Human body model reconstruction and body measurement method and system based on three-dimensional Gaussian splashing

Abstract

According to the human body model reconstruction and body measurement method and system based on three-dimensional Gaussian splashing, the 3D GS technology is adopted, a scene is modeled through a group of Gaussian ellipsoids, and therefore efficient rendering and reconstruction are achieved. Compared with a traditional neural implicit representation method, the rendering speed and the reconstruction quality of the 3DGS are remarkably improved. The method is not limited by specific postures of the SMPL human body parameterization model, and more complex and diversified postures can be processed. Therefore, a more vivid and natural human body model can be generated in applications such as virtual reality, animation and games. According to the method, multi-view depth supervision is introduced, depth information is obtained from multiple views, and the precision of the model is further improved by using multi-modal data. By obtaining the RGB images and the depth maps of the multiple visual angles at the same time, the quality of the reconstruction result can be effectively improved, and the method is used for measuring body parameters.

Description

Technical Field

[0001] The present invention relates to the fields of computer graphics, 3D computer vision and reconstruction technology, and particularly to a human body model reconstruction and body measurement method and system based on 3D Gaussian splashing. Background Art

[0002] In recent years, with the progress of depth camera technology, 3D reconstruction methods based on multi-view data have gradually become a research hotspot. Traditional 3D reconstruction methods are mainly divided into explicit methods and implicit methods. Explicit methods (such as methods based on the parametric human body model SMPL) generate human meshes by directly optimizing model parameters and clothing offsets to fit the input images. Implicit methods represent the human body through continuous functions (such as occupancy fields, SDFs or NeRFs). These methods perform well in modeling flexible topologies, but they have limitations in terms of scalability and efficiency due to the high computational cost of training and inference.

[0003] 1. Efficiency issues:

[0004] Explicit methods: Traditional explicit methods, such as methods based on the SMPL model, usually require a long optimization time, are difficult to handle complex clothing styles, and are less efficient in multi-view inputs. For example, some methods require time-consuming optimization for each instance, which limits their applicability and efficiency in actual scenarios.

[0005] Implicit methods: Although implicit methods perform well in modeling flexible topologies, they have limitations in terms of scalability and efficiency due to the high computational cost of training and inference. For example, some studies based on diffusion models elevate the 2D diffusion model prior to 3D through score distillation sampling (SDS), but each instance usually requires an optimization time of up to several hours.

[0006] 2. Pose limitations:

[0007] The SMPL model is a widely used parametric human body model, but its reconstruction results are usually limited to specific poses. For example, the SMPL model may be inaccurate when dealing with complex poses and dynamic changes. This limits the applicability of the SMPL model in some application scenarios that require high precision and diverse poses.

[0008] 3. Accuracy issues:

[0009] Single-view input: Reconstruction methods with single-view input face challenges in dealing with invisible regions. For example, some methods may produce blurry or inaccurate results when generating invisible regions, affecting the overall reconstruction quality.

[0010] Multi-view input: Although traditional multi-view reconstruction methods can improve the reconstruction quality, they usually require a large number of multi-view image inputs, which may be difficult to obtain in practical applications. Moreover, three-dimensional reconstruction using only single-modal image data limits the reconstruction accuracy. Summary of the Invention

[0011] To achieve the above and other advantages of the present invention, the first object of the present invention is to provide a method for reconstructing a human body model and body measurement based on three-dimensional Gaussian splashing, including the following steps:

[0012] Obtain a multi-view human body video dataset; wherein, the multi-view human body video dataset includes an RGB video stream and a corresponding depth video stream;

[0013] Extract image frames from the RGB video and the depth video respectively, and use a segmentation algorithm to extract the two-dimensional human body;

[0014] Generate a high-quality human body model based on a three-dimensional Gaussian ellipsoid through the segmented depth map;

[0015] Mesh the high-quality human body model;

[0016] Perform three-dimensional space key point sampling on the high-quality human body model;

[0017] Based on the high-quality human body model and the key point sampling results, calculate various body parameters to achieve measurement.

[0018] Further, the multi-view human body video dataset is obtained by multi-view scanning of the human body by multiple depth cameras arranged in a vertical column and through a rotating table.

