights of the neighbor nodes for the first child nodes includes: Based on the fourth formula, the first scaling factor, and the initial prediction weights, the final prediction weights of the neighbor nodes for the first child nodes are determined, wherein the fourth formula is: weight_final=weight_org×scale>>precisebit Wherein, weight_final represents the final predicted weight, weight_org represents the initial predicted weight, scale represents the first scaling ratio, and precisebit is the number of shift bits used to ensure the precision of scale. In the third and fourth formulas above, ">>" is a shift symbol used to ensure accuracy. It is understood that scaling and shifting the initial prediction weights according to the first scaling ratio can yield the final prediction weights more quickly. It is understandable that the above-mentioned correction of the initial prediction weight is performed on each neighbor node of the first child node. Finally, the intra-frame prediction attribute value is obtained by weighted averaging of the corrected final prediction weight and the attribute values ​​of the neighbor nodes for each first child node. Step 703: Predict the first child node according to the final prediction weight of each neighbor node for the first child node. In some embodiments, predicting the first child node based on the final prediction weights of each neighbor node for the first child node may include: determining the intra-frame prediction attribute value of the first child node based on the attribute values ​​of each neighbor node and the final prediction weights of each neighbor node for the first child node. In some embodiments, the intra-frame prediction attribute value of the first child node is: the attribute value of each of the neighboring nodes and the weighted average of the final prediction weights of each of the neighboring nodes for the first child node. In some embodiments, predicting the first child node based on the final prediction weights of each of the neighboring nodes may further include: The reconstruction residual of the first child node is obtained by decoding the input attribute-encoded bitstream; The reconstruction attribute value of the first child node is determined based on the intra-frame prediction attribute value and the reconstruction residual of the first child node. In some embodiments, decoding the input attribute-encoded bitstream to obtain the reconstruction residual of the first child node may include: The input attribute-encoded bitstream is decoded and dequantized to obtain the residual transform coefficients of the first child node; The inverse transformation of the residual transformation coefficients of the first child node is performed using the region adaptive hierarchical transformation to obtain the reconstruction residual of the first child node. Optionally, in some embodiments, the first child node to be decoded satisfies one of the following: The first child node belongs to the nodes of the first A layers of the transformation tree; The first child node belongs to the node of the last B layer of the transformation tree; The first child node does not belong to the nodes of the first O layers of the transformation tree; The first child node does not belong to the nodes of the later P layer of the transformation tree; Among them, A, B, O, and P are greater than or equal to 0. Furthermore, the point cloud decoding method shown in Figure 7 may further include: obtaining a first parameter from the attribute-encoded bitstream, so that the decoding device can determine whether or not to perform cross-attribute prediction on the nodes. The first parameter is used to indicate one of the following: The nodes of the first A layers of the transformation tree perform cross-attribute prediction; The nodes of the last B layer of the transformation tree perform cross-attribute prediction; The nodes in the first O layers of the transformation tree do not perform cross-attribute prediction; The nodes in the last P layers of the transformation tree do not perform cross-attribute prediction. It is understandable that the parsing is performed layer by layer and node by node from top to bottom. When the bottom layer is reached, the reconstructed attribute values ​​of all nodes are obtained, thus completing the attribute decoding. The point cloud decoding method proposed in the embodiment shown in Figure 7, corresponding to the point cloud encoding method, introduces the correlation between the reference attribute values ​​of the neighboring nodes and the child node to be encoded (the first child node) when determining the prediction weight of the child node of the current node to be decoded, that is, cross-attribute prediction is performed, which makes the obtained prediction weight more accurate, and thus can further improve the decompression efficiency. Figures 8A, 8B, 8C, and 8D illustrate the test results of point cloud encoding using the point cloud encoding method provided in the embodiments of this application. Figures 8A and 8B show the effect of predicting reflectance using color as a reference attribute, while Figures 8C and 8D show the effect of predicting color using reflectance. As can be seen from Figures 8A, 8B, 8C, and 8D, when using color to predict reflectance, there is a 4.9% gain under geometrically lossless but attribute-lossy conditions, and a 3.7% gain under both geometrically and attribute-lossy conditions. When using reflectance to predict color, the Luma, chromaB, and chromaC channels have gains of 1.6%, 1.2%, and 1.2% respectively under geometrically lossless but attribute-lossy conditions, and 1.3%, 1.0%, and 1.1% respectively under both geometrically and attribute-lossy conditions. This demonstrates that cross-attribute prediction can improve the accuracy of prediction weights, thereby improving compression efficiency. This application provides a point cloud encoding method or a point cloud decoding method, the execution subject of which can be a virtual device. This application uses a virtual device executing the point cloud encoding method or a point cloud decoding method as an example to illustrate the point cloud encoding device or a point cloud decoding device provided in this application. As shown in Figure 9, one embodiment of this application proposes a point cloud encoding device 900, which can be used in encoding end devices. The device 900 may include: an acquisition module 901, a weight correction module 902, and an intra-frame prediction module 903. The acquisition module 901 is used to acquire the initial prediction weights of the neighboring nodes to the first child node in the transform tree corresponding to the point cloud to be encoded, provided that intra-frame prediction is allowed for the current node to be encoded, and to acquire the reference attribute values ​​of the neighboring nodes and the reference attribute values ​​of the first child node, wherein the first child node is any child node of the current node to be encoded. Optionally, the device 900 may further include: a transformation tree generation module, used to generate a transformation tree corresponding to the point cloud to be encoded based on the geometric information of the point cloud to be encoded. Specifically, the point cloud is reordered based on its geometric coordinates, and a P-layer transformation tree is constructed from the reordered point cloud data. A bottom-up construction method can be used, with the bottom layer containing all nodes and the top layer being the root node layer, containing only one node. Optionally, during the construction of the transformation tree, corresponding Morton code information, attribute information, and weight information are generated for the node after the current attribute merging. In some embodiments, the acquisition module 901 may determine whether the current node to be encoded is allowed to perform intra-frame prediction by means of methods in related technologies. In some embodiments, the initial prediction weights of the neighbor nodes to the first child node obtained in the acquisition module 901 may be prediction weights obtained in related technologies based on the temporal and spatial correlations of the reconstructed neighbor nodes. In some embodiments, assuming that the attribute to be encoded is the first attribute, the reference attribute value of the neighboring node obtained by the acquisition module 901 can be the reference attribute value of the first attribute of the neighboring node, and the reference attribute value of the first child node can be the reference attribute value of the first attribute of the first child node. The reference attribute can be one or more attributes other than the first attribute. The weight correction module 902 determines the similarity information between the neighbor node and the first child node based on the reference attribute values ​​of the neighbor node and the reference attribute values ​​of the first child node, and corrects the initial prediction weight based on the similarity information to obtain the final prediction weight of the neighbor node for the first child node. The similarity information is used to characterize the degree of similarity. In some embodiments, a reference attribute of a node may include at least one channel. For example, if the reference attribute of a node is color, it may include three channels: R, G, and B; if the reference attribute of a node is reflectivity, it may include only one channel. In some embodiments, the similarity information between the neighbor node and the first child node may include: the error between the reference attribute values ​​of the neighbor node and the first child node. The weight correction module 902 can be used to: determine the error between the neighbor node and the first child node in at least one channel based on the reference attribute values ​​of the neighbor node and the first child node in at least one channel, wherein one channel corresponds to one error, the larger the error, the smaller the similarity, and the smaller the error, the greater the similarity. It is understood that the error between the reference attribute values ​​of the neighboring node and the first child node can be used to quickly and accurately characterize the similarity between the two, so that the correlation performance between the two can better influence the final prediction weight of the neighboring node for the first child node, thereby improving the accuracy of the final prediction weight and thus improving the coding efficiency. In one example, the reference attribute value of the neighboring node can be set to neibor_ref. When the reference attribute value of the first child node is curChild_ref, the initial prediction weight of the neighboring node for the first child node is weight_org, and the attribute value of the neighboring node is neibor_cur. The weight correction module 902 can specifically be used for: Based on the first formula, the reference attribute values ​​of the neighbor node and the first child node under the at least one channel, determine the error between