Three-dimensional Gaussian data processing method and device, equipment and medium

Abstract

The invention discloses a three-dimensional Gaussian data processing method and device, equipment and a medium, and relates to the technical field of three-dimensional Gaussian data processing, and the method comprises the steps: obtaining a three-dimensional Gaussian coding result outputted by a coding end, and the three-dimensional Gaussian coding result comprises coded three-dimensional Gaussian splash representation and packaged data cutting information; the encoded three-dimensional Gaussian splash representation is obtained by encoding the original three-dimensional Gaussian splash representation by the encoding end, and the original three-dimensional Gaussian splash representation is obtained by performing three-dimensional Gaussian splash training modeling processing on the target shooting content by the encoding end; the encapsulated data cutting information is obtained by encapsulating target data cutting information by the encoding end in a preset encapsulation mode, and is used for indicating cutting operation; analyzing the packaged data cutting information to obtain target data cutting information, and decoding the encoded three-dimensional Gaussian splash representation to obtain an original three-dimensional Gaussian splash representation; and cutting the original three-dimensional Gaussian splash representation to obtain a cut three-dimensional Gaussian splash representation, and transmitting the cut three-dimensional Gaussian splash representation to a preset user interface for display.

Description

Technical Field

[0001] This invention relates to the field of three-dimensional Gaussian data processing technology, and particularly to a three-dimensional Gaussian data processing method, apparatus, device, and medium. Background Technology

[0002] 3D Gaussian data is a special data structure used to represent 3D scenes. When using 3D Gaussian data to represent a 3D scene, millions of Gaussian primitives are typically required to ensure a realistic reconstruction of the scene. Therefore, representing a 3D scene using 3D Gaussian data often implies a huge amount of stored data. To address this challenge, academia and industry have been continuously exploring methods to reduce the data volume of 3D Gaussian scenes, including reducing the number of Gaussian primitives and compressing the data. In existing technologies, cropping 3D Gaussian data is a common method to reduce the data volume of 3D Gaussian representations. In practice, the following objective constraints exist: There are various training methods for generating 3D Gaussian scenes, resulting in inconsistent redundancy of 3D Gaussian primitives in the original scene; 3D Gaussian data from different subjects cannot be processed uniformly, leading to compatibility issues; There are also various algorithms for cropping the original 3D Gaussian scene, and the resulting rendering effect varies depending on the degree of cropping. For example, even with a 10% cropping of 3D Gaussian primitives, algorithm A might produce a rendering effect close to the original scene, while algorithm B might significantly reduce the viewing experience, leading to uncontrollable cropping effects and low efficiency and accuracy.

[0003] In conclusion, optimizing 3D Gaussian data processing methods to address the issues of data processing compatibility, cropping efficiency, and accuracy is a pressing problem that needs to be solved. Summary of the Invention

[0004] In view of this, the purpose of this invention is to provide a three-dimensional Gaussian data processing method, apparatus, device, and medium, which can optimize the three-dimensional Gaussian data processing method to solve the problems of data processing compatibility, cropping efficiency, and accuracy. The specific solution is as follows:

[0005] Firstly, this application provides a three-dimensional Gaussian data processing method, applied at the decoding end, including:

[0006] The encoding process obtains the 3D Gaussian encoding result output by the encoding end; the 3D Gaussian encoding result includes the encoded 3D Gaussian splash representation and the encapsulated data cropping information; the encoded 3D Gaussian splash representation is the encoding result obtained by the encoding end encoding the original 3D Gaussian splash representation, and the original 3D Gaussian splash representation is the processing result obtained by the encoding end performing 3D Gaussian splash training modeling on the target captured content; the encapsulated data cropping information is the information obtained by the encoding end encapsulating the target data cropping information through a preset encapsulation method, and the target data cropping information is used to indicate the cropping operation on the original 3D Gaussian splash representation;

[0007] The encapsulated data pruning information in the three-dimensional Gaussian encoding result is parsed to obtain the target data pruning information, and the encoded three-dimensional Gaussian splash representation in the three-dimensional Gaussian encoding result is decoded to obtain the original three-dimensional Gaussian splash representation;

[0008] The original 3D Gaussian splash representation is cropped according to the target data cropping information to obtain a cropped 3D Gaussian splash representation. The cropped 3D Gaussian splash representation is then transmitted to a preset user interface for display, thus completing the presentation of the target captured content.

[0009] Optionally, the target data pruning information includes pruning constraint information, pruning operation information, and pruning marker information; the pruning constraint information includes indication information characterizing whether the original 3D Gaussian splash representation is allowed to be pruned, and the range of the number of 3D Gaussian primitives allowed to be pruned in the original 3D Gaussian splash representation; the pruning operation information includes identification information characterizing whether the 3D Gaussian primitive has been pruned, and the number of primitives of the 3D Gaussian primitive that has been pruned; the pruning marker information includes primitive information of the 3D Gaussian primitives that are currently not pruned but are allowed to be pruned.

[0010] Optionally, the primitive information includes the coordinate position information of the three-dimensional Gaussian primitive, the hierarchical identifier associated with the three-dimensional Gaussian primitive, and the attribute component identifier.

[0011] Optionally, the step of trimming the original 3D Gaussian splash representation according to the target data trimming information to obtain the trimmed 3D Gaussian splash representation includes:

[0012] If the clipping mark information is not present in the target data clipping information, then based on the clipping constraint information and the clipping operation information, and according to the device performance and network bandwidth, the original three-dimensional Gaussian splash representation is clipped using a preset clipping algorithm, and the clipping results are reordered to obtain the clipped three-dimensional Gaussian splash representation.

[0013] If the target data clipping information contains the clipping mark information, then the original three-dimensional Gaussian splash representation is clipped based on the device performance and network bandwidth using the clipping mark information, and the clipping results are reordered to obtain the clipped three-dimensional Gaussian splash representation.

[0014] Secondly, this application provides a three-dimensional Gaussian data processing method, applied at the encoding end, including:

[0015] The target image content is subjected to 3D Gaussian splash training modeling processing to determine the processing result as the original 3D Gaussian splash representation, and the original 3D Gaussian splash representation is encoded to determine the encoded 3D Gaussian splash representation.

[0016] The target data clipping information is encapsulated using a preset encapsulation method to obtain the encapsulated data clipping information; the target data clipping information is used to indicate the clipping operation on the original three-dimensional Gaussian splash representation;

[0017] The output includes a 3D Gaussian encoding result comprising the encoded 3D Gaussian splash representation and the encapsulated data cropping information. This allows the decoding end to parse the encapsulated data cropping information in the 3D Gaussian encoding result to obtain the target data cropping information. The decoder then decodes the encoded 3D Gaussian splash representation in the 3D Gaussian encoding result to obtain the original 3D Gaussian splash representation. Furthermore, the decoder performs cropping processing on the original 3D Gaussian splash representation based on the target data cropping information to obtain a cropped 3D Gaussian splash representation. This cropped 3D Gaussian splash representation is then transmitted to a preset user interface for display, completing the presentation of the target captured content.

