Gaussian splash data level detail model construction method and system

Abstract

This invention discloses a method and system for constructing a hierarchical detail model of Gaussian splash data. The method includes: acquiring multi-view images of a target scene; generating Gaussian splash data based on the multi-view images; treating each Gaussian ellipsoid in the Gaussian splash data as a Gaussian ellipsoid and performing preprocessing; calculating the saliency score of each Gaussian ellipsoid; determining a threshold range based on the saliency score; dividing the Gaussian ellipsoid into multiple levels according to the threshold range to construct a multi-level detail model, where each level corresponds to a different level of detail; performing spatial meshing on the Gaussian ellipsoid of each level of the multi-level detail model and setting overlapping regions at the mesh boundaries; optimizing the overlapping regions of each mesh; standardizing and structuring the optimized Gaussian ellipsoid data, and constructing a hierarchical tile data structure to obtain a Gaussian splash model with hierarchical tile organization. This invention can improve the rendering speed of Gaussian splash data.

Description

Technical Field

[0001] This invention relates to the field of real-scene 3D reconstruction and rendering technology, and in particular to a method, system, storage medium and electronic device for constructing a Gaussian splash data hierarchical detail model. Background Technology

[0002] With the deepening of the "Digital China" and smart city construction, real-scene 3D reconstruction is transforming towards high fidelity and real-time interaction. 3D Gaussian Splatting (3DGS), with its explicit radiative field representation and efficient differentiable rasterization pipeline, combines rendering speed and quality advantages, becoming a key technology for large-scale scene construction.

[0003] However, applying 3DGS to large-scale city scenes faces significant challenges. First, there's the resource bottleneck: 3DGS models consist of a massive number of discrete ellipsoids carrying high-order spherical harmonics. Expanding the scene leads to exploding memory usage and a sharp drop in rendering frame rates, making it difficult to meet real-time interactive requirements. Second, there are scheduling and slicing challenges: existing Gaussian splashing level of detail (LOD) techniques lack a unified standard, and simple spatial partitioning fails to effectively reduce the rendering load. More critically, 3DGS ellipsoids have volumetric and semi-transparent properties; directly applying traditional 3dTiles geometric clipping strategies results in truncated boundary ellipsoids, producing a noticeable "black border effect" and disrupting visual consistency. Summary of the Invention

[0004] This invention provides a method, system, storage medium, and electronic device for constructing a Gaussian splash data hierarchical detail model, which can improve the rendering speed of Gaussian splash data and realize on-demand loading, efficient scheduling, and seamless real-time rendering of large-scale Gaussian scenes on the Web.

[0005] This invention provides a method for constructing a Gaussian splash data hierarchy detail model, including: Acquire multi-view images of the target scene, generate Gaussian splash data based on the multi-view images, treat each Gaussian in the Gaussian splash data as a Gaussian ellipsoid, and perform preprocessing. Calculate the saliency score for each Gaussian ellipsoid, determine a threshold range based on the saliency score, and divide the Gaussian ellipsoid into multiple levels based on the threshold range to construct a multi-level detail model, where each level corresponds to a different level of detail. Spatial meshing is performed on the Gaussian ellipsoid of each level of the multi-level detail model, and overlapping regions are set at the mesh boundaries; Optimize the overlapping areas of each grid; The optimized Gaussian ellipsoid data is standardized, encapsulated, and structured, and a hierarchical tile data structure is constructed to obtain a hierarchically organized Gaussian splash model.

[0006] Furthermore, according to the method for constructing a hierarchical detail model of Gaussian splash data, each Gaussian element in the Gaussian splash data is considered as a Gaussian ellipsoid and preprocessed, including: A KD tree model is constructed using the center point of the Gaussian ellipsoid as an index; Traverse all Gaussian ellipsoids, query the k nearest neighbor ellipsoids of each Gaussian ellipsoid using the KD tree model, and calculate the average distance between the current Gaussian ellipsoid and its k nearest neighbor ellipsoids; Filter and delete Gaussian ellipsoids that exceed the distance threshold; Obtain the coordinates of the center points of all Gaussian ellipsoids, calculate the arithmetic mean of all center point coordinates, and determine the corresponding mean point as the centroid of the scene; Using the scene centroid as a reference, the position and range of each Gaussian ellipsoid are scaled.