the neighbor node and the first child node under the at least one channel; The first formula is: error=abs(neibor_ref-curChild_ref) Here, error represents the error; the larger the error, the worse the similarity, and vice versa. In some embodiments, the weight correction module 902 can be specifically used for: A first scaling factor is determined for the initial prediction weights based on the similarity information, wherein the higher the similarity information, the larger the first scaling factor, and the lower the similarity information, the smaller the first scaling factor. The initial prediction weights are corrected according to the first scaling ratio to obtain the final prediction weights of the neighbor nodes for the first child node. Optionally, the reference attributes of each node include at least one channel, wherein determining the first scaling factor for the initial prediction weights based on the similarity information may include: 1) Obtain N pre-set thresholds: th1, th2, th3, ..., thN, where the N thresholds are N positive integers in ascending order; 2) Based on the second formula, the pre-set N thresholds, and the reference attribute values ​​of the first child node under the at least one channel, determine the N reference attribute thresholds corresponding to each of the at least one channel; Wherein, the N thresholds are N positive integers that increase sequentially, and the second formula is: refAttrThi=curChild_ref×thi Where refAttrThi represents the threshold of the i-th reference attribute, curChild_ref represents the reference attribute value of the first child node, and thi represents the i-th threshold among the N pre-set thresholds, i = 1, 2, ..., N. 3) Based on the N reference attribute thresholds corresponding to each of the at least one channel, determine the M reference attribute threshold intervals corresponding to each of the at least one channel, where M is a positive integer and M can be less than, equal to or greater than N. For example, the following M reference attribute threshold intervals can be determined: [0, refAttrTh1], [refAttrTh1, refAttrTh2], ..., [refAttrThN-1, refAttrThN], [refAttrThN, ∞]. It is easy to see that at this time, M = N+1. In this configuration, a reference attribute threshold interval within a single channel corresponds to a scaling factor. Optionally, M reference attribute threshold intervals within a single channel correspond to M scaling factors, and these M scaling factors can be the same or different. For example, the correspondence between reference attribute threshold intervals and scaling factors can be: [0,refAttrTh1]:scale=8 [refAttrTh1,refAttrTh2]:scale=7 … [refAttrThN-1,refAttrThN]:scale=6 [refAttrThN,∞]:scale=1 It is understandable that by setting a reasonable reference attribute threshold range and the corresponding scaling ratio, the first scaling ratio for the initial prediction weight can be accurately determined, resulting in an accurate final prediction weight, thereby improving the accuracy of the final prediction weight and thus improving coding efficiency. Optionally, determining the first scaling ratio for the initial prediction weights based on the similarity information may further include: 4) Multiply the error in the at least one channel by a first factor, wherein the first factor is used to ensure the accuracy of the error. For example, multiply the error by 10, that is: error = 10 × error It should be noted that the purpose of the above four steps is to simulate the process of calculating the relative error between the neighbor node and the first child node: abs(neibor_ref-curChild_ref) / curChild_ref. This formula can calculate the relative error between the reference attribute of the neighbor node and the reference attribute of the first child node, reflecting the similarity between the two, and making the error as integer as possible to facilitate calculation, thereby further improving coding efficiency. Optionally, determining the first scaling ratio for the initial prediction weights based on the similarity information may further include: From the M reference attribute threshold intervals corresponding to each of the at least one channel, determine the first reference attribute threshold interval where the error under the at least one channel is located, and obtain at least one first reference attribute threshold interval, wherein the larger the error, the lower the similarity, and the smaller the error, the higher the similarity; for example, assuming that the reference attribute is color and includes three channels, three first reference attribute threshold intervals can be obtained; A first scaling factor for the initial prediction weights is determined based on the threshold range of the at least one first reference attribute. The step of determining the first scaling ratio for the initial prediction weights based on the at least one first reference attribute threshold range may include: At least one scaling ratio for the initial prediction weight is determined based on the at least one first reference attribute threshold interval, wherein one reference attribute threshold interval corresponds to one scaling ratio. For example, assuming the reference attribute is color and includes three channels, three scaling ratios can be obtained. A first scaling factor for the initial prediction weights is determined based on the at least one scaling factor. In some embodiments, determining the first scaling factor for the initial prediction weights based on the at least one scaling factor may include: determining the average of the at least one scaling factor as the first scaling factor for the initial prediction weights. For example, assuming the reference attribute is color, including three channels, resulting in three scaling factors, the average of these three scaling factors can be determined as the first scaling factor for the initial prediction weights. In some embodiments, determining the first scaling factor for the initial prediction weights based on the at least one scaling factor may include: determining the maximum value among the at least one scaling factor as the first scaling factor for the initial prediction weights. In some embodiments, the point cloud encoding device shown in FIG9 may further include: a first sending module, configured to send the first information in the attribute encoding bitstream to the decoding end device so that the decoding end device can perform decoding. The first information includes at least one of the following: N thresholds are used to determine the threshold intervals of the M reference attributes corresponding to each of the at least one channel; Each of the at least one channel corresponds to M reference attribute threshold intervals; The scaling ratio corresponding to each of the M reference attribute threshold intervals of the at least one channel. Each channel corresponds to M reference attribute threshold intervals, and each reference attribute threshold interval corresponds to a scaling ratio. In some embodiments, the first information may also be embedded in both the encoding and decoding devices. It is understood that embedding the first information in the decoding device can reduce the size of the attribute-encoded bitstream and eliminate the need for the decoding device to retrieve the first information from the attribute-encoded bitstream. In the first embodiment, the weight correction module 902 is specifically used for: Based on the first scaling ratio and the initial prediction weight, determine a correction value for the initial prediction weight; The initial influence weight is adjusted based on the correction value to obtain the final influence weight of the neighbor node on the first child node. Wherein, determining the correction value for the initial prediction weight based on the first scaling ratio and the initial prediction weight includes: Based on the third formula, the first scaling factor, and the initial prediction weights, a correction value for the initial prediction weights is determined, wherein the third formula is: δw=weight_org×scale>>precisebit-weight_org Where δw represents the correction value, weight_org represents the initial prediction weight, scale represents the first scaling ratio, and precisebit is the number of shift bits used to ensure the accuracy of scale. Based on this embodiment, the step of correcting the initial prediction weight according to the correction value to obtain the final prediction weight of the neighbor node for the first child node includes: summing the correction value and the initial prediction weight to obtain the final prediction weight of the neighbor node for the first child node, that is: weight_final = weight_org + δw Wherein, weight_final represents the final predicted weight of the neighbor node for the first child node. In the second embodiment, the weight correction module 902 can be specifically used for: Based on the fourth formula, the first scaling factor, and the initial prediction weights, the final prediction weights of the neighbor nodes for the first child nodes are determined, wherein the fourth formula is: weight_final=weight_org×scale>>precisebit Wherein, weight_final represents the final predicted weight, weight_org represents the initial predicted weight, scale represents the first scaling ratio, and precisebit is the number of shift bits used to ensure the precision of scale. In the third and fourth formulas above, ">>" is a shift symbol used to ensure accuracy. It is understood that scaling and shifting the initial prediction weights according to the first scaling ratio can yield the final prediction weights more quickly. It is understandable that the above-mentioned correction of the initial prediction weight is performed on each neighbor node of the first child node. Finally, the intra-frame prediction attribute value is obtained by weighted averaging of the corrected final prediction weight and the attribute values ​​of the neighbor nodes for each first child node. The intra-frame prediction module 903 is used to predict the first child node based on the final prediction weights of each of the neighboring nodes for the first child node. In some embodiments, the intra-frame prediction module 903 can be specifically used for: The intra-frame prediction attribute value of the first child node is determined based on the attribute values ​​of each neighbor node and the final prediction weight of each neighbor node for the first child node. Based on the original attribute value and intra-predicted attribute value of the first child node, determine the intra-predicted residual of the first child node; The intra-frame prediction residual of the first child node is encoded to obtain the attribute-coded bitstream. In some embodiments, the intra-frame prediction attribute value of the first child node is: the attribute value of each of the neighboring nodes and the weighted average of the final prediction weights of each of the neighboring nodes for the first child node. In some embodiments, the intra-prediction residual of the first child node is the difference