[0018] Optionally, the step of encapsulating the target data trimming information using a preset encapsulation method to obtain the encapsulated data trimming information includes:

[0019] The target data pruning information is encapsulated into a target SEI message, so that the target SEI message is identified as the encapsulated data pruning information;

[0020] Accordingly, the output includes the encoded 3D Gaussian sputtering representation and the 3D Gaussian encoding result of the encapsulated data clipping information, including:

[0021] The encoded 3D Gaussian splash representation is output as a video stream, and the target SEI message is embedded in the video stream.

[0022] Optionally, the step of encapsulating the target data trimming information using a preset encapsulation method to obtain the encapsulated data trimming information includes:

[0023] The encoded 3D Gaussian splash representation is encapsulated into a target ISOBMFF file, and the target data clipping information is encapsulated into a clipping metadata track of the target ISOBMFF file, so that the clipping metadata track is determined as the encapsulated data clipping information;

[0024] Accordingly, the output includes the encoded 3D Gaussian sputtering representation and the 3D Gaussian encoding result of the encapsulated data clipping information, including:

[0025] The output includes the target ISOBMFF file containing the trimmed metadata track.

[0026] Thirdly, this application provides a three-dimensional Gaussian data processing device, applied at the decoding end, comprising:

[0027] The result acquisition module is used to acquire the 3D Gaussian encoding result output by the encoding end; the 3D Gaussian encoding result includes the encoded 3D Gaussian splash representation and the encapsulated data cropping information; the encoded 3D Gaussian splash representation is the encoding result obtained by the encoding end encoding the original 3D Gaussian splash representation, and the original 3D Gaussian splash representation is the processing result obtained by the encoding end performing 3D Gaussian splash training modeling on the target captured content; the encapsulated data cropping information is the information obtained by the encoding end encapsulating the target data cropping information through a preset encapsulation method, and the target data cropping information is used to indicate the cropping operation on the original 3D Gaussian splash representation;

[0028] The decoding module is used to parse the encapsulated data clipping information in the three-dimensional Gaussian encoding result to obtain the target data clipping information, and to decode the encoded three-dimensional Gaussian splatter representation in the three-dimensional Gaussian encoding result to obtain the original three-dimensional Gaussian splatter representation;

[0029] The cropping module is used to crop the original three-dimensional Gaussian splash representation according to the target data cropping information to obtain a cropped three-dimensional Gaussian splash representation, and then transmit the cropped three-dimensional Gaussian splash representation to a preset user interface for display, thereby completing the presentation operation of the target captured content.

[0030] Fourthly, this application provides an electronic device, comprising:

[0031] Memory, used to store computer programs;

[0032] A processor is used to execute the computer program to implement the aforementioned three-dimensional Gaussian data processing method.

[0033] Fifthly, this application provides a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the aforementioned three-dimensional Gaussian data processing method.

[0034] In this application, a 3D Gaussian encoding result output by the encoding end is obtained; the 3D Gaussian encoding result includes an encoded 3D Gaussian splash representation and encapsulated data cropping information; the encoded 3D Gaussian splash representation is the encoding result obtained by the encoding end encoding the original 3D Gaussian splash representation, and the original 3D Gaussian splash representation is the processing result obtained by the encoding end performing 3D Gaussian splash training and modeling on the target captured content; the encapsulated data cropping information is the information obtained by the encoding end encapsulating the target data cropping information through a preset encapsulation method, and the target data cropping information... This is used to indicate the cropping operation of the original 3D Gaussian splash representation; to parse the encapsulated data cropping information in the 3D Gaussian encoding result to obtain the target data cropping information, and to decode the encoded 3D Gaussian splash representation in the 3D Gaussian encoding result to obtain the original 3D Gaussian splash representation; to perform cropping processing on the original 3D Gaussian splash representation according to the target data cropping information to obtain the cropped 3D Gaussian splash representation, and to transmit the cropped 3D Gaussian splash representation to a preset user interface for display, thereby completing the presentation operation of the target captured content. As can be seen from the above, the encoding end of this application first performs 3D Gaussian splash training modeling on the target captured content to obtain the original 3D Gaussian splash representation, and then encodes it to generate the encoded 3D Gaussian splash representation. At the same time, the target data clipping information used to indicate the clipping operation is encapsulated through a preset encapsulation method to obtain the encapsulated data clipping information. The two together constitute the 3D Gaussian encoding result and are output by the encoding end. Subsequently, the decoding end first parses the encapsulated data clipping information in the encoding result to extract the target data clipping information, and at the same time decodes the encoded 3D Gaussian splash representation to recover the original 3D Gaussian splash representation. Finally, according to the target data clipping information, the original 3D Gaussian splash representation is clipped to obtain the clipped 3D Gaussian representation, thereby realizing the transmission or presentation of the clipped 3D Gaussian representation and completing the presentation operation of the target captured content. In this way, through the process described in this application, the encoded output content, standardized parsing and decoding steps, and the clipping operation based on standardized clipping information are clearly defined. This ensures the uniformity of 3D Gaussian data processing and transmission, avoids compatibility issues when different subjects process the data, and ensures controllable clipping effects through precise clipping instructions. It effectively balances data volume optimization and scene presentation quality, while adapting to different transmission scenarios and terminal requirements. This improves the flexibility and reliability of 3D Gaussian data applications, thereby optimizing 3D Gaussian data processing methods to solve the problems of data processing compatibility, clipping efficiency, and accuracy. Attached Figure Description

[0035] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0036] Figure 1 This is a flowchart of a three-dimensional Gaussian data processing method disclosed in this application;

[0037] Figure 2 This is a flowchart and timing diagram of a Gaussian data processing method for the encoding end disclosed in this application;

[0038] Figure 3 This is a flowchart of a specific three-dimensional Gaussian data processing method disclosed in this application;

[0039] Figure 4 This is a schematic diagram of a primitive three-dimensional Gaussian splash representation disclosed in this application;

[0040] Figure 5 This is a schematic diagram of the arrangement of planar frames disclosed in this application;

[0041] Figure 6 This is a schematic diagram of the structure of a three-dimensional Gaussian data processing device disclosed in this application;

[0042] Figure 7 This is a schematic diagram of the structure of a three-dimensional Gaussian data processing device disclosed in this application;

[0043] Figure 8 This is a structural diagram of an electronic device disclosed in this application. Detailed Implementation

[0044] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0045] Representing 3D scenes using 3D Gaussian data often implies a massive amount of stored data. Faced with this challenge, both academia and industry are constantly exploring methods to reduce the data volume of 3D Gaussian scenes, including reducing the number of 3D Gaussian primitives and compressing the 3D Gaussian primitive data. In existing technologies, cropping 3D Gaussian data is a common method to reduce the amount of data in 3D Gaussian representation. However, in practice, the following objective constraints exist: There are various training methods for generating 3D Gaussian scenes, resulting in inconsistent redundancy of 3D Gaussian primitives in the original scene; 3D Gaussian data produced by different entities cannot be processed uniformly, leading to compatibility issues; There are various algorithms for cropping the original 3D Gaussian scene, and the resulting rendering effect varies depending on the degree of cropping. For example, even with a 10% cropping of 3D Gaussian primitives, algorithm A might produce a rendering close to the original scene, while algorithm B might significantly degrade the viewing experience, leading to uncontrollable cropping effects and low cropping efficiency and accuracy.