[0007] Furthermore, based on the Gaussian splash data hierarchical detail model construction method, Gaussian ellipsoids exceeding the distance threshold are filtered and deleted, as expressed by the following formula:

[0008]

[0009] in, This represents the set of ellipsoids obtained through filtering. Indicates the average distance threshold. Indicates the first An ellipsoid and The average Euclidean distance of the nearest ellipsoid Indicates the total number of ellipsoids. Represents an ellipsoid Position parameters, Represents an ellipsoid j Position parameters, This indicates the number of ellipsoids involved in the calculation of the average Euclidean distance. Indicates distance from the first The nearest ellipsoid A set of ellipsoids.

[0010] Furthermore, according to the Gaussian splash data hierarchical detail model construction method, the Gaussian ellipsoid of each level of the multi-level detail model is spatially meshed, and overlapping regions are set at the mesh boundaries, including: Obtain all Gaussian ellipsoids in the current level, calculate the projection of all Gaussian ellipsoids onto the XOY plane to obtain the two-dimensional projection region, calculate the axis-aligned bounding box of all two-dimensional projection regions, and determine the rectangular region formed by the axis-aligned bounding box as the root node coverage of the current level mesh. The grid density is determined based on the current partitioning level. The two-dimensional projection area is divided into square grid cells of fixed size. Each square grid cell serves as the spatial boundary. All Gaussian ellipsoids falling within the range are filtered and indexed. Set the overlap region extension for each grid; Assign a unique identifier to each grid.

[0011] Furthermore, based on the Gaussian splash data hierarchy detail model construction method, where the overlapping region extension is set for each grid, it is expressed by the following formula:

[0012]

[0013] in, Indicates the first Layer OK The set of Gaussian ellipsoids within the grid tiles Represents the set of all Gaussian ellipsoids. Indicates the position coordinates of the ellipsoidal plane projection. They represent the first OK The center of the grid tiles, This represents the width and height of the bounding rectangle of all ellipsoidal projection ranges, considering the ellipsoidal scale parameters. This indicates the current grid tile's level.

[0014] Furthermore, based on the Gaussian splash data hierarchy detail model construction method, the overlapping areas of each grid are optimized, including: Based on the unique identifier of each grid, the corresponding opacity filtering threshold and volume parameter filtering threshold are retrieved from the preset codebook; Gaussian ellipsoids in the overlapping region whose opacity does not reach the opacity filtering threshold are filtered out and deleted. Gaussian ellipsoids in the overlapping region whose volume parameter exceeds the volume parameter filtering threshold are also filtered out and deleted.

[0015] Furthermore, according to the Gaussian splash data hierarchy detail model construction method, the opacity filtering threshold and the volume parameter filtering threshold are calculated by the following formula:

[0016]

[0017] in, These represent the levels. The opacity filtering threshold and volume parameter filtering threshold for the lower grid tiles. These represent the baseline threshold for opacity and the baseline threshold for volume parameters at the baseline level, respectively. This represents the exponential scaling factor that varies with the level depth.

[0018] This invention also provides a system for constructing a Gaussian splash data hierarchy detail model, comprising: The acquisition and preprocessing module is used to acquire multi-view images of the target scene, generate Gaussian splash data based on the multi-view images, treat each Gaussian in the Gaussian splash data as a Gaussian ellipsoid, and perform preprocessing. A multi-level detail model construction module is used to calculate the saliency score of each Gaussian ellipsoid, determine a threshold range based on the saliency score, and divide the Gaussian ellipsoid into multiple levels based on the threshold range to construct a multi-level detail model, wherein each level corresponds to a different level of detail. The spatial mesh generation module is used to perform spatial mesh generation on the Gaussian ellipsoid of each level of the multi-level detail model and set overlapping areas at the mesh boundaries. The optimization module is used to optimize the overlapping areas of each grid. The encapsulation module is used to standardize and structure the optimized Gaussian ellipsoid data, and to construct a hierarchical tile data structure to obtain a hierarchically organized Gaussian splash model.

[0019] The present invention also provides a computer-readable storage medium storing a plurality of instructions adapted for loading by a processor to execute any of the above-described methods for constructing a Gaussian splash data hierarchy detail model.