[0019] Further, the step of using a segmentation algorithm to extract the two-dimensional human body includes:

[0020] Use a deep learning algorithm to generate a corresponding mask to accurately segment the human body part and reduce background interference.

[0021] Further, the step of generating a high-quality human body model based on a three-dimensional Gaussian ellipsoid through the segmented depth map includes:

[0022] Perform multi-view fusion based on the segmented depth map to generate a human body three-dimensional point cloud;

[0023] Initialize each point in the human body three-dimensional point cloud and assign it an explicit three-dimensional Gaussian ellipsoid expression.

[0024] Further, after the step of generating the human body three-dimensional point cloud, it further includes:

[0025] Remove point cloud outliers through a filtering algorithm.

[0026] Further, after the step of initializing each point in the three-dimensional human point cloud and endowing it with an explicit three-dimensional Gaussian ellipsoid representation, the following steps are further included:

[0027] Project the three-dimensional Gaussian onto a two-dimensional plane according to the known camera pose, and use rasterization technology to render the human body image from the corresponding perspective.

[0028] Further, after the step of initializing each point in the three-dimensional human point cloud and endowing it with an explicit three-dimensional Gaussian ellipsoid representation, the following steps are further included:

[0029] Introduce a depth supervision optimization loss function to ensure that the generated image matches the actual depth information.

[0030] The second object of the present invention is to provide a computer device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the steps of the above method are implemented.

[0031] The third object of the present invention is to provide a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above method are implemented.

[0032] The fourth object of the present invention is to provide a human body model reconstruction and body measurement system based on three-dimensional Gaussian splashing, which implements the above method, including a data acquisition module, a data processing module, a human body reconstruction module, a human body three-dimensional model meshing module, a human body key point extraction module, and a body parameter measurement module; wherein,

[0033] The data acquisition module is used to collect a multi-perspective human body video dataset; wherein, the multi-perspective human body video dataset includes an RGB video stream and a corresponding depth video stream;

[0034] The data processing module is used to extract image frames from the RGB video and the depth video respectively, and use a segmentation algorithm to extract the two-dimensional human body;

[0035] The human body reconstruction module is used to generate a high-quality human body model based on three-dimensional Gaussian ellipsoids through the segmented depth map;

[0036] The human body three-dimensional model meshing module is used to mesh the high-quality human body model;

[0037] The human body key point extraction module is used to sample three-dimensional space key points of the high-quality human body model;

[0038] The body parameter measurement module is used to calculate various body parameters based on the high-quality human body model and the key point sampling results to achieve measurement.

[0039] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0040] The present invention adopts the 3D Gaussian Splattering (3DGS) technology to model the scene through a group of Gaussian ellipsoids, thereby achieving efficient rendering and reconstruction. Compared with traditional neural implicit representation methods (such as NeRF), 3DGS has significant improvements in both rendering speed and reconstruction quality.

[0041] The method provided by the present invention is not limited to the specific postures of the SMPL human parameterization model and can handle more complex and diverse postures. This enables the generation of more realistic and natural human models in applications such as virtual reality, animation, and games.

[0042] The present invention introduces multi-view depth supervision. By obtaining depth information from multiple perspectives and utilizing multi-modal data, the accuracy of the model is further improved. By simultaneously obtaining RGB images and depth maps from multiple perspectives, the quality of the reconstruction results can be effectively enhanced for body parameter measurement.

[0043] Based on the three-dimensional Gaussian splashing technology, the present invention realizes low-cost, high-fidelity, and high-precision three-dimensional human body reconstruction through the efficient processing of multi-view RGB pictures and depth information.

[0044] The above description is only an overview of the technical solution of the present invention. In order to be able to more clearly understand the technical means of the present invention and implement it in accordance with the content of the specification, the following takes the preferred embodiments of the present invention and combines the accompanying drawings to elaborate in detail as follows. The specific implementation manners of the present invention are given in detail by the following embodiments and their accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The accompanying drawings described herein are used to provide a further understanding of the present invention, form a part of this application, and the illustrative embodiments and descriptions of the present invention are used to explain the present invention and do not constitute an improper limitation to the present invention. In the drawings:

[0046] Figure 1 is the flowchart of the method for human body model reconstruction and body measurement based on three-dimensional Gaussian splashing Figure 1 ;