[0046] To overcome the aforementioned technical problems, this application provides a three-dimensional Gaussian data processing method that can optimize the three-dimensional Gaussian data processing method to solve the problems of data processing compatibility, clipping efficiency, and accuracy.

[0047] See Figure 1 As shown, this embodiment of the invention discloses a three-dimensional Gaussian data processing method, applied at the decoding end, including:

[0048] Step S11: Obtain the 3D Gaussian encoding result output by the encoding end; the 3D Gaussian encoding result includes the encoded 3D Gaussian splash representation and the encapsulated data cropping information; the encoded 3D Gaussian splash representation is the encoding result obtained by the encoding end encoding the original 3D Gaussian splash representation, and the original 3D Gaussian splash representation is the processing result obtained by the encoding end performing 3D Gaussian splash training modeling on the target captured content; the encapsulated data cropping information is the information obtained by the encoding end encapsulating the target data cropping information through a preset encapsulation method, and the target data cropping information is used to indicate the cropping operation on the original 3D Gaussian splash representation.

[0049] In this embodiment, the 3D Gaussian encoding result output by the encoding end is obtained. This result includes the encoded 3D Gaussian splash representation and the encapsulated data cropping information. The encoded 3D Gaussian splash representation is the result obtained by the encoding end encoding the original 3D Gaussian splash representation, where the original 3D Gaussian splash representation is the processing result obtained by the encoding end performing 3D Gaussian splash training and modeling on the target captured content. The encapsulated data cropping information is the content obtained by the encoding end encapsulating the target data cropping information used to indicate the cropping operation of the original 3D Gaussian splash representation using a preset encapsulation method.Figure 2 The diagram shown is a flowchart of a Gaussian data processing method for the encoding end provided in this application.

[0050] It should be noted that the target data pruning information includes pruning constraint information, pruning operation information, and pruning marker information. The pruning constraint information includes indication information indicating whether the original 3D Gaussian splash representation is allowed to be pruned, and the range of the number of 3D Gaussian primitives allowed to be pruned in the original 3D Gaussian splash representation. The pruning operation information includes identifier information indicating whether the 3D Gaussian primitive has been pruned, and the number of primitives of the pruned 3D Gaussian primitive. The pruning marker information includes primitive information of 3D Gaussian primitives that are currently not pruned but are allowed to be pruned. The primitive information includes the coordinate position information of the 3D Gaussian primitive, the hierarchy identifier associated with the 3D Gaussian primitive, and the attribute component identifier. It is understood that in the prior art, pruning 3D Gaussian data is a common method to reduce the amount of 3D Gaussian representation data. In practice, the following objective constraints exist: There are various training methods for generating 3D Gaussian scenes, resulting in inconsistent redundancy of 3D Gaussian primitives in the original scene; there are also various algorithms for cropping the original 3D Gaussian scene, leading to inconsistent rendering effects corresponding to different cropping degrees. For example, even with the same 10% cropping of 3D Gaussian primitives, algorithm A might produce a rendering effect close to the original scene, while algorithm B might significantly reduce the viewing experience. Therefore, to ensure that 3D Gaussian data (3DGS data) produced by different content creators adheres to a unified and defined standard during processing, this application proposes a 3D Gaussian data processing method, including the following signaling instruction information, namely the target data cropping information, encompassing cropping constraint information, cropping operation information, and cropping marker information. The clipping constraint information includes indication information indicating whether clipping is allowed, and the range of the number of 3D Gaussian primitives that are allowed to be clipped; the clipping operation information includes identification information indicating whether a 3D Gaussian primitive has been clipped, and the number of primitives of the 3D Gaussian primitives that have been clipped; the clipping marker information includes primitive information of 3D Gaussian primitives that are not currently clipped but are allowed to be clipped, and the primitive information covers the coordinate position information of the 3D Gaussian primitive, that is, its position information within the image frame, as well as associated hierarchical identifiers and attribute component identifiers, etc., and the information of Gaussian primitives that can be clipped can be indicated by an additional placeholder map.In this way, this embodiment combines core data encoding with clipping instruction encapsulation to obtain the 3D Gaussian encoding result output by the encoding end. This enables efficient compression of 3D Gaussian data while accurately carrying the clipping control information required for subsequent data processing, providing a basis for data restoration and optimization at the decoding end, and improving the overall efficiency of 3D Gaussian data transmission and processing. By adopting a multi-dimensional and refined method for dividing and defining target data clipping information, it can provide an execution basis for the 3D Gaussian splash representation clipping processing at the decoding end, accurately controlling the scope, state, and object attributes of clipping, and ensuring the validity and integrity of the clipped 3D Gaussian data.

[0051] Step S12: Parse the encapsulated data pruning information in the three-dimensional Gaussian encoding result to obtain the target data pruning information, and decode the encoded three-dimensional Gaussian splash representation in the three-dimensional Gaussian encoding result to obtain the original three-dimensional Gaussian splash representation.

[0052] In this embodiment, the encapsulated data clipping information in the 3D Gaussian encoding result is parsed to obtain the corresponding target data clipping information. Simultaneously, the encoded 3D Gaussian sputtering representation in the encoding result is decoded to reconstruct the original 3D Gaussian sputtering representation. In this way, this embodiment simultaneously completes the clipping instruction parsing and core data decoding, laying a data foundation for subsequent precise clipping operations based on the target data clipping information, and improving the overall coherence and reliability of data processing.

[0053] Step S13: Based on the target data cropping information, the original three-dimensional Gaussian splash representation is cropped to obtain a cropped three-dimensional Gaussian splash representation. The cropped three-dimensional Gaussian splash representation is then transmitted to a preset user interface for display, thus completing the presentation of the target captured content.

[0054] In this embodiment, the original three-dimensional Gaussian splash representation is cropped according to the target data cropping information to obtain a cropped three-dimensional Gaussian splash representation. The cropped three-dimensional Gaussian splash representation is then transmitted to a preset user interface for display. Specifically, based on the user's viewing / operation behavior, several three-dimensional Gaussian primitives corresponding to the current viewpoint are obtained, and their corresponding two-dimensional images, i.e., the target captured content, are obtained through a projection algorithm. The two-dimensional image is then presented to the user, completing the presentation operation of the target captured content.