between the original attribute value and the intra-prediction attribute value of the first child node. In some embodiments, encoding the intra-frame prediction residual of the first child node to obtain an attribute-coded bitstream may include: The intra-frame prediction residual of the first child node is subjected to Region Adaptive Hierarchical Transform (RAHT) to obtain the residual transformation coefficients of the first child node. The residual transform coefficients of the first child node are encoded (e.g., quantization entropy encoding) to obtain the attribute-encoded bitstream. The process of performing Region Adaptive Hierarchical Transformation (RAHT) on the intra-frame prediction residual of the first sub-node to obtain the residual transformation coefficients of the first sub-node can be referred to the explanation of Figure 5 above. The encoding method for encoding the residual transform coefficients of the first child node can include, but is not limited to, exponential Golomb coding, arithmetic coding, etc. The position in the bitstream can be in the attribute parameter set (aps) or the slice parameter set (abh), without restriction. Optionally, in some embodiments, the first child node to be encoded satisfies one of the following: The first child node belongs to the nodes of the first A layers of the transformation tree; The first child node belongs to the node of the last B layer of the transformation tree; The first child node does not belong to the nodes of the first O layers of the transformation tree; The first child node does not belong to the nodes of the later P layer of the transformation tree; Among them, A, B, O, and P are greater than or equal to 0. Furthermore, the point cloud encoding device shown in Figure 9 may also include: a second sending module, used to send the first parameter in the attribute encoding bitstream to the decoding end device, so that the decoding end device can determine whether or not to perform cross-attribute prediction. The first parameter is used to indicate one of the following: The nodes of the first A layers of the transformation tree perform cross-attribute prediction; The nodes of the last B layer of the transformation tree perform cross-attribute prediction; The nodes in the first O layers of the transformation tree do not perform cross-attribute prediction; The nodes in the last P layers of the transformation tree do not perform cross-attribute prediction. In other words, the point cloud encoding device 900 provided in this application embodiment can perform cross-attribute prediction on nodes of some layers, or it can not perform cross-attribute prediction on nodes of some layers, thus flexibly setting the layers to perform cross-attribute prediction. In other words, the embodiments of this application can perform cross-attribute prediction on nodes of some layers, or not perform cross-attribute prediction on nodes of some layers, thus flexibly setting the layers to perform cross-attribute prediction. The point cloud encoding device 900 provided in this application embodiment can implement the various processes implemented in the method embodiment of FIG6 and achieve the same technical effect. To avoid repetition, it will not be described again here. As shown in Figure 10, an embodiment of this application proposes a point cloud decoding device 1000, which can be used in decoding end devices. The device 1000 may include: an acquisition module 1001, a weight correction module 1002, and an intra-frame prediction module 1003. The acquisition module 1001 is used to acquire the initial prediction weights of the neighboring nodes to the first child node in the transform tree corresponding to the point cloud to be decoded, provided that intra-frame prediction is allowed for the current node to be decoded, and to acquire the reference attribute values ​​of the neighboring nodes and the reference attribute values ​​of the first child node, wherein the first child node is any child node of the current node to be decoded. Optionally, the point cloud decoding device 1000 may further include a transformation tree construction module, used to generate a transformation tree corresponding to the point cloud to be decoded based on the reconstructed geometric information of the point cloud to be decoded. Specifically, the point cloud is reordered based on the reconstructed geometric coordinates of the point cloud to be decoded, and a P-layer transformation tree is constructed on the reordered point cloud data. A bottom-up construction method can be adopted, with the bottom layer containing all nodes and the top layer being the root node layer, containing only one node. Optionally, during the construction of the transformation tree, corresponding Morton code information and weight information are generated for the nodes after the current attribute merging. In some embodiments, it can be determined whether the current node to be decoded is allowed to perform intra-frame prediction using methods found in related technologies. In some embodiments, the initial prediction weights of neighbor nodes to the first child node obtained in the acquisition module 1001 may be prediction weights obtained in related technologies based on the temporal and spatial correlations of the reconstructed neighbor nodes. In some embodiments, assuming that the attribute to be decoded is the first attribute, the reference attribute value of the neighboring node obtained by the acquisition module 1001 can be the reference attribute value of the first attribute of the neighboring node, and the reference attribute value of the first child node can be the reference attribute value of the first attribute of the first child node. The reference attribute can be one or more attributes other than the first attribute. The weight correction module 1002 is used to determine the similarity information between the neighbor node and the first child node based on the reference attribute values ​​of the neighbor node and the reference attribute values ​​of the first child node, and to correct the initial prediction weight based on the similarity information to obtain the final prediction weight of the neighbor node for the first child node. The similarity information is used to characterize the degree of similarity. In some embodiments, a reference attribute of a node may include at least one channel. For example, if the reference attribute of a node is color, it may include three channels: R, G, and B; if the reference attribute of a node is reflectivity, it may include only one channel. In some embodiments, the similarity information between the neighbor node and the first child node may include: the error between the reference attribute values ​​of the neighbor node and the first child node. The weight correction module 1002 can be used to: determine the error between the neighbor node and the first child node in at least one channel based on the reference attribute values ​​of the neighbor node and the first child node in at least one channel, wherein one channel corresponds to one error, the larger the error, the smaller the similarity, and the smaller the error, the greater the similarity. It is understood that the error between the reference attribute values ​​of the neighboring node and the first child node can be used to quickly and accurately characterize the similarity between the two, so that the correlation performance between the two can better influence the final prediction weight of the neighboring node for the first child node, thereby improving the accuracy of the final prediction weight and thus improving the coding efficiency. In one example, the reference attribute value of the neighboring node can be set to neibor_ref. When the reference attribute value of the first child node is curChild_ref, the initial prediction weight of the neighboring node for the first child node is weight_org, and the attribute value of the neighboring node is neibor_cur. Specifically, the weight correction module 1002 can be used for: Based on the first formula, the reference attribute values ​​of the neighbor node and the first child node under the at least one channel, determine the error between the neighbor node and the first child node under the at least one channel; The first formula is: error=abs(neibor_ref-curChild_ref) Here, error represents the error; the larger the error, the worse the similarity, and vice versa. In some embodiments, the weight correction module 1002 can be specifically used for: A first scaling factor is determined for the initial prediction weights based on the similarity information, wherein the higher the similarity information, the larger the first scaling factor, and the lower the similarity information, the smaller the first scaling factor. The initial prediction weights are corrected according to the first scaling ratio to obtain the final prediction weights of the neighbor nodes for the first child node. Optionally, the reference attributes of each node include at least one channel, wherein determining the first scaling factor for the initial prediction weights based on the similarity information may include: 1) Obtain N pre-set thresholds: th1, th2, th3, ..., thN, where the N thresholds are N positive integers in ascending order; 2) Based on the second formula, the pre-set N thresholds, and the reference attribute values ​​of the first child node under the at least one channel, determine the N reference attribute thresholds corresponding to each of the at least one channel; Wherein, the N thresholds are N positive integers that increase sequentially, and the second formula is: refAttrThi=curChild_ref×thi Where refAttrThi represents the threshold of the i-th reference attribute, curChild_ref represents the reference attribute value of the first child node, and thi represents the i-th threshold among the N pre-set thresholds, i = 1, 2, ..., N. 3) Based on the N reference attribute thresholds corresponding to each of the at least one channel, determine the M reference attribute threshold intervals corresponding to each of the at least one channel, where M is a positive integer and M can be less than, equal to or greater than N. For example, the following M reference attribute threshold intervals can be determined: [0, refAttrTh1], [refAttrTh1, refAttrTh2], ..., [refAttrThN-1, refAttrThN], [refAttrThN, ∞]. It is easy to see that at this time, M = N+1. In this configuration, a reference attribute threshold interval within a single channel corresponds to a scaling factor. Optionally, M reference attribute threshold intervals within a single channel correspond to M scaling factors, and these M scaling factors can be the same or different. For example, the correspondence between reference attribute threshold intervals and scaling factors can be: [0,refAttrTh1]:scale=8 [refAttrTh1,refAttrTh2]:scale=7 … [refAttrThN-1,refAttrThN]:scale=6 [refAttrThN,∞]:scale=1 It is understandable that by setting a reasonable reference attribute threshold range and the corresponding scaling ratio, the first scaling ratio for the initial prediction weight can be accurately