[0020] The present invention also provides an electronic device, including a processor and a memory, wherein the processor is electrically connected to the memory, the memory is used to store instructions and data, and the processor is used in the steps of the Gaussian splash data hierarchy detail model construction method described in any of the preceding claims.

[0021] This invention provides a method, system, storage medium, and electronic device for constructing a Gaussian splash data hierarchical detail model. The invention divides the Gaussian ellipsoid into multiple layers to construct a multi-layered detail model. Through a Gaussian splash layered and block-based data slicing strategy based on the hierarchical detail model, it alleviates the front-end rendering pressure and improves the user interaction experience in large-scene Gaussian data scenarios. Addressing the "black border" rendering problem at tile seams, this invention employs a strategy of spatial meshing of the Gaussian ellipsoid at each layer of the multi-layered detail model, setting overlapping regions at the mesh boundaries, and optimizing these overlapping regions. This strategy effectively alleviates the artifact problem in current rendering sorting methods. Attached Figure Description

[0022] The technical solution and other beneficial effects of the present invention will become apparent from the following detailed description of specific embodiments of the invention, in conjunction with the accompanying drawings.

[0023] Figure 1 A flowchart illustrating the method for constructing a Gaussian splash data hierarchy detail model provided in an embodiment of the present invention.

[0024] Figure 2 A rasterized projection diagram of saliency scores provided in an embodiment of the present invention.

[0025] Figure 3 This is a schematic diagram of the overlapping range grid segmentation provided in an embodiment of the present invention.

[0026] Figure 4 A flowchart illustrating the process of concurrent processing of multiple tiles.

[0027] Figure 5 A schematic diagram of the structure of the Gaussian splash data hierarchy detail model construction system provided in an embodiment of the present invention.

[0028] Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0029] 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.

[0030] This invention provides a method, system, storage medium, and electronic device for constructing a Gaussian splash data hierarchical detail model. The Gaussian splash data hierarchical detail model construction system provided by this invention can be integrated into an electronic device, such as a terminal or server. The terminal can include a tablet computer, laptop computer, personal computer (PC), microprocessor box, or other devices.

[0031] Please see Figure 1 , Figure 1 The flowchart illustrates a method for constructing a Gaussian splash data hierarchical detail model, provided in an embodiment of the present invention. This method, applied in electronic devices, includes the following steps: S1. Acquire multi-view images of the target scene, generate Gaussian splash data based on the multi-view images, treat each Gaussian in the Gaussian splash data as a Gaussian ellipsoid, and perform preprocessing.

[0032] In one embodiment, step S1 includes the following steps: S11, using the center point of the Gaussian ellipsoid as an index, constructs a KD tree model.

[0033] Extract the 3D coordinates of the center points of all Gaussian ellipsoids, and construct a KD tree model using the center points. The KD tree recursively divides the 3D space into a binary tree structure, which reduces the time complexity of nearest neighbor query from O(n) to O(log n).

[0034] S12: Traverse all Gaussian ellipsoids, query the k nearest neighbor ellipsoids of each Gaussian ellipsoid using the KD tree model, and calculate the average distance between the current Gaussian ellipsoid and its k nearest neighbor ellipsoids.

[0035] For each Gaussian ellipsoid, query its k nearest neighbor ellipsoids using the KD tree model, calculate the Euclidean distance to the center of these neighbor ellipsoids, and calculate the average distance to the remaining k neighbor ellipsoids.

[0036] S13, filter and delete Gaussian ellipsoids that exceed the distance threshold.

[0037] Specifically, it is expressed by the following formula:

[0038]

[0039] in, This represents the set of ellipsoids obtained through filtering. Indicates the average distance threshold. Indicates the first An ellipsoid and The average Euclidean distance of the nearest ellipsoid Indicates the total number of ellipsoids. Represents an ellipsoid Position parameters, Represents an ellipsoid j Position parameters, This indicates the number of ellipsoids involved in the calculation of the average Euclidean distance. Indicates distance from the first The nearest ellipsoid A set of ellipsoids.

[0040] S14: Obtain the coordinates of the center points of all Gaussian ellipsoids, calculate the arithmetic mean of the coordinates of all center points, and determine the corresponding mean point as the centroid of the scene.

[0041] S15, using the scene centroid as a reference, performs position scaling and range scaling on each Gaussian ellipsoid.