[0055] It should be noted that the process of pruning the original 3D Gaussian splash representation based on the target data pruning information to obtain the pruned 3D Gaussian splash representation is as follows: If the pruning marker information is not present in the target data pruning information, then based on the pruning constraint information and the pruning operation information, and according to device performance and network bandwidth, the original 3D Gaussian splash representation is pruned using a preset pruning algorithm, and the pruning results are reordered to obtain the pruned 3D Gaussian splash representation; if the pruning marker information is present in the target data pruning information, then the original 3D Gaussian splash representation is pruned using the pruning marker information, according to device performance and network bandwidth, and the pruning results are reordered to obtain the pruned 3D Gaussian splash representation. The preset pruning algorithm includes, but is not limited to, a greedy pruning algorithm based on contribution scoring, an adaptive pruning algorithm based on machine learning, and a hierarchical pruning algorithm based on an octree. That is, if the target data pruning information does not contain the pruning marker information, then based on the pruning constraint information and the pruning operation information, within the constraint range and considering device performance and network bandwidth, the original 3D Gaussian splash representation is pruned using a preset pruning algorithm, and the pruning results are reordered to obtain the pruned 3D Gaussian splash representation, while the target data pruning information is updated; if the target data pruning information contains the pruning marker information, then with the help of the pruning marker information, combined with device performance and network bandwidth, the original 3D Gaussian splash representation is directly pruned, the pruningable 3D Gaussian primitives are directly discarded, the processing results are reordered, the target data pruning information is updated, and the pruned 3D Gaussian splash representation is obtained. In this way, this embodiment performs 3D Gaussian data processing based on precise clipping instructions, which can effectively remove redundant information from the original data. While ensuring the 3D visual presentation effect, it reduces the cost of data transmission and rendering, and improves the display efficiency and smoothness of 3D Gaussian content in the user interface. The differentiated clipping strategy based on the presence or absence of clipping mark information can flexibly adapt to different levels of clipping information completeness. At the same time, it can perform targeted processing based on the actual conditions of the device and network, ensuring effective simplification of 3D Gaussian data while taking into account the flexibility and adaptability of data processing, and improving the transmission and display effect of the clipped data.

[0056] As can be seen from the above, in this embodiment, the encoding end first performs 3D Gaussian splash training modeling on the target captured content to obtain the original 3D Gaussian splash representation, and then generates an encoded 3D Gaussian splash representation. At the same time, the target data clipping information used to indicate the clipping operation is encapsulated through a preset encapsulation method to obtain encapsulated data clipping information. The two together constitute the 3D Gaussian encoding result and are output by the encoding end. Subsequently, the decoding end first parses the encapsulated data clipping information in the encoding result to extract the target data clipping information, and at the same time decodes the encoded 3D Gaussian splash representation to recover the original 3D Gaussian splash representation. Finally, based on the target data clipping information, the original 3D Gaussian splash representation is clipped to obtain the clipped 3D Gaussian representation, thereby realizing the transmission or presentation of the clipped 3D Gaussian representation and completing the presentation operation of the target captured content. In this way, through the above-described process of the embodiments of this application, the encoded output content, the standardized parsing and decoding steps, and the clipping operation based on standardized clipping information are clearly defined. This ensures the uniformity of 3D Gaussian data processing and transmission, avoids compatibility issues when different subjects process the data, and ensures controllable clipping effects through precise clipping instructions. It effectively balances data volume optimization and scene presentation quality, while adapting to different transmission scenarios and terminal requirements. This improves the flexibility and reliability of 3D Gaussian data applications, thereby optimizing the 3D Gaussian data processing method to solve the problems of data processing compatibility, clipping efficiency, and accuracy.

[0057] Further, see Figure 3 As shown in the embodiments of this application, a three-dimensional Gaussian data processing method is also provided, applied at the encoding end, including:

[0058] Step S21: Perform three-dimensional Gaussian splash training modeling on the target captured content to determine the processing result as the original three-dimensional Gaussian splash representation, and encode the original three-dimensional Gaussian splash representation to determine the encoded result as the encoded three-dimensional Gaussian splash representation.

[0059] In this embodiment, in order to utilize existing video encoding and decoding tools, it is usually chosen to map the three-dimensional Gaussian data to a two-dimensional plane. That is, to perform three-dimensional Gaussian splash training modeling processing on the three-dimensional Gaussian data of the target shooting content, the processing result is defined as the original three-dimensional Gaussian splash representation, and then the original three-dimensional Gaussian splash representation is encoded. The generated encoding result is determined as the encoded three-dimensional Gaussian splash representation.

[0060] It should be noted that the specific process for performing 3D Gaussian splash training and modeling on the aforementioned 3D Gaussian data is as follows: Each attribute component of the 3D Gaussian primitives in the 3D Gaussian data is considered as a pixel. The data of N 3D Gaussian primitives are mapped to K planar frames, each planar frame containing M pixels (M>=N). Here, K depends on the number of attribute components in each Gaussian primitive and the representation of the attributes. For example... Figure 4 The diagram shown illustrates a primitive three-dimensional Gaussian splash representation provided in this application. Taking this example, K = 4 + 3 + 3 + 1 + 48 = 59. Setting M to be greater than or equal to N ensures that the planar frames are rectangular. If a rectangle cannot be formed based solely on N three-dimensional Gaussian primitives, padding is required to complete the rectangle. Furthermore, it is possible to rearrange the K planar frames corresponding to different attribute components into L planar frames (K > L), such as... Figure 5 The diagram illustrates a planar frame arrangement provided in this application. This arrangement can be spatial or temporal. Furthermore, existing video encoding / decoding technologies employ layering techniques, where a video frame contains N layers, each layer representing a portion of the complete video frame's data. For example, K attribute components can be considered as K layers. Therefore, this embodiment first performs morphological transformation of the 3D Gaussian data before implementing encoding compression. This allows the 3D Gaussian data to adapt to the encoding processing logic, effectively reducing data storage and transmission volume while preserving the core features of the 3D Gaussian data, providing reliable support for subsequent decoding, reconstruction, and visualization.

[0061] Step S22: Encapsulate the target data clipping information using a preset encapsulation method to obtain encapsulated data clipping information; the target data clipping information is used to indicate the clipping operation on the original three-dimensional Gaussian splash representation.

[0062] In this embodiment, the target data clipping information used to indicate the clipping operation of the original three-dimensional Gaussian splash representation is encapsulated using a preset encapsulation method to obtain encapsulated data clipping information. It should be noted that, in order to obtain the encapsulated data clipping information, this embodiment can employ various different preset encapsulation methods to encapsulate the target data clipping information.

[0063] In one specific implementation, the encoding end can encapsulate based on SEI (Supplemental Enhancement Information) messages. The processing flow is as follows: the target data clipping information is encapsulated into a target SEI message, and the target SEI message is determined as the encapsulated data clipping information. That is, the target data clipping information is encapsulated into a target SEI message, and the target SEI message is determined as the encapsulated data clipping information. Table 1 below shows a three-dimensional Gaussian clipping message and descriptor table for SEI message extension provided in this application.