determined, resulting in an accurate final prediction weight, thereby improving the accuracy of the final prediction weight and thus improving coding efficiency. Optionally, determining the first scaling ratio for the initial prediction weights based on the similarity information may further include: 4) Multiply the error in the at least one channel by a first factor, wherein the first factor is used to ensure the accuracy of the error. For example, multiply the error by 10, that is: error = 10 × error It should be noted that the purpose of the above four steps is to simulate the process of calculating the relative error between the neighbor node and the first child node: abs(neibor_ref-curChild_ref) / curChild_ref. This formula can calculate the relative error between the reference attribute of the neighbor node and the reference attribute of the first child node, reflecting the similarity between the two, and making the error as integer as possible to facilitate calculation, thereby further improving coding efficiency. Optionally, determining the first scaling ratio for the initial prediction weights based on the similarity information may further include: From the M reference attribute threshold intervals corresponding to each of the at least one channel, determine the first reference attribute threshold interval where the error under the at least one channel is located, and obtain at least one first reference attribute threshold interval, wherein the larger the error, the lower the similarity, and the smaller the error, the higher the similarity; for example, assuming that the reference attribute is color and includes three channels, three first reference attribute threshold intervals can be obtained; A first scaling factor for the initial prediction weights is determined based on the threshold range of the at least one first reference attribute. The step of determining the first scaling ratio for the initial prediction weights based on the at least one first reference attribute threshold range may include: At least one scaling ratio for the initial prediction weight is determined based on the at least one first reference attribute threshold interval, wherein one reference attribute threshold interval corresponds to one scaling ratio. For example, assuming the reference attribute is color and includes three channels, three scaling ratios can be obtained. A first scaling factor for the initial prediction weights is determined based on the at least one scaling factor. In some embodiments, determining the first scaling factor for the initial prediction weights based on the at least one scaling factor may include: determining the average of the at least one scaling factor as the first scaling factor for the initial prediction weights. For example, assuming the reference attribute is color, including three channels, resulting in three scaling factors, the average of these three scaling factors can be determined as the first scaling factor for the initial prediction weights. In some embodiments, determining the first scaling factor for the initial prediction weights based on the at least one scaling factor may include: determining the maximum value among the at least one scaling factor as the first scaling factor for the initial prediction weights. In some embodiments, the point cloud decoding device shown in FIG10 may further include: a first receiving module, configured to receive first information carried in the attribute encoding bitstream, so as to facilitate decoding by the decoding end device. The first information includes at least one of the following: N thresholds are used to determine the threshold intervals of the M reference attributes corresponding to each of the at least one channel; Each of the at least one channel corresponds to M reference attribute threshold intervals; The scaling ratio corresponding to each of the M reference attribute threshold intervals of the at least one channel. Each channel corresponds to M reference attribute threshold intervals, and each reference attribute threshold interval corresponds to a scaling ratio. In some embodiments, the first information may also be embedded in both the encoding and decoding devices. It is understood that embedding the first information in the decoding device can reduce the size of the attribute-encoded bitstream and eliminate the need for the decoding device to retrieve the first information from the attribute-encoded bitstream. In the first embodiment, the weight correction module 1002 is specifically used for: Based on the first scaling ratio and the initial prediction weight, determine a correction value for the initial prediction weight; The initial influence weight is adjusted based on the correction value to obtain the final influence weight of the neighbor node on the first child node. Wherein, determining the correction value for the initial prediction weight based on the first scaling ratio and the initial prediction weight includes: Based on the third formula, the first scaling factor, and the initial prediction weights, a correction value for the initial prediction weights is determined, wherein the third formula is: δw=weight_org×scale>>precisebit-weight_org Where δw represents the correction value, weight_org represents the initial prediction weight, scale represents the first scaling ratio, and precisebit is the number of shift bits used to ensure the accuracy of scale. Based on this embodiment, the step of correcting the initial prediction weight according to the correction value to obtain the final prediction weight of the neighbor node for the first child node includes: summing the correction value and the initial prediction weight to obtain the final prediction weight of the neighbor node for the first child node, that is: weight_final = weight_org + δw Wherein, weight_final represents the final predicted weight of the neighbor node for the first child node. In the second embodiment, the weight correction module 1002 can be specifically used to: determine the final predicted weight of the neighbor node to the first child node according to the fourth formula, the first scaling ratio and the initial prediction weight. The fourth formula is: weight_final=weight_org×scale>>precisebit Wherein, weight_final represents the final predicted weight, weight_org represents the initial predicted weight, scale represents the first scaling ratio, and precisebit is the number of shift bits used to ensure the precision of scale. In the third and fourth formulas above, ">>" is a shift symbol used to ensure accuracy. It is understood that scaling and shifting the initial prediction weights according to the first scaling ratio can yield the final prediction weights more quickly. It is understandable that the above-mentioned correction of the initial prediction weight is performed on each neighbor node of the first child node. Finally, the intra-frame prediction attribute value is obtained by weighted averaging of the corrected final prediction weight and the attribute values ​​of the neighbor nodes for each first child node. The intra-frame prediction module 1003 is used to predict the first child node based on the final prediction weights of each neighbor node for the first child node. In some embodiments, the intra-frame prediction module 1003 may specifically be used for: The intra-frame prediction attribute value of the first child node is determined based on the attribute values ​​of each neighbor node and the final prediction weight of each neighbor node for the first child node. In some embodiments, the intra-frame prediction attribute value of the first child node is: the attribute value of each of the neighboring nodes and the weighted average of the final prediction weights of each of the neighboring nodes for the first child node. In some embodiments, the intra-frame prediction module 1003 can also be used to: decode the input attribute-encoded bitstream to obtain the reconstruction residual of the first child node, and determine the reconstruction attribute value of the first child node based on the intra-frame prediction attribute value of the first child node and the reconstruction residual. In some embodiments, the intra-frame prediction module 1003 may specifically be used for: The input attribute-encoded bitstream is decoded and dequantized to obtain the residual transform coefficients of the first child node; The inverse transformation of the residual transformation coefficients of the first child node is performed using the region adaptive hierarchical transformation to obtain the reconstruction residual of the first child node. Optionally, in some embodiments, the first child node to be decoded satisfies one of the following: The first child node belongs to the nodes of the first A layers of the transformation tree; The first child node belongs to the node of the last B layer of the transformation tree; The first child node does not belong to the nodes of the first O layers of the transformation tree; The first child node does not belong to the nodes of the later P layer of the transformation tree; Among them, A, B, O, and P are greater than or equal to 0. Furthermore, the point cloud decoding device shown in Figure 10 may further include: a parameter receiving module, used to obtain a first parameter from the attribute encoded bitstream, so that the decoding end device can determine whether or not to perform cross-attribute prediction on the node. The first parameter is used to indicate one of the following: The nodes of the first A layers of the transformation tree perform cross-attribute prediction; The nodes of the last B layer of the transformation tree perform cross-attribute prediction; The nodes in the first O layers of the transformation tree do not perform cross-attribute prediction; The nodes in the last P layers of the transformation tree do not perform cross-attribute prediction. It is understandable that the parsing is performed layer by layer and node by node from top to bottom. When the bottom layer is reached, the reconstructed attribute values ​​of all nodes are obtained, thus completing the attribute decoding. The point cloud decoding device 1000 provided in this application embodiment can implement the various processes implemented in the method embodiment of FIG7 and achieve the same technical effect. To avoid repetition, it will not be described again here. As shown in Figure 11, this application embodiment also provides a communication device 1100, including a processor 1101 and a memory 1102. The memory 1102 stores a program or instructions that can run on the processor 1101. For example, when the communication device 1100 is a terminal, the program or instructions executed by the processor 1101 implement the various steps of the above-described point cloud encoding method embodiment and achieve the same technical effect. When the communication device 1100 is a network-side device, the program or instructions executed by the processor 1101 implement the various steps of the above-described point cloud encoding method embodiment and achieve the same technical effect. To avoid