[0047] Figure 2 is the flowchart of the method for human body model reconstruction and body measurement based on three-dimensional Gaussian splashing Figure 2 ;

[0048] Figure 3 is the flowchart for generating a high-quality human body model based on three-dimensional Gaussian ellipsoids through the segmented depth map;

[0049] Figure 4Schematic diagram of a human body model reconstruction and body measurement system based on three-dimensional Gaussian splashing;

[0050] Figure 5 Schematic diagram of a computer device;

[0051] Figure 6 Schematic diagram of a computer-readable storage medium. Specific implementation manners

[0052] Next, in combination with the accompanying drawings and specific implementation manners, the present invention will be further described. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all the embodiments. It should be noted that, on the premise of no conflict, the following-described embodiments or technical features can be combined arbitrarily to form new embodiments.

[0053] All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts belong to the scope of protection of the present invention.

[0054] In this application, the attached drawing numbers are only used to distinguish each step in the solution and are not used to limit the execution order of each step. The specific execution order shall be subject to the description in the specification.

[0055] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which the present invention belongs. The terms used in the specification of the present invention herein are only for the purpose of describing specific embodiments and are not intended to limit the present invention.

[0056] With the development of computer technology, three-dimensional reconstruction technology has been widely applied in the fields of virtual reality, film and television production, game development, digital humans, etc. However, the prior art still has limitations in the accuracy, efficiency, and parameter extraction of human body reconstruction. The present invention provides a method and system for human body model reconstruction and body parameter measurement based on three-dimensional Gaussian splashing, aiming to solve these technical problems and improve the accuracy and efficiency of human body three-dimensional reconstruction.

[0057] Embodiment 1

[0058] A method for human body model reconstruction and body measurement based on three-dimensional Gaussian splashing, as Figure 1 、 Figure 2 shown, includes the following steps:

[0059] S1. Obtain a multi-view human body video dataset; wherein, the multi-view human body video dataset includes an RGB video stream and a corresponding depth video stream;

[0060] In some embodiments, the multi-view human video dataset is obtained by multi-view scanning of a human body using multiple (two or more) depth cameras arranged in a vertical column and a turntable. Optionally, the subject is located on the turntable and remains in the A-shaped pose without movement. Each camera simultaneously collects the RGB video stream and the corresponding depth video stream and uploads them to the terminal.

[0061] S2. Extract image frames from the RGB video and the depth video respectively, and use a segmentation algorithm to extract the two-dimensional human body;

[0062] By converting the image frames in the multi-view video stream into static pictures, it is used for subsequent image segmentation and reconstruction.

[0063] Furthermore, the step of using the segmentation algorithm to extract the two-dimensional human body includes:

[0064] Apply a deep learning algorithm (such as SAM) to generate the corresponding mask, accurately segment the human body part, reduce background interference, and improve the accuracy of subsequent point cloud generation.

[0065] S3. Generate a high-quality human body model based on a three-dimensional Gaussian ellipsoid from the segmented depth map;

[0066] Adopting multi-view data combined with a three-dimensional Gaussian model can accurately capture the subtle features of the human body and ensure a high-fidelity reconstruction result.

[0067] In some embodiments, as Figure 3 shown, the step of generating a high-quality human body model based on a three-dimensional Gaussian ellipsoid from the segmented depth map includes:

[0068] S31. Perform multi-view fusion based on the segmented depth map to generate a three-dimensional human body point cloud;

[0069] Furthermore, after the step of generating the three-dimensional human body point cloud, it further includes:

[0070] Remove the outliers in the point cloud through a filtering algorithm.

[0071] S32. Initialize each point in the three-dimensional human body point cloud and endow it with an explicit three-dimensional Gaussian ellipsoid expression.

[0072] Furthermore, after the step of initializing each point in the three-dimensional human body point cloud and endowing it with an explicit three-dimensional Gaussian ellipsoid expression, it further includes:

[0073] According to the known camera pose, project the three-dimensional Gaussian onto the two-dimensional plane, and use rasterization technology to render the human body image in the corresponding view.

[0074] Apply rasterization technology to achieve two-dimensional projection and rendering, and improve the quality and realism of the generated image.