[0064] Table 1. SEI Message Extension: 3D Gaussian Clipping Messages and Descriptors

[0065] gaussian_splatting_pruning_information( payloadSize ) { Descriptor gspi_cancel_flag u(1) if( !gspi_cancel_flag ) { gspi_persistence_flag u(1) gspi_pruning_enable_flag u(1) if(gspi_pruning_enable_flag){ gspi_pruning_percent_max ue(v) gspi_pruning_percent_min ue(v) gspi_pruned_flag u(1) if(gspi_pruned_flag){ gspi_pruned_percent ue(v) } gspi_pruning_location_flag u(1) if(gspi_pruning_location_flag){ gspi_num_pruning_level u(8) gspi_attr_comp_ref_idc u(4) if(gspi_pruning_location_idc == 1){ gspi_num_attr_component u(8) } for(i = 0; i < gspi_num_pruning_level; i++ ){ gspi_percent[i] ue(v) if(gspi_attr_comp_ref_idc == 0){ gspi_region_top_left_x[i] ue(v) gspi_region_top_left_y[i] ue(v) gspi_region_width[i] ue(v) gspi_region_height[i] ue(v) } if(gspi_attr_comp_ref_idc != 0){ for(j = 0; i < gspi_num_attr_component; j++ ){ gspi_attr_comp_ref_id[i][j] ue(v) gspi_region_top_left_x[i][j] ue(v) gspi_region_top_left_y[i][j] ue(v) gspi_region_width[i][j] ue(v) gspi_region_height[i][j] ue(v) } } }

[0066] The 3D Gaussian clipping message is used to indicate information when the 3D Gaussian splash representation is clipped. When `gspi_cancel_flag` is 1, it means the current SEI message cancels the continued effect of any 3D Gaussian clipping message preceding the current SEI message in the output order; a value of 0 indicates the specific information of the current 3D Gaussian clipping message. `gspi_persistence_flag` indicates the range of the current 3D Gaussian clipping message's continued effect; a value of 0 means the current 3D Gaussian clipping message only affects the current access unit; a value of 1 means the current 3D Gaussian clipping message affects the current access unit and continues to be effective until one of the following occurs: a new video sequence begins, the current bitstream ends, or an access unit whose output order follows the current access unit contains a new 3D Gaussian clipping message. The access unit is a set of image units (PUs) belonging to different layers. This set of image units contains several interrelated encoded frames that need to be simultaneously output from the decoded image buffer (DPB). Essentially, it is a basic unit encapsulating the encoded image and associated data that need to be processed collaboratively at a certain moment. A single access unit can be simply understood as a video frame.

[0067] Furthermore, a value of 1 for `gspi_pruning_enable_flag` indicates that pruning of the 3D Gaussian splash representation is permitted within the current SEI message's effective range; a value of 0 indicates that pruning is not permitted within the current SEI message's effective range. `gspi_pruning_percent_max` and `gspi_pruning_percent_min` indicate the maximum and minimum allowed pruning range for the 3D Gaussian splash representation, respectively, in percentage or number of 3D Gaussian primitives. A value of 1 for `gspi_pruned_flag` indicates that pruning of the 3D Gaussian splash representation has been performed within the current SEI message's effective range; a value of 0 indicates that pruning has not been performed within the current SEI message's effective range. `gspi_pruned_percent` indicates the percentage of 3D Gaussian primitives in the 3D Gaussian splash representation that has been pruned, in percentage or number of 3D Gaussian primitives. `gspi_pruning_location_flag` is 1 when it indicates the location information of the croppable 3D Gaussian primitives in the 3D Gaussian splash representation; a value of 0 indicates no location information of the croppable 3D Gaussian primitives in the 3D Gaussian splash representation. `gspi_num_pruning_level` indicates the number of predefined croppable levels in the 3D Gaussian splash representation. Each croppable level corresponds to a certain number of 3D Gaussian primitives. `gspi_attr_comp_ref_idc` indicates the method of referencing specific attribute components in the 3D Gaussian splash representation. A value of 0 indicates that no identifier for each attribute component needs to be explicitly indicated, and the croppable Gaussian primitives corresponding to each attribute component are all in the same position in the corresponding video frame; a value other than 0 indicates that the identifier for each attribute component needs to be explicitly indicated. `gspi_num_attr_component` indicates the number of attribute components in the 3D Gaussian splash representation. `gspi_percent[i]` indicates the percentage of Gaussian primitives or the number of 3D Gaussian primitives corresponding to the i-th cropping level. `gspi_region_top_left_x[i]` and `gspi_region_top_left_y[i]` indicate the horizontal and vertical positions of the i-th cropping layer within the image frame, respectively. `gspi_region_width[i]` and `gspi_region_height[i]` indicate the horizontal width and vertical height of the i-th cropping layer within the image frame, respectively. `gspi_attr_comp_ref_id[i][j]` indicates the identifier of the j-th attribute component within the i-th cropping layer. The meaning of this field depends on the value of `gspi_attr_comp_ref_idc`.

[0068] It should be further noted that when the value of gspi_attr_comp_ref_idc is not 0, the meaning of the identifier of each attribute component is as follows: When the value of gspi_attr_comp_ref_idc is 1, the attribute component data of the 3D Gaussian splash is distinguished by different layers, and the value of the gspi_attr_comp_ref_id[i][j] field should correspond to the layer identifier (layer id) in the image frame; when the value of gspi_attr_comp_ref_idc is 2, the attribute component data of the 3D Gaussian splash is distinguished by different layers and YCbCr (a color space) components. The high 6 bits of this field correspond to the layer identifier (layer id) in the image frame, and the low 2 bits correspond to the YCbCr component indication. The lower 2 bits represent the luminance component Y when 0, the chrominance component Cb when 1, and the chrominance component Cr when 2. When gspi_attr_comp_ref_idc is 3, the attribute components of the 3D Gaussian splash are distinguished by different levels and sub-pictures. The higher 6 bits of this field correspond to the layer identifier (layer id) in the image frame, and the remaining lower bits correspond to the identifier of the sub-picture. When gspi_attr_comp_ref_idc is 4, the attribute components of the 3D Gaussian splash are distinguished by different tiles, and this field is valued as tile_id. It should be noted that in the above attribute component identifier index, values ​​1 to 3 mainly correspond to data partitioning in traditional video encoding and decoding; value 4, in addition to corresponding to data partitioning in traditional video encoding and decoding, can also correspond to data partitioning in geometry-based point cloud encoding and decoding, that is, dividing the region into different tiles in space. In addition, metadata in SEI messages can also be indicated in the form of parameters within the bitstream, such as the set of geometric parameters or the set of attribute parameters in geometry-based point cloud coding.