repetition, this will not be described again here. This application also provides a terminal, including a processor and a communication interface, wherein the communication interface is coupled to the processor, and the processor is used to run programs or instructions to implement the steps in the method embodiments shown in FIG6 or FIG7. All implementation processes and methods of the above method embodiments can be applied to this terminal embodiment and can achieve the same technical effect. The terminal can be the point cloud encoding device shown in FIG9, or the terminal can be the point cloud decoding device shown in FIG10. Specifically, FIG12 is a schematic diagram of the hardware structure of a terminal implementing an embodiment of this application. The terminal 1200 includes, but is not limited to, at least some of the following components: radio frequency unit 1201, network module 1202, audio output unit 1203, input unit 1204, sensor 1205, display unit 1206, user input unit 1207, interface unit 1208, memory 1209, and processor 1210. Those skilled in the art will understand that terminal 1200 may also include a power supply (such as a battery) for powering various components. The power supply can be logically connected to processor 1210 through a power management system, thereby enabling functions such as charging, discharging, and power consumption management through the power management system. The terminal structure shown in Figure 12 does not constitute a limitation on the terminal. The terminal may include more or fewer components than shown, or combine certain components, or have different component arrangements, which will not be elaborated here. It should be understood that, in this embodiment, the input unit 1204 may include a graphics processor 12041 and a microphone 12042. The graphics processor 12041 processes image data of still images or videos obtained by an image capture device (such as a camera) in video capture mode or image capture mode. The display unit 1206 may include a display panel 12061, which may be configured in the form of a liquid crystal display, an organic light-emitting diode, or the like. The user input unit 1207 includes a touch panel 12071 and at least one of other input devices 12072. The touch panel 12071 is also called a touch screen. The touch panel 12071 may include a touch detection device and a touch controller. Other input devices 12072 may include, but are not limited to, physical keyboards, function keys (such as volume control buttons, power buttons, etc.), trackballs, mice, and joysticks, which will not be described in detail here. In this embodiment, after receiving downlink data from the network-side device, the radio frequency unit 1201 can transmit it to the processor 1210 for processing; in addition, the radio frequency unit 1201 can send uplink data to the network-side device. Typically, the radio frequency unit 1201 includes, but is not limited to, antennas, amplifiers, transceivers, couplers, low-noise amplifiers, duplexers, etc. The memory 1209 can be used to store software programs or instructions, as well as various data. The memory 1209 may primarily include a first storage area for storing programs or instructions and a second storage area for storing data. The first storage area may store the operating system, application programs or instructions required for at least one function (such as sound playback, image playback, etc.). Furthermore, the memory 1209 may include volatile memory or non-volatile memory. The non-volatile memory may be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory can be random access memory (RAM), static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct memory bus RAM (DRRAM). The memory 1209 in this embodiment includes, but is not limited to, these and any other suitable types of memory. Processor 1210 may include one or more processing units; optionally, processor 1210 integrates an application processor and a modem processor, wherein the application processor mainly handles operations involving the operating system, user interface, and applications, and the modem processor mainly handles wireless communication signals, such as a baseband processor. It is understood that the aforementioned modem processor may also not be integrated into processor 1210. The processor 1210 is configured to, when intra-frame prediction is allowed for the current node to be encoded in the transform tree corresponding to the point cloud to be encoded, obtain the initial prediction weights of neighboring nodes for a first child node, and obtain the reference attribute values ​​of the neighboring nodes and the first child node, wherein the first child node is any child node of the current node to be encoded; determine the similarity information between the neighboring nodes and the first child node based on the reference attribute values ​​of the neighboring nodes and the first child node, and correct the initial prediction weights based on the similarity information to obtain the final prediction weights of the neighboring nodes for the first child node, wherein the similarity information is used to characterize the degree of similarity; and predict the first child node based on the final prediction weights of each neighboring node for the first child node. The terminal proposed in this application, when determining the prediction weight of the child node of the current node to be encoded, also considers the correlation between attributes by introducing the reference attribute values ​​of neighboring nodes and the child node to be encoded (the first child node), so that the obtained prediction weight is more accurate and thus can further improve the compression efficiency. Alternatively, the processor 1210 is configured to, when intra-frame prediction is allowed for the current node to be decoded in the transform tree corresponding to the point cloud to be decoded, obtain the initial prediction weights of neighboring nodes for the first child node, and obtain the reference attribute values ​​of the neighboring nodes and the first child node, wherein the first child node is any child node of the current node to be decoded; determine the similarity information between the neighboring nodes and the first child node based on the reference attribute values ​​of the neighboring nodes and the first child node, and correct the initial prediction weights based on the similarity information to obtain the final prediction weights of the neighboring nodes for the first child node, wherein the similarity information is used to characterize the degree of similarity; and predict the first child node based on the final prediction weights of each neighboring node for the first child node. The terminal proposed in this application, when determining the prediction weight of the child node of the current node to be decoded, also considers the correlation between attributes by introducing the reference attribute values ​​of neighboring nodes and the child node to be encoded (the first child node), so that the obtained prediction weight is more accurate and thus the decoding efficiency can be further improved. It is understood that the implementation process of each implementation method mentioned in this embodiment can refer to the relevant description of the method embodiment and achieve the same or corresponding technical effect. To avoid repetition, it will not be described again here. This application also provides a readable storage medium storing a program or instructions. When the program or instructions are executed by a processor, they implement the various processes of the above-described point cloud encoding method embodiments and achieve the same technical effect. To avoid repetition, they will not be described again here. The processor mentioned above is the processor in the terminal described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk. In some examples, the readable storage medium may be a non-transient readable storage medium. This application embodiment also provides a chip, which includes a processor and a communication interface. The communication interface is coupled to the processor. The processor is used to run programs or instructions to implement the various processes of the above-described point cloud encoding method embodiments and can achieve the same technical effect. To avoid repetition, it will not be described again here. It should be understood that the chip mentioned in the embodiments of this application may also be referred to as a system-on-a-chip, system chip, chip system, or system-on-a-chip, etc. This application also provides a computer program / program product, which is stored in a storage medium and executed by at least one processor to implement the various processes of the above-described point cloud encoding method or point cloud decoding method embodiments, and can achieve the same technical effect. To avoid repetition, it will not be described again here. This application embodiment also provides a communication system, including: an encoding end device and a decoding end device, wherein the encoding end device can be used to execute the steps of the point cloud encoding method shown in FIG6, and the decoding end device can be used to execute the steps of the point cloud decoding method shown in FIG7. This application embodiment also provides a chip, the chip including a processor and a communication interface, the communication interface being coupled to the processor, the processor being used to run programs or instructions to implement the point cloud encoding method as shown in FIG6, or to implement the point cloud decoding method as shown in FIG7. This application also provides a computer program / program product, which is stored in a storage medium and executed by at least one processor to implement the point cloud encoding method as shown in FIG6, or to implement the point cloud decoding method as shown in FIG7. It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element. Furthermore, it should be noted that the scope of the methods and apparatuses in the embodiments of this application is not limited to performing functions in the order shown or discussed, but may also include performing functions substantially simultaneously or in the reverse order, depending on the functions involved. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples. From the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of computer software products plus necessary general-purpose hardware platforms, and of course, they can also be implemented by hardware. The computer software product is stored in a storage medium (such as ROM, RAM, magnetic disk, optical disk, etc.) and includes several instructions to cause the terminal or network-side device to execute the methods described in the various embodiments of this application. The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other implementations under the guidance of this application without departing from the spirit and scope of the claims. All of these implementations are within the protection scope of this application.