[0075] Further, after the step of initializing each point in the three-dimensional human body point cloud and endowing it with an explicit three-dimensional Gaussian ellipsoid representation, the following steps are also included:

[0076] Introduce a depth supervision optimization loss function to ensure that the generated image matches the actual depth information, thereby improving the reconstruction accuracy.

[0077] By introducing multi-view depth information and optimizing the loss function, the 3DGS reconstruction process has real-depth supervision, thereby achieving higher reconstruction accuracy.

[0078] S4. Mesh the high-quality human body model; meshing the three-dimensional human body model facilitates subsequent downstream task processing, such as editing, key point detection, measurement calculation, etc.

[0079] S5. Sample three-dimensional space key points of the high-quality human body model for subsequent body parameter calculation and measurement.

[0080] Implement the extraction and calculation of key parts of the human body model through a three-dimensional key point extraction algorithm, and then achieve accurate measurement of body parameters.

[0081] S6. Calculate various body parameters based on the high-quality human body model and the key point sampling results to achieve measurement, including height, trunk length, waist circumference, chest circumference, shoulder width, etc.

[0082] The present invention combines multi-view data with a three-dimensional Gaussian model, can accurately capture the subtle features of the human body, and ensure a high-fidelity reconstruction result; uses rasterization technology to achieve two-dimensional projection and rendering, and improves the quality and realism of the generated image.

[0083] The present invention optimizes the loss function by introducing multi-view depth information and uses multi-modal data to improve the reconstruction accuracy.

[0084] The present invention is not limited to the specific postures of the SMPL human body parameterization model and can handle more complex and diverse postures; at the same time, the built low-cost scanning system is not only limited to human body reconstruction, but can also be extended to the three-dimensional reconstruction of other objects, and has strong versatility.

[0085] Embodiment 2

[0086] A human body model reconstruction and body measurement system based on three-dimensional Gaussian splash realizes the above method. For the detailed description of the method, reference can be made to the corresponding description in the above method embodiment, which will not be repeated here. As Figure 4As shown, the system 700 includes a data acquisition module 710, a data processing module 720, a human body reconstruction module 730, a human body three-dimensional model meshing module 740, a human body key point extraction module 750, and a body parameter measurement module 760; wherein,

[0087] The data acquisition module is used to acquire a multi-view human body video data set; wherein the multi-view human body video data set includes an RGB image video stream and a corresponding depth image video stream;

[0088] The data processing module is used to extract image frames from the RGB image video and the depth image video respectively, and extract the two-dimensional human body using a segmentation algorithm;

[0089] The human body reconstruction module is used to generate a high-quality human body model based on a three-dimensional Gaussian ellipsoid through the segmented depth map;

[0090] The human body three-dimensional model meshing module is used to mesh the high-quality human body model;

[0091] The human body key point extraction module is used to perform three-dimensional space key point sampling on the high-quality human body model;

[0092] The body parameter measurement module is used to calculate various body parameters based on the high-quality human body model and key point sampling results to achieve measurement.

[0093] Based on the technical solution of the above embodiment, optionally, the multi-view human body video dataset is obtained by arranging multiple depth cameras in vertical columns and performing multi-view scanning on the human body through a rotating stage.

[0094] Based on the technical solution of the above embodiment, optionally, the step of extracting the two-dimensional human body by using a segmentation algorithm includes:

[0095] Use deep learning algorithms to generate corresponding masks to accurately segment human body parts and reduce background interference.

[0096] Based on the technical solution of the above embodiment, optionally, the step of generating a high-quality human body model based on a three-dimensional Gaussian ellipsoid through the segmented depth map includes:

[0097] Based on the segmented depth map, multi-view fusion is performed to generate a 3D point cloud of the human body;

[0098] Each point in the three-dimensional point cloud of the human body is initialized and given an explicit three-dimensional Gaussian ellipsoid expression.

[0099] Based on the technical solution of the above embodiment, optionally, after the step of generating a three-dimensional point cloud of a human body, the method further includes:

[0100] Remove outliers from the point cloud through a filtering algorithm.

[0101] Based on the technical solution of the above embodiment, optionally, after the step of initializing each point in the three-dimensional point cloud of the human body and endowing it with an explicit three-dimensional Gaussian ellipsoid representation, the following steps are further included:

[0102] Project the three-dimensional Gaussian onto a two-dimensional plane according to the known camera pose, and use rasterization technology to render the human body image from the corresponding perspective.