[0069] In another specific implementation, the encoding end can encapsulate based on an ISOBMFF (ISO Base Media FileFormat, an extensible container file format) file. The processing flow is as follows: the encoded 3D Gaussian splash representation is encapsulated into a target ISOBMFF file, and the target data clipping information is encapsulated into a clipping metadata track of the target ISOBMFF file, thus defining the clipping metadata track as the encapsulated data clipping information. That is, the encoded 3D Gaussian splash representation is encapsulated into a target ISOBMFF file, and the target data clipping information is simultaneously encapsulated into a clipping metadata track of the target ISOBMFF file, with this clipping metadata track defined as the encapsulated data clipping information. The clipping metadata track is used to indicate information during the clipping of the 3D Gaussian splash representation, and its sample entry syntax is as follows:

[0070] aligned(8) class 3DGSPruningInfoSampleEntry extendsMetaDataSampleEntry(' gspi ') {

[0071] }

[0072] The sample syntax is as follows:

[0073] aligned(8) 3DGSPruningInfoSample () {

[0074] unsigned int(1) gspi_cancel_flag;

[0075] if (!gspi_cancel_flag) {

[0076] unsigned int(1) gspi_persistence_flag

[0077] unsigned int(1) gspi_pruning_enable_flag

[0078] if(gspi_pruning_enable_flag){

[0079] unsigned int(16) gspi_pruning_percent_max

[0080] unsigned int(16) gspi_pruning_percent_min

[0081] unsigned int(1) gspi_pruned_flag

[0082] if(gspi_pruned_flag){

[0083] unsigned int(16) gspi_pruned_percent

[0084] }

[0085] unsigned int(1) gspi_pruning_location_flag

[0086] if(gspi_pruning_location_flag){

[0087] unsigned int(8) gspi_num_pruning_level

[0088] unsigned int(8) gspi_attr_comp_ref_idc

[0089] if(gspi_pruning_location_idc == 1){

[0090] unsigned int(8) gspi_num_attr_component

[0091] }

[0092] if(gspi_attr_comp_ref_idc == 0){

[0093] for(i = 0; i < gspi_num_pruning_level; i++ ){

[0094] unsigned int(16) gspi_percent[i]

[0095] if(gspi_attr_comp_ref_idc == 0){

[0096] unsigned int(16) gspi_region_top_left_x[i]

[0097] unsigned int(16) gspi_region_top_left_y[i]

[0098] unsigned int(16) gspi_region_width[i]

[0099] unsigned int(16) gspi_region_height[i]

[0100] }

[0101] if(gspi_attr_comp_ref_idc != 0){

[0102] for(j = 0; i < gspi_num_attr_component; j++ ){

[0103] unsigned int(16) gspi_attr_comp_ref_id[i][j]

[0104] unsigned int(16) gspi_region_top_left_x[i][j]

[0105] unsigned int(16) gspi_region_top_left_y[i][j]

[0106] unsigned int(16) gspi_region_width[i][j]

[0107] unsigned int(16) gspi_region_height[i][j]

[0108] }

[0109] }

[0110] }

[0111] }

[0112] }

[0113] It is understandable that the sample semantics in ISOBMFF are the same as those in the SEI message extension. Thus, this embodiment standardizes the encapsulation of clipping control information, enabling clipping instructions and encoded 3D Gaussian core data to form a unified transmission carrier. This ensures the integrity and accuracy of clipping information during transmission, providing a basis for precise clipping operations at the decoding end. Relying on the SEI message's clipping information encapsulation method, the transmission mechanism of supplementary enhancement information in the encoding standard can be used to achieve synchronous transmission of clipping control information and encoded 3D Gaussian splash representation, ensuring the standardization and efficiency of information transmission while avoiding interference with the core encoded data. Based on the ISOBMFF standard's multi-track encapsulation method, integrated storage and transmission of 3D Gaussian core encoded data and clipping control information can be achieved. The standardized file structure ensures the convenience and compatibility of data parsing, while the design of independent metadata tracks avoids interference from clipping information with the core data, improving the overall standardization and efficiency of data processing.

[0114] Step S23: Output a 3D Gaussian encoding result including the encoded 3D Gaussian splash representation and the encapsulated data cropping information, so that the decoding end can parse the encapsulated data cropping information in the 3D Gaussian encoding result to obtain the target data cropping information, and decode the encoded 3D Gaussian splash representation in the 3D Gaussian encoding result to obtain the original 3D Gaussian splash representation, and crop the original 3D Gaussian splash representation according to the target data cropping information to obtain the cropped 3D Gaussian splash representation, so as to transmit the cropped 3D Gaussian splash representation to a preset user interface for display, thereby completing the presentation operation of the target captured content.

[0115] It should be noted that if the encapsulated data cropping information is based on SEI message encapsulation, the processing flow for outputting the 3D Gaussian encoding result is as follows: the encoded 3D Gaussian sputtering representation is output as a video stream, and the target SEI message is embedded in the video stream. That is, the encoded 3D Gaussian sputtering representation is output as a video stream, and the target SEI message is embedded in the video stream. If the encapsulated data cropping information is based on ISOBMFF file encapsulation, the processing flow for outputting the 3D Gaussian encoding result is as follows: the target ISOBMFF file including the cropping metadata track is output. That is, the target ISOBMFF file containing the cropping metadata track is output. In this way, the method of embedding the clipping control information into the video stream output in the form of SEI messages in this embodiment can realize the synchronous transmission of 3D Gaussian core encoded data and clipping instructions. Relying on the supplementary and enhanced characteristics of SEI messages in the encoding standard, it not only ensures the standardization of clipping information transmission, but also avoids interference with the parsing of the core video stream, improving the coherence and efficiency of the transmission of encoded results and subsequent decoding processing. The integrated file output method based on the ISOBMFF standard can integrate the encoded 3D Gaussian splash representation and clipping control information into a unified file structure. With the help of independent metadata tracks, the orderly storage and retrieval of clipping information can be realized, ensuring accurate parsing and efficient processing of data at the decoding end, and improving the standardization of 3D Gaussian data transmission and visualization.

[0116] As can be seen from the above, the encoding end of this application performs 3D Gaussian splash training modeling on the target captured content, determines the processing result as the original 3D Gaussian splash representation, and then performs encoding operation on the original 3D Gaussian splash representation to obtain the encoded 3D Gaussian splash representation. At the same time, the target data clipping information used to indicate the clipping operation of the original 3D Gaussian splash representation is encapsulated through a preset encapsulation method to generate encapsulated data clipping information. Finally, a 3D Gaussian encoding result containing the encoded 3D Gaussian splash representation and the encapsulated data clipping information is output, so that the decoding end can parse the encapsulated data clipping information in the encoding result to obtain the target data clipping information, decode the encoded 3D Gaussian splash representation to restore the original 3D Gaussian splash representation, and perform clipping processing on the original 3D Gaussian representation according to the target data clipping information to obtain the clipped 3D Gaussian representation. After obtaining the clipped 3D Gaussian representation, it is transmitted to the preset user interface for display, thus completing the presentation operation of the target captured content. In this way, through the above-described process of the embodiments of this application, a processing flow integrating data conversion, encoding compression, information encapsulation, and result output is constructed, establishing a complete link for 3D Gaussian data from the encoding end to the decoding end and then to visualization. This not only achieves efficient data compression and transmission, but also ensures the 3D visual presentation effect by relying on precise cropping control information, improving the data processing efficiency and standardization of the entire process, and thus optimizing the 3D Gaussian data processing method to solve the problems of data processing compatibility, cropping efficiency, and accuracy.