Claims

1. A point cloud encoding method, the method comprising: If intra-frame prediction is allowed for the current node to be encoded in the transform tree corresponding to the point cloud to be encoded, the initial prediction weights of the neighboring nodes for the first child node are obtained, and the reference attribute values ​​of the neighboring nodes and the reference attribute values ​​of the first child node are obtained, wherein the first child node is any child node of the current node to be encoded. Based on the reference attribute values ​​of the neighboring nodes and the reference attribute values ​​of the first child node, the similarity information between the neighboring nodes and the first child node is determined, and the initial prediction weight is corrected based on the similarity information to obtain the final prediction weight of the neighboring nodes for the first child node. The similarity information is used to characterize the degree of similarity. The first child node is predicted based on the final prediction weights of each of the neighboring nodes for the first child node.

2. The method according to claim 1, wherein, The step of predicting the first child node based on the final prediction weights of each of the neighboring nodes for the first child node includes: The intra-frame prediction attribute value of the first child node is determined based on the attribute values ​​of each neighbor node and the final prediction weight of each neighbor node for the first child node. Based on the original attribute value and intra-predicted attribute value of the first child node, determine the intra-predicted residual of the first child node; The intra-frame prediction residual of the first child node is encoded to obtain the attribute-coded bitstream.

3. The method according to claim 1 or 2, wherein, The reference attributes of a node include at least one channel, and the similarity information includes an error. The step of determining the similarity information between the neighboring node and the first child node based on the reference attribute values ​​of the neighboring nodes and the first child node includes: Based on the reference attribute values ​​of the neighbor node and the first child node under the at least one channel, the error between the neighbor node and the first child node under the at least one channel is determined, wherein one channel corresponds to one error, the larger the error, the lower the similarity, and the smaller the error, the higher the similarity.

4. The method according to claim 3, wherein, The step of determining the error between the neighbor node and the first child node in at least one channel based on the reference attribute values ​​of the neighbor node and the first child node in at least one channel includes: Based on the first formula, the reference attribute values ​​of the neighbor node and the first child node under the at least one channel, determine the error between the neighbor node and the first child node under the at least one channel; The first formula is: error=abs(neibor_ref-curChild_ref) Wherein, error represents the error, neibor_ref represents the reference attribute value of the neighbor node, and curChild_ref represents the reference attribute value of the first child node.

5. The method according to any one of claims 1-4, wherein, The step of correcting the initial prediction weights based on the similarity information to obtain the final prediction weights of the neighbor nodes for the first child node includes: A first scaling ratio is determined for the initial prediction weights based on the similarity information, wherein the higher the similarity, the larger the first scaling ratio, and the lower the similarity, the smaller the first scaling ratio. The initial prediction weights are corrected according to the first scaling ratio to obtain the final prediction weights of the neighbor nodes for the first child node.

6. The method according to claim 5, wherein, The node's reference attributes include at least one channel, and the similarity information includes an error, wherein determining a first scaling factor for the initial prediction weights based on the similarity information includes: From the M reference attribute threshold intervals corresponding to each of the at least one channel, determine the first reference attribute threshold interval where the error under the at least one channel is located, and obtain at least one first reference attribute threshold interval, wherein the error is inversely proportional to the similarity information; A first scaling factor for the initial prediction weights is determined based on the threshold range of the at least one first reference attribute.

7. The method according to claim 6, wherein, Determining the first scaling ratio for the initial prediction weights based on the at least one first reference attribute threshold range includes: At least one scaling ratio for the initial prediction weight is determined based on the at least one first reference attribute threshold interval, wherein one reference attribute threshold interval corresponds to one scaling ratio; A first scaling factor for the initial prediction weights is determined based on the at least one scaling factor.