[0103] Based on the technical solution of the above embodiment, optionally, after the step of initializing each point in the three-dimensional point cloud of the human body and endowing it with an explicit three-dimensional Gaussian ellipsoid representation, the following steps are further included:

[0104] Introduce a depth supervision optimization loss function to ensure that the generated image matches the actual depth information.

[0105] The present invention combines multi-view data with a three-dimensional Gaussian model, which can accurately capture the subtle features of the human body and ensure a high-fidelity reconstruction result; it uses rasterization technology to achieve two-dimensional projection and rendering, improving the quality and realism of the generated image.

[0106] The present invention optimizes the loss function by introducing multi-view depth information and uses multi-modal data to improve the reconstruction accuracy.

[0107] The present invention adopts a system modular design, including a data acquisition module, a data processing module, a human body model reconstruction module, and a body parameter measurement module, which is highly efficient in processing.

[0108] The present invention is not limited to the specific poses of the SMPL human parameterization model and can handle more complex and diverse poses; at the same time, the low-cost scanning system built can be extended to the three-dimensional reconstruction of other objects and is not limited to human body reconstruction, with strong versatility.

[0109] Embodiment 3

[0110] A computer device 800, as Figure 5 shown, includes a memory 810, a processor 820, and a computer program 830 stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of a method for human body model reconstruction and body measurement based on three-dimensional Gaussian splashing. For a detailed description of the method, reference can be made to the corresponding description in the above method embodiments and will not be repeated here.

[0111] Embodiment 4

[0112] A computer-readable storage medium, as Figure 6As shown, a computer program is stored thereon, and when the computer program is executed by a processor, it implements the steps of a method for reconstructing a human body model and body measurement based on three-dimensional Gaussian splashing. For a detailed description of the method, reference may be made to the corresponding description in the above method embodiments, which will not be elaborated here.

[0113] The number of devices and the scale of processing described here are used to simplify the description of the present invention. Applications, modifications, and variations of the present invention will be apparent to those skilled in the art.

[0114] Although the embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. It can be fully applied to various fields suitable for the present invention. For those skilled in the art, additional modifications can be easily made. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to specific details and the examples shown and described here.

[0115] The devices, computer devices, non-volatile computer storage media provided in the embodiments of this specification correspond to the method. Therefore, the devices, computer devices, and non-volatile computer storage media also have beneficial technical effects similar to the corresponding method. Since the beneficial technical effects of the method have been described in detail above, the beneficial technical effects of the corresponding devices, computer devices, and non-volatile computer storage media will not be elaborated here.

[0116] Those skilled in the art also know that in addition to implementing the controller in the form of pure computer-readable program code, the method steps can be logically programmed to enable the controller to implement the same function in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, embedded microcontrollers, etc. Therefore, such a controller can be considered a hardware component, and the devices included therein for implementing various functions can also be regarded as the structures within the hardware component. Or even, the devices for implementing various functions can be regarded as both software units for implementing the method and the structures within the hardware component.

[0117] The systems, devices, or units illustrated in the above embodiments can be specifically implemented by computer chips or entities, or by products with certain functions. For the convenience of description, when describing the above devices, they are described as various units according to their functions. Of course, when implementing one or more embodiments of this specification, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0118] Those skilled in the art should understand that the embodiments of this specification can be provided as a method, a system, or a computer program product. Therefore, the embodiments of this specification can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the embodiments of this specification can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0119] This specification is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of this specification. It should be understood that each flow and / or block in the flowchart and / or block diagram can be implemented by computer program instructions, and the combination of flows and / or blocks in the flowchart and / or block diagram can also be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for implementing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0120] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including an instruction device, and the instruction device implements the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0121] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0122] It should also be noted that the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, such that a process, method, commodity or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, commodity or device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, method, commodity or device comprising said element.

[0123] This specification may be described in the general context of computer-executable instructions executed by a computer, such as program units. Generally, program units include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The specification may also be practiced in a distributed computing environment where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program units may be located in local and remote computer storage media including storage devices.