[0117] Accordingly, see Figure 6 As shown in the illustration, this application also provides a three-dimensional Gaussian data processing device, applied at the decoding end, comprising:

[0118] Result acquisition module 11 is used to acquire the three-dimensional Gaussian encoding result output by the encoding end; the three-dimensional Gaussian encoding result includes the encoded three-dimensional Gaussian splash representation and the encapsulated data cropping information; the encoded three-dimensional Gaussian splash representation is the encoding result obtained by the encoding end encoding the original three-dimensional Gaussian splash representation, and the original three-dimensional Gaussian splash representation is the processing result obtained by the encoding end performing three-dimensional Gaussian splash training modeling on the target captured content; the encapsulated data cropping information is the information obtained by the encoding end encapsulating the target data cropping information through a preset encapsulation method, and the target data cropping information is used to indicate the cropping operation on the original three-dimensional Gaussian splash representation;

[0119] The decoding module 12 is used to parse the encapsulated data clipping information in the three-dimensional Gaussian encoding result to obtain the target data clipping information, and to decode the encoded three-dimensional Gaussian splatter representation in the three-dimensional Gaussian encoding result to obtain the original three-dimensional Gaussian splatter representation;

[0120] The cropping processing module 13 is used to crop the original three-dimensional Gaussian splash representation according to the target data cropping information to obtain a cropped three-dimensional Gaussian splash representation, and then transmit the cropped three-dimensional Gaussian splash representation to a preset user interface for display, thereby completing the presentation operation of the target captured content.

[0121] In some specific embodiments, the target data pruning information includes pruning constraint information, pruning operation information, and pruning marker information; the pruning constraint information includes indication information characterizing whether the original 3D Gaussian splash representation is allowed to be pruned, and the range of the number of 3D Gaussian sputtering primitives in the original 3D Gaussian sputtering representation that are allowed to be pruned; the pruning operation information includes identification information characterizing whether the 3D Gaussian primitives have been pruned, and the number of primitives of the 3D Gaussian primitives that have been pruned; the pruning marker information includes primitive information of the 3D Gaussian primitives that are currently not pruned but are allowed to be pruned.

[0122] In some specific implementations, the primitive information includes the coordinate position information of the three-dimensional Gaussian primitive, the hierarchical identifier associated with the three-dimensional Gaussian primitive, and the attribute component identifier.

[0123] In some specific embodiments, the cropping processing module 13 may specifically include:

[0124] The first trimming unit is used to trim the original three-dimensional Gaussian splash representation based on the trimming constraint information and the trimming operation information, according to the device performance and network bandwidth, and reorder the trimming results to obtain the trimmed three-dimensional Gaussian splash representation if the trimming mark information is not present in the target data trimming information.

[0125] The second trimming unit is used to trim the original three-dimensional Gaussian splash representation based on device performance and network bandwidth using the trimming mark information if the target data trimming information contains the trimming mark information, and then reorder the trimming results to obtain the trimmed three-dimensional Gaussian splash representation.

[0126] Accordingly, see Figure 7 As shown in the figure, this application embodiment also provides a three-dimensional Gaussian data processing device, applied at the encoding end, including:

[0127] The encoding module 21 is used to perform three-dimensional Gaussian splash training modeling on the target captured content, so as to determine the obtained processing result as the original three-dimensional Gaussian splash representation, and to encode the original three-dimensional Gaussian splash representation, so as to determine the obtained encoding result as the encoded three-dimensional Gaussian splash representation.

[0128] The information encapsulation module 22 is used to encapsulate the target data clipping information through a preset encapsulation method to obtain the encapsulated data clipping information; the target data clipping information is used to indicate the clipping operation on the original three-dimensional Gaussian splash representation;

[0129] The result output module 23 is used to output a 3D Gaussian encoding result including the encoded 3D Gaussian splash representation and the encapsulated data cropping information, so that the decoding end can parse the encapsulated data cropping information in the 3D Gaussian encoding result to obtain the target data cropping information, decode the encoded 3D Gaussian splash representation in the 3D Gaussian encoding result to obtain the original 3D Gaussian splash representation, and crop the original 3D Gaussian splash representation according to the target data cropping information to obtain the cropped 3D Gaussian splash representation, so as to transmit the cropped 3D Gaussian splash representation to a preset user interface for display, thereby completing the presentation operation of the target captured content.

[0130] In some specific embodiments, the information encapsulation module 22 may specifically include:

[0131] The first information encapsulation unit is used to encapsulate the target data trimming information into a target SEI message, so as to determine the target SEI message as the encapsulated data trimming information.

[0132] Accordingly, the result output module 23 may specifically include:

[0133] The message embedding unit is used to output the encoded three-dimensional Gaussian splash representation in the form of a video stream, and to embed the target SEI message in the video stream.

[0134] In some specific embodiments, the information encapsulation module 22 may specifically include:

[0135] The second information encapsulation unit is used to encapsulate the encoded three-dimensional Gaussian splash representation into a target ISOBMFF file and encapsulate the target data clipping information into a clipping metadata track of the target ISOBMFF file, so as to determine the clipping metadata track as the encapsulated data clipping information.

[0136] Accordingly, the result output module 23 may specifically include:

[0137] The file output unit is used to output the target ISOBMFF file including the trimmed metadata track.

[0138] Furthermore, embodiments of this application also disclose an electronic device, [[ID=This is a structural diagram of an electronic device 30 according to an exemplary embodiment. The content of the diagram should not be construed as limiting the scope of this application. The electronic device 30 may specifically include: at least one processor 31, at least one memory 32, a power supply 33, a communication interface 34, an input / output interface 35, and a communication bus 36. The memory 32 stores a computer program, which is loaded and executed by the processor 31 to implement the relevant steps in the three-dimensional Gaussian data processing method disclosed in any of the foregoing embodiments. Furthermore, 30 in this embodiment may specifically be an electronic computer.

[0139] In this embodiment, the power supply 33 is used to provide operating voltage for each hardware device on the electronic device 30; the communication interface 34 can create a data transmission channel between the electronic device 30 and external devices, and the communication protocol it follows can be any communication protocol applicable to the technical solution of this application, and is not specifically limited here; the input / output interface 35 is used to acquire external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs, and is not specifically limited here.

[0140] In addition, the memory 32, as a carrier for resource storage, can be a read-only memory, random access memory, disk or optical disk, etc. The resources stored thereon can include operating system 321, computer program 322, etc., and the storage method can be temporary storage or permanent storage.

[0141] The operating system 321 is used to manage and control the various hardware devices on the electronic device 30 and the computer program 322, which may be Windows Server, Netware, Unix, Linux, etc. In addition to including a computer program capable of performing the three-dimensional Gaussian data processing method executed by the electronic device 30 as disclosed in any of the foregoing embodiments, the computer program 322 may further include computer programs capable of performing other specific tasks.

[0142] Furthermore, this application also discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the aforementioned disclosed three-dimensional Gaussian data processing method. Specific steps of this method can be found in the corresponding content disclosed in the foregoing embodiments, and will not be repeated here.