[0042] Using the scene's centroid as a reference, the relative positions and scale parameters of each Gaussian ellipsoid are simultaneously and linearly scaled to reduce the overall target scene to a size that is easy to observe. The scaling formula is as follows:

[0043]

[0044] in, , , and These represent the model before and after scaling. Position and extent parameters of a Gaussian ellipsoid This indicates the scaling factor.

[0045] S2, calculate the significance score of each Gaussian ellipsoid, determine the threshold range based on the significance score, and divide the Gaussian ellipsoid into multiple levels based on the threshold range to construct a multi-level detail model, where each level corresponds to a different level of detail.

[0046] Figure 2 The rasterized projection diagram of the saliency score provided in the embodiment of the present invention has the following meanings: Training image A and training image B respectively refer to the input view used in the optimization process of the Gaussian splash model; This represents the ray direction vector defined under the current view frustum. Step S2 includes the following steps: S21, use the corresponding pixel view frustum to determine the contribution of the Gaussian ellipsoid, and then calculate its significance score;

[0047] in, This represents the rasterized ray-ellipsoid intersection function, used to determine whether the ellipsoid intersects with the current line of sight; Indicates the first The initial score of each ellipsoid; Indicates the global number The grating rays corresponding to each pixel; Indicates the preceding Transmittance of one Gauss; This is the current ellipsoid. Opacity. Indicates the adjusted number The volume of a Gaussian ellipsoid; The volume parameters of the ellipsoid are represented. These represent the range parameters in the three dimensions of the ellipsoid; Defined as the 90% largest in volume among all ellipsoid sets; Introduced as a hyperparameter to provide additional flexibility.

[0048] S22, the first step for multi-layered Gaussian splashing scenarios Multi-level detailed model, setting saliency score threshold All Gaussian ellipsoids are filtered using a significance score threshold, and those that meet the threshold criteria are included in the current accuracy level. The filtering formula is as follows:

[0049] in, Indicates the first A Gaussian ellipsoid; Represents the set of all Gaussian ellipsoids; Indicates the first The significance score of each ellipsoid; Indicates the first The set of Gaussian ellipsoids that the layer needs to retain; Indicates the first The pruning rate threshold of the layer.

[0050] S23, repeat step S22 until the number of fine-grained levels generated meets the preset requirements.

[0051] In this embodiment, the initial saliency score of each ellipsoid is set to 0, and the number of levels in the multi-level detail model is set to 6.

[0052] S3 performs spatial meshing on the Gaussian ellipsoid of each level of the multi-level detail model and sets overlapping regions at the mesh boundaries.

[0053] In one implementation, step S3 includes the following steps: S31, obtain all Gaussian ellipsoids in the current level, calculate the projection of all Gaussian ellipsoids onto the XOY plane to obtain the two-dimensional projection region, calculate the axis-aligned bounding box of all two-dimensional projection regions, and determine the rectangular region formed by the axis-aligned bounding box as the root node coverage of the current level mesh.

[0054] Specifically, obtain all Gaussian ellipsoids in the current level of detail (such as LOD0, LOD1, etc.), and project each 3D ellipsoid onto the XOY plane (horizontal plane) to obtain a 2D projection region. Calculate the axis-aligned bounding box (AABB) of all projection regions: determine the minimum and maximum X and Y coordinates of all projection points to form a rectangular region. Define this rectangular region as the root node coverage of the current level of meshing, serving as the basic coordinate system for subsequent meshing.

[0055] S32, determine the grid density based on the current partitioning level, divide the two-dimensional projection area into square grid cells of fixed size, use each square grid cell as the spatial boundary, and filter and index all Gaussian ellipsoids falling within the range.

[0056] Specifically, using the root node's coverage area (rectangular region) as the boundary, the two-dimensional projection region is divided into square grid cells of equal size, and the number of grid rows and columns is calculated. For each square grid cell, its two-dimensional boundary (rectangular region) is used as a spatial filtering condition. All Gaussian ellipsoids at the current level are traversed to determine whether their projection centers fall within the grid cell, establishing a mapping relationship between grid cells and the ellipsoid set: each grid cell records a list of ellipsoid indices that fall within it.

[0057] S33 sets the overlap region extension for each grid.