[0124] Each embodiment in this specification is described in a progressive manner. For the same or similar parts among the embodiments, reference may be made to each other. Each embodiment focuses on the differences from other embodiments. In particular, for system embodiments, since they are basically similar to method embodiments, the description is relatively simple, and reference may be made to the corresponding parts of the method embodiments for relevant details.

[0125] The above is only for the embodiments of this specification and is not intended to limit one or more embodiments of this specification. For those skilled in the art, various changes and modifications can be made to one or more embodiments of this specification. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of one or more embodiments of this specification shall be included within the scope of the claims of one or more embodiments of this specification.

Claims

1. A human body model reconstruction and body measurement method based on three-dimensional Gaussian splashing, characterized in that: The following steps are involved: Acquire a multi-view human body video data set; wherein the multi-view human body video data set includes an RGB image video stream and a corresponding depth image video stream; Extract image frames from RGB image video and depth image video respectively, and use segmentation algorithm to extract two-dimensional human body; Generate a high-quality human body model based on a three-dimensional Gaussian ellipsoid through the segmented depth map; Meshing the high-quality human body model; Performing three-dimensional spatial key point sampling on the high-quality human body model; Based on the high-quality human body model and key point sampling results, various body parameters are calculated to achieve measurement.

2. A method for human body model reconstruction and body measurement based on three-dimensional Gaussian splashing as claimed in claim 1, characterized in that: The multi-view human body video data set is obtained by performing multi-view scanning of the human body by using a rotating table and a plurality of depth cameras arranged in a vertical column.

3. The method for human body model reconstruction and body measurement based on three-dimensional Gaussian splashing as claimed in claim 1, characterized in that: The step of extracting a two-dimensional human body using a segmentation algorithm comprises: Use deep learning algorithms to generate corresponding masks to accurately segment human body parts and reduce background interference.

4. The method for human body model reconstruction and body measurement based on three-dimensional Gaussian splashing as claimed in claim 2, characterized in that: The step of generating a high-quality human body model based on a three-dimensional Gaussian ellipsoid through the segmented depth map comprises: Based on the segmented depth map, multi-view fusion is performed to generate a 3D point cloud of the human body; Each point in the three-dimensional point cloud of the human body is initialized and given an explicit three-dimensional Gaussian ellipsoid expression.

5. The method for human body model reconstruction and body measurement based on three-dimensional Gaussian splashing as claimed in claim 4, characterized in that: After the step of generating a three-dimensional point cloud of a human body, the method further includes: The point cloud outliers are removed through filtering algorithm.

6. The method for human body model reconstruction and body measurement based on three-dimensional Gaussian splashing as claimed in claim 4, characterized in that: After the step of initializing each point in the three-dimensional point cloud of the human body and giving it an explicit three-dimensional Gaussian ellipsoid expression, the method further includes: According to the known camera pose, the three-dimensional Gaussian is projected onto a two-dimensional plane, and the human body image under the corresponding perspective is rendered using rasterization technology.

7. The method for human body model reconstruction and body measurement based on three-dimensional Gaussian splashing as claimed in claim 4, characterized in that: After the step of initializing each point in the three-dimensional point cloud of the human body and giving it an explicit three-dimensional Gaussian ellipsoid expression, the method further includes: Deep supervision is introduced to optimize the loss function to ensure that the generated image matches the actual depth information.

8. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A human body model reconstruction and body measurement system based on three-dimensional Gaussian splashing, implementing the method according to any one of claims 1 to 7, characterized in that: It includes data acquisition module, data processing module, human body reconstruction module, human body 3D model meshing module, human body key point extraction module, and body parameter measurement module; among them, The data acquisition module is used to acquire a multi-view human body video data set; wherein the multi-view human body video data set includes an RGB image video stream and a corresponding depth image video stream; The data processing module is used to extract image frames from the RGB image video and the depth image video respectively, and extract the two-dimensional human body using a segmentation algorithm; The human body reconstruction module is used to generate a high-quality human body model based on a three-dimensional Gaussian ellipsoid through the segmented depth map; The human body three-dimensional model meshing module is used to mesh the high-quality human body model; The human body key point extraction module is used to perform three-dimensional space key point sampling on the high-quality human body model; The body parameter measurement module is used to calculate various body parameters based on the high-quality human body model and key point sampling results to achieve measurement.

Images