[0143] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the method section.

[0144] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0145] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0146] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, 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 said element.

[0147] The technical solutions provided in this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A three-dimensional Gaussian data processing method, characterized in that, Applied to the decoding end, including: The encoding process obtains the 3D Gaussian encoding result output by the encoding end; the 3D Gaussian encoding result includes the encoded 3D Gaussian splash representation and the encapsulated data cropping information; the encoded 3D Gaussian splash representation is the encoding result obtained by the encoding end encoding the original 3D Gaussian splash representation, and the original 3D Gaussian splash representation is the processing result obtained by the encoding end performing 3D Gaussian splash training modeling on the target captured content; the encapsulated data cropping information is the information obtained by the encoding end encapsulating the target data cropping information through a preset encapsulation method, and the target data cropping information is used to indicate the cropping operation on the original 3D Gaussian splash representation; The encapsulated data pruning information in the three-dimensional Gaussian encoding result is parsed to obtain the target data pruning information, and the encoded three-dimensional Gaussian splash representation in the three-dimensional Gaussian encoding result is decoded to obtain the original three-dimensional Gaussian splash representation; The original 3D Gaussian splash representation is cropped according to the target data cropping information to obtain a cropped 3D Gaussian splash representation. The cropped 3D Gaussian splash representation is then transmitted to a preset user interface for display, thus completing the presentation of the target captured content.

2. The three-dimensional Gaussian data processing method according to claim 1, characterized in that, The target data pruning information includes pruning constraint information, pruning operation information, and pruning marker information; the pruning constraint information includes indication information indicating whether the original 3D Gaussian splash representation is allowed to be pruned, and the range of the number of 3D Gaussian sputtering primitives allowed to be pruned in the original 3D Gaussian sputtering representation; the pruning operation information includes identifier information indicating whether the 3D Gaussian primitives have been pruned, and the number of primitives of the 3D Gaussian primitives that have been pruned; the pruning marker information includes primitive information of the 3D Gaussian primitives that are currently not pruned but are allowed to be pruned.

3. The three-dimensional Gaussian data processing method according to claim 2, characterized in that, The primitive information includes the coordinate position information of the three-dimensional Gaussian primitive, the hierarchical identifier associated with the three-dimensional Gaussian primitive, and the attribute component identifier.

4. The three-dimensional Gaussian data processing method according to claim 2 or 3, characterized in that, The step of trimming the original 3D Gaussian splash representation according to the target data trimming information to obtain the trimmed 3D Gaussian splash representation includes: If the clipping mark information is not present in the target data clipping information, then based on the clipping constraint information and the clipping operation information, and according to the device performance and network bandwidth, the original three-dimensional Gaussian splash representation is clipped using a preset clipping algorithm, and the clipping results are reordered to obtain the clipped three-dimensional Gaussian splash representation. If the target data clipping information contains the clipping mark information, then the original three-dimensional Gaussian splash representation is clipped based on the device performance and network bandwidth using the clipping mark information, and the clipping results are reordered to obtain the clipped three-dimensional Gaussian splash representation.

5. A three-dimensional Gaussian data processing method, characterized in that, Applied to the encoding end, including: The target image content is subjected to 3D Gaussian splash training modeling processing to determine the processing result as the original 3D Gaussian splash representation, and the original 3D Gaussian splash representation is encoded to determine the encoded 3D Gaussian splash representation. The target data clipping information is encapsulated using a preset encapsulation method to obtain the encapsulated data clipping information; the target data clipping information is used to indicate the clipping operation on the original three-dimensional Gaussian splash representation; The output includes a 3D Gaussian encoding result comprising the encoded 3D Gaussian splash representation and the encapsulated data cropping information. This allows the decoding end to parse the encapsulated data cropping information in the 3D Gaussian encoding result to obtain the target data cropping information. The decoder then decodes the encoded 3D Gaussian splash representation in the 3D Gaussian encoding result to obtain the original 3D Gaussian splash representation. Furthermore, the decoder performs cropping processing on the original 3D Gaussian splash representation based on the target data cropping information to obtain a cropped 3D Gaussian splash representation. This cropped 3D Gaussian splash representation is then transmitted to a preset user interface for display, completing the presentation of the target captured content.

6. The three-dimensional Gaussian data processing method according to claim 5, characterized in that, The process of encapsulating the target data trimming information using a preset encapsulation method to obtain the encapsulated data trimming information includes: The target data pruning information is encapsulated into a target SEI message, so that the target SEI message is identified as the encapsulated data pruning information; Accordingly, the output includes the encoded 3D Gaussian sputtering representation and the 3D Gaussian encoding result of the encapsulated data clipping information, including: The encoded 3D Gaussian splash representation is output as a video stream, and the target SEI message is embedded in the video stream.

7. The three-dimensional Gaussian data processing method according to claim 5, characterized in that, The process of encapsulating the target data trimming information using a preset encapsulation method to obtain the encapsulated data trimming information includes: The encoded 3D Gaussian splash representation is encapsulated into a target ISOBMFF file, and the target data clipping information is encapsulated into a clipping metadata track of the target ISOBMFF file, so that the clipping metadata track is determined as the encapsulated data clipping information; Accordingly, the output includes the encoded 3D Gaussian sputtering representation and the 3D Gaussian encoding result of the encapsulated data clipping information, including: The output includes the target ISOBMFF file containing the trimmed metadata track.

8. A three-dimensional Gaussian data processing device, characterized in that, Applied to the decoding end, including: The result acquisition module is used to acquire the 3D Gaussian encoding result output by the encoding end; the 3D Gaussian encoding result includes the encoded 3D Gaussian splash representation and the encapsulated data cropping information; the encoded 3D Gaussian splash representation is the encoding result obtained by the encoding end encoding the original 3D Gaussian splash representation, and the original 3D Gaussian splash representation is the processing result obtained by the encoding end performing 3D Gaussian splash training modeling on the target captured content; the encapsulated data cropping information is the information obtained by the encoding end encapsulating the target data cropping information through a preset encapsulation method, and the target data cropping information is used to indicate the cropping operation on the original 3D Gaussian splash representation; The decoding module is used to parse the encapsulated data clipping information in the three-dimensional Gaussian encoding result to obtain the target data clipping information, and to decode the encoded three-dimensional Gaussian splatter representation in the three-dimensional Gaussian encoding result to obtain the original three-dimensional Gaussian splatter representation; The cropping module is used to crop the original three-dimensional Gaussian splash representation according to the target data cropping information to obtain a cropped three-dimensional Gaussian splash representation, and then transmit the cropped three-dimensional Gaussian splash representation to a preset user interface for display, thereby completing the presentation operation of the target captured content.

9. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor for executing the computer program to implement the three-dimensional Gaussian data processing method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, Used to store computer programs; wherein, when the computer programs are executed by a processor, they implement the three-dimensional Gaussian data processing method as described in any one of claims 1 to 7.

Images