[0058] The extension percentage is set as a parameter. The calculation formula is:

[0059]

[0060] in, Indicates the first Layer OK The set of Gaussian ellipsoids within the grid tiles Represents the set of all Gaussian ellipsoids. Indicates the position coordinates of the ellipsoidal plane projection. They represent the first OK The center of the grid tiles, This represents the width and height of the bounding rectangle of all ellipsoidal projection ranges, considering the ellipsoidal scale parameters. This indicates the current grid tile level; the deeper the level, the higher the grid density.

[0061] Figure 3 This is a schematic diagram of the overlapping range grid segmentation provided in an embodiment of the present invention.

[0062] In this example, the number of meshes in layers 1-6 are as follows: Number, Percentage of extension Set to 0.05.

[0063] S34 assigns a unique identifier to each grid.

[0064] Each grid tile is assigned a unique 64-bit identifier. This encoding structure consists of a 4-bit hierarchy field and a 60-bit index field.

[0065] S4 optimizes the overlapping areas of each grid.

[0066] In one embodiment, step S4 includes the following steps: S41, based on the unique identifier of each grid, retrieves the corresponding opacity filtering threshold and volume parameter filtering threshold from the preset codebook.

[0067] In this embodiment, the unique identifier of each grid tile is constructed using 64-bit binary encoding. This encoding structure is divided into two logical bit fields: a hierarchy index field, occupying the first 4 bits of the higher-order bits, used to identify the detail level to which the current tile belongs; and a spatial location field, occupying the last 60 bits of the lower-order bits, implemented using Hilbert encoding. A pre-built codebook is constructed as a multi-level lookup table, based on hierarchy depth. and spatial index key As the join query key. When indexing grid tiles, the layer depth to which the tile belongs is first resolved through bitwise operations. Spatial Index Then, based on the depth of the hierarchy Calculate the current tile's opacity and volume parameter filtering threshold. The specific threshold calculation formula is as follows:

[0068]

[0069] in, These represent the levels. The opacity filtering threshold and volume parameter filtering threshold for the lower grid tiles. These represent the baseline threshold for opacity and the baseline threshold for volume parameters at the baseline level, respectively. This represents the exponential scaling factor that varies with hierarchy depth. In this embodiment, The values ​​are 0.1 and 2.5 respectively; The values ​​are 0.5 and 1.5 respectively.

[0070] S42: Filter and delete Gaussian ellipsoids in the overlapping region whose opacity does not reach the opacity filtering threshold, and filter and delete Gaussian ellipsoids in the overlapping region whose volume parameter exceeds the volume parameter filtering threshold.

[0071] Initiate a multi-threaded concurrency mechanism to perform the Gaussian ellipsoid filtering operation. Figure 4 A flowchart illustrating multi-watt concurrent processing, such as Figure 4 As shown, the filtering criterion formula for the Gaussian ellipsoid is defined as follows:

[0072] in, This represents the set of Gaussian ellipsoids retained after filtering. Indicates the first The volume of an ellipsoid; These represent the opacity and volume threshold for filtering, respectively.

[0073] S5 standardizes and structures the optimized Gaussian ellipsoid data, and constructs a hierarchical tile data structure to obtain a hierarchically organized Gaussian splash model.

[0074] Specifically, the first step is to encapsulate the position coordinates of the Gaussian ellipsoid into a three-dimensional floating-point vector; the second step is to divide the zero-order spherical harmonic coefficients by the spherical harmonic constant to solve for the color components, and apply the Logistic activation function to the opacity parameter; the third step is to merge the processed color and opacity, and encapsulate them into a four-dimensional unsigned byte vector; the fourth step is to apply the Logistic activation function to the scaling parameter of the Gaussian ellipsoid, mapping its value to the [0,1] interval, and encapsulating it into a three-dimensional floating-point vector; the fifth step is to perform normalization processing on the rotation quaternion of the Gaussian ellipsoid and encapsulate it into a four-dimensional floating-point vector. The sixth step is to reorganize the above Gaussian splash ellipsoid information according to the tile layered and block structure, and calculate the bounding box range, geographic error and other metadata of the corresponding tile in the configuration file to obtain the tree-like Gaussian splash tile structure, that is, the Gaussian splash model after hierarchical tile organization.