8. The method according to claim 6 or 7, wherein, Before determining the first reference attribute threshold interval in which the error under the at least one channel lies from the M reference attribute threshold intervals corresponding to each of the at least one channel, the method further includes: Based on the second formula, the pre-set N thresholds, and the reference attribute values ​​of the first child node under the at least one channel, determine the N reference attribute thresholds corresponding to each of the at least one channel; Based on the N reference attribute thresholds corresponding to each of the at least one channel, determine the M reference attribute threshold ranges corresponding to each of the at least one channel; Wherein, the N thresholds are N positive integers that increase sequentially, and the second formula is: refAttrThi=curChild_ref×thi Where refAttrThi represents the threshold of the i-th reference attribute, curChild_ref represents the reference attribute value of the first child node, and thi represents the i-th threshold among the N pre-set thresholds, i = 1, 2, ..., N.

9. The method according to any one of claims 6-8, wherein, The method further includes: The first information is carried in the attribute-encoded bitstream and sent to the decoding device; The first information includes at least one of the following: N thresholds are used to determine the threshold intervals of the M reference attributes corresponding to each of the at least one channel; Each of the at least one channel corresponds to M reference attribute threshold intervals; The scaling ratio corresponding to each of the M reference attribute threshold intervals of the at least one channel.

10. The method according to any one of claims 6-9, wherein, Before determining the first reference attribute threshold interval in which the error under the at least one channel lies from the M reference attribute threshold intervals corresponding to each of the at least one channel, the method further includes: The errors in the at least one channel are each multiplied by a first factor, wherein the first factor is used to ensure the accuracy of the errors.

11. The method according to any one of claims 5-10, wherein, The step of correcting the initial prediction weights according to the first scaling ratio to obtain the final prediction weights of the neighbor nodes for the first child node includes: Based on the first scaling ratio and the initial prediction weight, determine a correction value for the initial prediction weight; The initial influence weight is adjusted based on the correction value to obtain the final influence weight of the neighbor node on the first child node.

12. The method according to claim 11, wherein, The step of determining a correction value for the initial prediction weights based on the first scaling ratio and the initial prediction weights includes: Based on the third formula, the first scaling factor, and the initial prediction weights, a correction value for the initial prediction weights is determined, wherein the third formula is: δw=weight_org×scale>>precisebit-weight_org Where δw represents the correction value, weight_org represents the initial prediction weight, scale represents the first scaling ratio, and precisebit is the number of shift bits used to ensure the accuracy of scale.

13. The method according to claim 11 or 12, wherein, The step of correcting the initial prediction weights based on the correction value to obtain the final prediction weights of the neighbor nodes for the first child node includes: The final prediction weight of the neighbor node for the first child node is obtained by summing the correction value and the initial prediction weight.

14. The method according to any one of claims 5-10, wherein, The step of correcting the initial prediction weights according to the first scaling ratio to obtain the final prediction weights of the neighbor nodes for the first child node includes: Based on the fourth formula, the first scaling factor, and the initial prediction weights, the final prediction weights of the neighbor nodes for the first child nodes are determined, wherein the fourth formula is: weight_final=weight_org×scale>>precisebit Wherein, weight_final represents the final predicted weight, weight_org represents the initial predicted weight, scale represents the first scaling ratio, and precisebit is the number of shift bits used to ensure the precision of scale.

15. The method according to any one of claims 1-14, wherein, The step of encoding the intra-frame prediction residual of the first child node to obtain the attribute-coded bitstream includes: Perform a region-adaptive hierarchical transformation on the intra-frame prediction residual of the first child node to obtain the residual transformation coefficients of the first child node. The residual transform coefficients of the first child node are encoded to obtain the attribute-encoded bitstream.

16. The method according to any one of claims 1-15, wherein, The first child node to be encoded satisfies one of the following: The first child node belongs to the nodes of the first A layers of the transformation tree; The first child node belongs to the node of the last B layer of the transformation tree; The first child node does not belong to the nodes of the first O layers of the transformation tree; The first child node does not belong to the nodes of the later P layer of the transformation tree; Among them, A, B, O, and P are greater than or equal to 0.

17. The method according to claim 16, wherein, The method further includes: The first parameter is carried in the attribute-encoded bitstream and sent to the decoding device; The first parameter is used to indicate one of the following: The nodes of the first A layers of the transformation tree perform cross-attribute prediction; The nodes of the last B layer of the transformation tree perform cross-attribute prediction; The nodes in the first O layers of the transformation tree do not perform cross-attribute prediction; The nodes in the last P layers of the transformation tree do not perform cross-attribute prediction.

18. A point cloud decoding method, the method comprising: If intra-frame prediction is allowed for the current node to be decoded in the transformation tree corresponding to the point cloud to be decoded, the initial prediction weights of the neighboring nodes for the first child node are obtained, and the reference attribute values ​​of the neighboring nodes and the reference attribute values ​​of the first child node are obtained, wherein the first child node is any child node of the current node to be decoded. Based on the reference attribute values ​​of the neighboring nodes and the reference attribute values ​​of the first child node, the similarity information between the neighboring nodes and the first child node is determined, and the initial prediction weight is corrected based on the similarity information to obtain the final prediction weight of the neighboring nodes for the first child node. The similarity information is used to characterize the degree of similarity. The first child node is predicted based on the final prediction weights of each of the neighboring nodes for the first child node.

19. The method according to claim 18, wherein, The step of predicting the first child node based on the final prediction weights of each of the neighboring nodes for the first child node includes: The intra-frame prediction attribute value of the first child node is determined based on the attribute values ​​of each neighbor node and the final prediction weight of each neighbor node for the first child node. The reconstruction residual of the first child node is obtained by decoding the input attribute-encoded bitstream; The reconstruction attribute value of the first child node is determined based on the intra-frame prediction attribute value and the reconstruction residual of the first child node.

20. The method according to claim 18 or 19, wherein, The reference attributes of a node include at least one channel, and the similarity information includes an error. The step of determining the similarity information between the neighboring node and the first child node based on the reference attribute values ​​of the neighboring nodes and the first child node includes: Based on the reference attribute values ​​of the neighbor node and the first child node under the at least one channel, the error between the neighbor node and the first child node under the at least one channel is determined, wherein one channel corresponds to one error, the larger the error, the lower the similarity, and the smaller the error, the higher the similarity.

21. The method according to claim 20, wherein, The step of determining the error between the neighbor node and the first child node in at least one channel based on the reference attribute values ​​of the neighbor node and the first child node in at least one channel includes: Based on the first formula, the reference attribute values ​​of the neighbor node and the first child node under the at least one channel, determine the error between the neighbor node and the first child node under the at least one channel; The first formula is: error=abs(neibor_ref-curChild_ref) Wherein, error represents the error, neibor_ref represents the reference attribute value of the neighbor node, and curChild_ref represents the reference attribute value of the first child node.

22. The method according to any one of claims 18-19, wherein, The step of correcting the initial prediction weights based on the similarity information to obtain the final prediction weights of the neighbor nodes for the first child node includes: A first scaling ratio is determined for the initial prediction weights based on the similarity information, wherein the higher the similarity, the larger the first scaling ratio, and the lower the similarity, the smaller the first scaling ratio. The initial prediction weights are corrected according to the first scaling ratio to obtain the final prediction weights of the neighbor nodes for the first child node.