[0075] Based on the method described in the above embodiments, this embodiment will further describe it from the perspective of a Gaussian splash data hierarchical detail model construction system. This Gaussian splash data hierarchical detail model construction system can be implemented as an independent entity or integrated into an electronic device. The electronic device can be a terminal, server, or other device. The terminal can include a tablet computer, a laptop computer, a personal computer (PC), a microprocessor box, or other devices.

[0076] Please see Figure 5 , Figure 5 This invention specifically describes a Gaussian splash data hierarchical detail model construction system provided in an embodiment of the invention, which is applied in electronic devices. The Gaussian splash data hierarchical detail model construction system may include: The acquisition and preprocessing module is used to acquire multi-view images of the target scene, generate Gaussian splash data based on the multi-view images, treat each Gaussian in the Gaussian splash data as a Gaussian ellipsoid, and perform preprocessing. A multi-level detail model construction module is used to calculate the saliency score of each Gaussian ellipsoid, determine a threshold range based on the saliency score, and divide the Gaussian ellipsoid into multiple levels based on the threshold range to construct a multi-level detail model, wherein each level corresponds to a different level of detail. The spatial mesh generation module is used to perform spatial mesh generation on the Gaussian ellipsoid of each level of the multi-level detail model and set overlapping areas at the mesh boundaries. The optimization module is used to optimize the overlapping areas of each grid. The encapsulation module is used to standardize and structure the optimized Gaussian ellipsoid data, and to construct a hierarchical tile data structure to obtain a hierarchically organized Gaussian splash model.

[0077] In specific implementation, the above modules and / or units can be implemented as independent entities, or they can be arbitrarily combined and implemented as the same or several entities. For the specific implementation of the above modules and / or units, please refer to the previous method embodiments. For the specific beneficial effects that can be achieved, please also refer to the beneficial effects in the previous method embodiments, which will not be repeated here.

[0078] In addition, this embodiment of the invention also provides an electronic device, which may be a computer, tablet computer, or other similar device. This electronic device can implement the steps in any embodiment of the Gaussian splash data level detail model construction method provided by this embodiment of the invention. Therefore, it can achieve the beneficial effects that any Gaussian splash data level detail model construction method provided by this embodiment of the invention can achieve, as detailed in the preceding embodiments, and will not be repeated here.

[0079] Figure 6 A specific structural block diagram of an electronic device provided in an embodiment of the present invention is shown. This electronic device can be used to implement the Gaussian splash data hierarchy detail model construction method provided in the above embodiments. The electronic device 500 can be a terminal, server, or other device. The terminal can include a tablet computer, laptop computer, personal computer (PC), microprocessor box, or other devices.

[0080] The memory 520 can be used to store software programs and modules, such as the program instructions / modules corresponding to those in the above embodiments. The processor 580 executes various functional applications and data processing by running the software programs and modules stored in the memory 520. The memory 520 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 520 may further include memory remotely located relative to the processor 580, and these remote memories can be connected to the electronic device 500 via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0081] The input unit 530 can be used to receive input numeric or character information, and to generate a keyboard and mouse related to user settings and function control. Display unit 540 can be used to display information input by the user or information provided to the user, as well as various graphical user interfaces, which can be composed of graphics, text, icons, video, and any combination thereof. Display unit 540 may include display panel 541, which may optionally be configured in the form of LCD (Liquid Crystal Display), OLED (Organic Light-Emitting Diode), or other similar forms.

[0082] Electronic device 500, through transmission module 570 (e.g., Wi-Fi module), can help users receive requests, send information, etc., providing users with wireless broadband internet access. Although transmission module 570 is shown in the figure, it is understood that it is not an essential component of electronic device 500 and can be omitted as needed without changing the essence of the invention.

[0083] The processor 580 is the control center of the electronic device 500. It connects to various parts of the phone via various interfaces and lines, and performs various functions and processes data of the electronic device 500 by running or executing software programs and / or modules stored in the memory 520, and by calling data stored in the memory 520, thereby providing overall monitoring of the electronic device. Optionally, the processor 580 may include one or more processing cores; in some embodiments, the processor 580 may integrate an application processor and a modem processor, wherein the application processor mainly handles the operating system, user interface, and applications, and the modem processor mainly handles wireless communication. It is understood that the modem processor may also not be integrated into the processor 580.

[0084] Electronic device 500 also includes a power supply 590 (such as a battery) that supplies power to various components. In some embodiments, the power supply may be logically connected to processor 580 through a power management system, thereby enabling functions such as charging, discharging, and power consumption management through the power management system. The power supply 590 may also include one or more DC or AC power supplies, recharging systems, power fault detection circuits, power converters or inverters, power status indicators, and other arbitrary components.

[0085] Although not shown, the electronic device 500 also includes cameras (such as front-facing cameras and rear-facing cameras), Bluetooth modules, etc., which will not be described in detail here. Specifically, in this embodiment, the display unit of the electronic device is a touch screen display, and the mobile terminal also includes a memory and one or more programs, wherein one or more programs are stored in the memory and configured to be executed by one or more processors. One or more programs contain instructions for performing the following operations: Acquire multi-view images of the target scene, generate Gaussian splash data based on the multi-view images, treat each Gaussian in the Gaussian splash data as a Gaussian ellipsoid, and perform preprocessing. Calculate the saliency score for each Gaussian ellipsoid, determine a threshold range based on the saliency score, and divide the Gaussian ellipsoid into multiple levels based on the threshold range to construct a multi-level detail model, where each level corresponds to a different level of detail. Spatial meshing is performed on the Gaussian ellipsoid of each level of the multi-level detail model, and overlapping regions are set at the mesh boundaries; Optimize the overlapping areas of each grid; The optimized Gaussian ellipsoid data is standardized, encapsulated, and structured, and a hierarchical tile data structure is constructed to obtain a hierarchically organized Gaussian splash model.

[0086] In practice, the above modules can be implemented as independent entities or combined in any way to be implemented as the same or several entities. For the specific implementation of the above modules, please refer to the previous method implementation examples, which will not be repeated here.

[0087] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by instructions, or by instructions controlling related hardware. These instructions can be stored in a computer-readable storage medium and loaded and executed by a processor. Therefore, embodiments of the present invention provide a storage medium storing multiple instructions that can be loaded by a processor to execute the steps of any embodiment of the Gaussian splash data hierarchy detail model construction method provided by the present invention.

[0088] The computer-readable storage medium may include: read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0089] Since the instructions stored in the storage medium can execute the steps in any embodiment of the Gaussian splash data hierarchy detail model construction method provided in the embodiments of the present invention, the beneficial effects that any Gaussian splash data hierarchy detail model construction method provided in the embodiments of the present invention can achieve can be realized, as detailed in the preceding embodiments, and will not be repeated here.

[0090] The foregoing has provided a detailed description of a Gaussian splash data hierarchy detail model construction method, system, storage medium, and electronic device provided by embodiments of the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. 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 the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for constructing a Gaussian splash data hierarchy detail model, characterized in that, The method includes: Acquire multi-view images of the target scene, generate Gaussian splash data based on the multi-view images, treat each Gaussian in the Gaussian splash data as a Gaussian ellipsoid, and perform preprocessing. Calculate the saliency score for each Gaussian ellipsoid, determine a threshold range based on the saliency score, and divide the Gaussian ellipsoid into multiple levels based on the threshold range to construct a multi-level detail model, where each level corresponds to a different level of detail. Spatial meshing is performed on the Gaussian ellipsoid of each level of the multi-level detail model, and overlapping regions are set at the mesh boundaries; Optimize the overlapping areas of each grid; The optimized Gaussian ellipsoid data is standardized, encapsulated, and structured, and a hierarchical tile data structure is constructed to obtain a hierarchically organized Gaussian splash model.

2. The method for constructing a Gaussian splash data hierarchy detail model according to claim 1, characterized in that, Each Gaussian in the Gaussian splash data is treated as a Gaussian ellipsoid and preprocessed, including: A KD tree model is constructed using the center point of the Gaussian ellipsoid as an index; Traverse all Gaussian ellipsoids, query the k nearest neighbor ellipsoids of each Gaussian ellipsoid using the KD tree model, and calculate the average distance between the current Gaussian ellipsoid and its k nearest neighbor ellipsoids; Filter and delete Gaussian ellipsoids that exceed the distance threshold; Obtain the coordinates of the center points of all Gaussian ellipsoids, calculate the arithmetic mean of all center point coordinates, and determine the corresponding mean point as the centroid of the scene; Using the scene centroid as a reference, the position and range of each Gaussian ellipsoid are scaled.