23. The method according to claim 22, wherein, The node's reference attributes include at least one channel, and the similarity information includes an error, wherein determining a first scaling factor for the initial prediction weights based on the similarity information includes: From the M reference attribute threshold intervals corresponding to each of the at least one channel, determine the first reference attribute threshold interval where the error under the at least one channel is located, and obtain at least one first reference attribute threshold interval, wherein the error is inversely proportional to the similarity information; A first scaling factor for the initial prediction weights is determined based on the threshold range of the at least one first reference attribute.

24. The method according to claim 23, wherein, Determining the first scaling ratio for the initial prediction weights based on the at least one first reference attribute threshold range includes: At least one scaling ratio for the initial prediction weight is determined based on the at least one first reference attribute threshold interval, wherein one reference attribute threshold interval corresponds to one scaling ratio; A first scaling factor for the initial prediction weights is determined based on the at least one scaling factor.

25. The method according to claim 23 or 24, wherein, Before determining the first reference attribute threshold interval in which the error under the at least one channel lies from the M reference attribute threshold intervals corresponding to each of the at least one channel, the method further includes: Based on the second formula, the pre-set N thresholds, and the reference attribute values ​​of the first child node under the at least one channel, determine the N reference attribute thresholds corresponding to each of the at least one channel; Based on the N reference attribute thresholds corresponding to each of the at least one channel, determine the M reference attribute threshold ranges corresponding to each of the at least one channel; Wherein, the N thresholds are N positive integers that increase sequentially, and the second formula is: refAttrThi=curChild_ref×thi Where refAttrThi represents the threshold of the i-th reference attribute, curChild_ref represents the reference attribute value of the first child node, and thi represents the i-th threshold among the N pre-set thresholds, i = 1, 2, ..., N.

26. The method according to any one of claims 23-25, wherein, The method further includes: Receive the first information carried in the attribute-encoded bitstream; The first information includes at least one of the following: N thresholds are used to determine the threshold intervals of the M reference attributes corresponding to each of the at least one channel; Each of the at least one channel corresponds to M reference attribute threshold intervals; The scaling ratio corresponding to each of the M reference attribute threshold intervals of the at least one channel.

27. The method according to any one of claims 23-26, wherein, Before determining the first reference attribute threshold interval in which the error under the at least one channel lies from the M reference attribute threshold intervals corresponding to each of the at least one channel, the method further includes: The errors in the at least one channel are each multiplied by a first factor, wherein the first factor is used to ensure the accuracy of the errors.

28. The method according to any one of claims 22-27, wherein, The step of correcting the initial prediction weights according to the first scaling ratio to obtain the final prediction weights of the neighbor nodes for the first child node includes: Based on the first scaling ratio and the initial prediction weight, determine a correction value for the initial prediction weight; The initial influence weight is adjusted based on the correction value to obtain the final influence weight of the neighbor node on the first child node.

29. The method according to claim 28, wherein, The step of determining a correction value for the initial prediction weights based on the first scaling ratio and the initial prediction weights includes: Based on the third formula, the first scaling factor, and the initial prediction weights, a correction value for the initial prediction weights is determined, wherein the third formula is: δw=weight_org×scale>>precisebit-weight_org Where δw represents the correction value, weight_org represents the initial prediction weight, scale represents the first scaling ratio, and precisebit is the number of shift bits used to ensure the accuracy of scale.

30. The method according to claim 28 or 29, wherein, The step of correcting the initial prediction weights based on the correction value to obtain the final prediction weights of the neighbor nodes for the first child node includes: The final prediction weight of the neighbor node for the first child node is obtained by summing the correction value and the initial prediction weight.

31. The method according to any one of claims 22-27, wherein, The step of correcting the initial prediction weights according to the first scaling ratio to obtain the final prediction weights of the neighbor nodes for the first child node includes: Based on the fourth formula, the first scaling factor, and the initial prediction weights, the final prediction weights of the neighbor nodes for the first child nodes are determined, wherein the fourth formula is: weight_final=weight_org×scale>>precisebit Wherein, weight_final represents the final predicted weight, weight_org represents the initial predicted weight, scale represents the first scaling ratio, and precisebit is the number of shift bits used to ensure the precision of scale.

32. The method according to any one of claims 18-31, wherein, The step of decoding the input attribute-encoded bitstream to obtain the reconstruction residual of the first child node includes: The input attribute-encoded bitstream is decoded and dequantized to obtain the residual transform coefficients of the first child node; The inverse transformation of the residual transformation coefficients of the first child node is performed using the region adaptive hierarchical transformation to obtain the reconstruction residual of the first child node.

33. The method according to any one of claims 18-32, wherein, The first child node to be decoded satisfies one of the following: The first child node belongs to the nodes of the first A layers of the transformation tree; The first child node belongs to the node of the last B layer of the transformation tree; The first child node does not belong to the nodes of the first O layers of the transformation tree; The first child node does not belong to the nodes of the later P layer of the transformation tree; Among them, A, B, O, and P are greater than or equal to 0.

34. The method according to claim 33, wherein, The method further includes: Obtain the first parameter from the attribute-encoded bitstream; The first parameter is used to indicate one of the following: The nodes of the first A layers of the transformation tree perform cross-attribute prediction; The nodes of the last B layer of the transformation tree perform cross-attribute prediction; The nodes in the first O layers of the transformation tree do not perform cross-attribute prediction; The nodes in the last P layers of the transformation tree do not perform cross-attribute prediction.

35. A point cloud encoding device, comprising: The acquisition module is used to acquire the initial prediction weights of the neighboring nodes to the first child node in the transform tree corresponding to the point cloud to be encoded, provided that intra-frame prediction is allowed for the current node to be encoded; and to acquire the reference attribute values ​​of the neighboring nodes and the reference attribute values ​​of the first child node, wherein the first child node is any child node of the current node to be encoded. The weight correction module determines the similarity information between the neighbor node and the first child node based on the reference attribute values ​​of the neighbor node and the reference attribute values ​​of the first child node, and corrects the initial prediction weight based on the similarity information to obtain the final prediction weight of the neighbor node for the first child node. The similarity information is used to characterize the degree of similarity. The intra-frame prediction module is used to predict the first child node based on the final prediction weights of each of the neighboring nodes for the first child node.

36. A point cloud decoding device, comprising: The acquisition module is used to acquire the initial prediction weights of the neighboring nodes to the first child node in the transform tree corresponding to the point cloud to be decoded, provided that intra-frame prediction is allowed for the current node to be decoded; and to acquire the reference attribute values ​​of the neighboring nodes and the reference attribute values ​​of the first child node, wherein the first child node is any child node of the current node to be decoded. The weight correction module is used to determine the similarity information between the neighbor node and the first child node based on the reference attribute values ​​of the neighbor node and the reference attribute values ​​of the first child node, and to correct the initial prediction weight based on the similarity information to obtain the final prediction weight of the neighbor node for the first child node. The similarity information is used to characterize the degree of similarity. The intra-frame prediction module is used to predict the first child node based on the final prediction weights of each of the neighboring nodes for the first child node.

37. A communication device comprising a processor and a memory, the memory storing a program or instructions executable on the processor, the program or instructions, when executed by the processor, implementing the steps of the point cloud encoding method as claimed in any one of claims 1 to 34.

38. A readable storage medium storing a program or instructions that, when executed by a processor, implement the point cloud encoding method as described in any one of claims 1 to 34.

Images