3. The method for constructing a Gaussian splash data hierarchy detail model according to claim 2, characterized in that, The Gaussian ellipsoids exceeding the distance threshold are filtered and deleted, as expressed by the following formula: in, This represents the set of ellipsoids obtained through filtering. Indicates the average distance threshold. Indicates the first An ellipsoid and The average Euclidean distance of the nearest ellipsoid Indicates the total number of ellipsoids. Represents an ellipsoid Position parameters, Represents an ellipsoid j Position parameters, This indicates the number of ellipsoids involved in the calculation of the average Euclidean distance. Indicates distance from the first The nearest ellipsoid A set of ellipsoids.

4. The method for constructing a Gaussian splash data hierarchy detail model according to claim 1, characterized in that, The Gaussian ellipsoid of each level of the multi-level detail model is spatially meshed, and overlapping regions are set at the mesh boundaries, including: Obtain all Gaussian ellipsoids in the current level, calculate the projection of all Gaussian ellipsoids onto the XOY plane to obtain the two-dimensional projection region, calculate the axis-aligned bounding box of all two-dimensional projection regions, and determine the rectangular region formed by the axis-aligned bounding box as the root node coverage of the current level mesh. The grid density is determined based on the current partitioning level. The two-dimensional projection area is divided into square grid cells of fixed size. Each square grid cell serves as the spatial boundary. All Gaussian ellipsoids falling within the range are filtered and indexed. Set the overlap region extension for each grid; Assign a unique identifier to each grid.

5. The method for constructing a Gaussian splash data hierarchy detail model according to claim 4, characterized in that, The overlapping region extension is set for each grid, expressed by the following formula: in, Indicates the first Layer OK The set of Gaussian ellipsoids within the grid tiles Represents the set of all Gaussian ellipsoids. Represents the position coordinates of the ellipsoidal plane projection. They represent the first OK The center of the grid tiles, This represents the width and height of the bounding rectangle of all ellipsoidal projection ranges, considering the ellipsoidal scale parameters. This indicates the current grid tile's level.

6. The method for constructing a Gaussian splash data hierarchy detail model according to claim 4, characterized in that, Optimize the overlapping areas of each grid, including: Based on the unique identifier of each grid, the corresponding opacity filtering threshold and volume parameter filtering threshold are retrieved from the preset codebook; Gaussian ellipsoids in the overlapping region whose opacity does not reach the opacity filtering threshold are filtered out and deleted. Gaussian ellipsoids in the overlapping region whose volume parameter exceeds the volume parameter filtering threshold are also filtered out and deleted.

7. The method for constructing a Gaussian splash data hierarchy detail model according to claim 6, characterized in that, The opacity filtering threshold and the volume parameter filtering threshold are calculated using the following formula: in, These represent the levels. The opacity filtering threshold and volume parameter filtering threshold for the lower grid tiles. These represent the baseline threshold for opacity and the baseline threshold for volume parameters at the baseline level, respectively. This represents the exponential scaling factor that varies with the level depth.

8. A system for constructing a hierarchical detail model of Gaussian splash data, characterized in that, include: The acquisition and preprocessing module is used to acquire multi-view images of the target scene, generate Gaussian splash data based on the multi-view images, treat each Gaussian in the Gaussian splash data as a Gaussian ellipsoid, and perform preprocessing. A multi-level detail model construction module is used to calculate the saliency score of each Gaussian ellipsoid, determine a threshold range based on the saliency score, and divide the Gaussian ellipsoid into multiple levels based on the threshold range to construct a multi-level detail model, wherein each level corresponds to a different level of detail. The spatial mesh generation module is used to perform spatial mesh generation on the Gaussian ellipsoid of each level of the multi-level detail model and set overlapping areas at the mesh boundaries. The optimization module is used to optimize the overlapping areas of each grid. The encapsulation module is used to standardize and structure the optimized Gaussian ellipsoid data, and to construct a hierarchical tile data structure to obtain a hierarchically organized Gaussian splash model.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a plurality of instructions adapted for loading by a processor to execute the Gaussian splash data hierarchy detail model construction method according to any one of claims 1 to 7.

10. An electronic device, characterized in that, The method includes a processor and a memory, the processor being electrically connected to the memory, the memory being used to store instructions and data, and the processor being used to execute the steps in the Gaussian splash data hierarchy detail model construction method according to any one of claims 1 to 7.

Images