Indoor scene wireless channel modeling method based on laser radar sensing reconstruction

Abstract

The invention is suitable for the technical field of cross fusion of three-dimensional scene reconstruction and wireless channel modeling, and provides an indoor scene wireless channel modeling method based on laser radar perception reconstruction, which comprises the following steps: firstly, collecting indoor scene point cloud data by using a laser radar, then carrying out noise reduction sampling processing on the point cloud data, and obtaining a noise reduction sampling result; then, structural elements including walls, ceilings and floors are extracted based on random sampling consistency and a Hough transform algorithm, a wall opening structure is detected by adopting an image semantic segmentation model aiming at internal elements, furniture reconstruction is achieved based on template deformation, three-dimensional surface reconstruction is achieved by utilizing a surface element subdivision algorithm, and the wall opening structure is obtained. And finally, matching electromagnetic parameters on the surfaces of different materials, and completing wireless channel modeling through ray tracing simulation. Therefore, a complex scene is divided into structural elements and internal elements, and corresponding algorithms are designed for different elements to extract key geometric features, so that the precision and robustness of reconstruction are improved.

Description

Technical Field

[0001] The present application belongs to the technical field of cross-integration of three-dimensional scene reconstruction and wireless channel modeling, and in particular, relates to a method for indoor scene wireless channel modeling based on laser radar perception reconstruction. Background Art

[0002] With the in-depth research on 6G communication technology, how to achieve high-precision and high-efficiency scene reconstruction and wireless channel modeling has become a key technical challenge in the field of communications. Due to its closedness and diversity, the indoor environment has complex signal propagation characteristics, so it is necessary to accurately model the key reflectors and scatterers in the scene. LiDAR technology achieves accurate measurement of the distance to the target object by emitting laser pulses and accurately measuring their return time. On this basis, the point cloud data formed provides us with detailed three-dimensional information of the indoor scene. This information covers the specific location and shape of key reflectors and scatterers, and constructs an accurate three-dimensional environmental model. The application of this technology is crucial for in-depth understanding and effective modeling of the propagation characteristics of indoor wireless signals. It can more accurately assist the channel simulator in calculating the reflection, refraction and scattering electromagnetic propagation effects of electromagnetic signals in the room, thereby further exploring the complex mapping relationship between channel characteristics and the environment.

[0003] However, in the reconstruction of communication scenes based on LiDAR perception assistance, the complexity of the scene makes the acquisition and processing of point cloud data face huge challenges. On the one hand, the amount of point cloud data in large-scale scenes is huge, which limits the speed of data transmission and processing; on the other hand, accurately extracting key geometric structures from huge point cloud data is a challenging task. This not only requires the algorithm to have efficient data screening and noise reduction capabilities, but also requires the ability to accurately identify and distinguish key features in the scene, such as edges, corners, and planes, to facilitate subsequent communication scene analysis and modeling. In addition, the possible occlusion, reflection, and noise in the point cloud data further increase the difficulty of extracting key geometric structures, and put forward higher requirements on the robustness and accuracy of the algorithm.

[0004] As an effective method for predicting electromagnetic wave propagation, ray tracing simulation technology has important application value in 6G communication systems. However, when dealing with complex scenes, traditional ray tracing methods often require a lot of manual measurements and meticulous manual modeling to build accurate three-dimensional scene models, which is not only time-consuming but also extremely labor-intensive. These tedious steps include accurate recording of the size, material, and position parameters of each object in the scene, as well as proofreading of multiple details in the model one by one, which places extremely high demands on researchers and, to a certain extent, limits the efficiency of ray tracing technology in rapid response and large-scale scene applications. Therefore, lidar point cloud-assisted scene reconstruction provides an efficient, accurate, and cost-effective solution for indoor scene ray tracing simulation, which has important practical application value for promoting the development of 6G communication systems. Summary of the invention

[0005] The embodiment of the present application provides a method for modeling indoor scene wireless channels based on laser radar perception reconstruction, which can solve the problems of low robustness and accuracy in the current communication scene reconstruction based on laser radar perception assistance.

[0006] In a first aspect, an embodiment of the present application provides a method for modeling wireless channels of indoor scenes based on laser radar perception reconstruction, including: step S1, using laser radar to collect dense three-dimensional point cloud data of indoor communication scenes, the three-dimensional point cloud data including three-dimensional point cloud data of ceilings, floors, walls, columns, doors, windows and furniture surfaces; step S2, according to the three-dimensional point cloud data in step S1, performing region cropping, filtering, denoising and point cloud downsampling preprocessing; step S3, projecting the three-dimensional point cloud data processed in step S2 onto a two-dimensional plane and performing grid segmentation to obtain a point cloud density map, using an algorithm combining a random sampling consistency algorithm with a Hough transform to extract key facade structure parameters from the point cloud density map, the key facade structure parameters including wall structure parameters and cylindrical structure parameters, and detecting and extracting height parameters of the ceiling structure and height parameters of the floor structure from the point cloud density map by calculating the statistical histogram on the height axis; step Step S4: According to the key facade structure parameters in step S3, the key facade structure is projected onto a two-dimensional plane to obtain a wall image, and the door and window structure on the wall image is extracted based on the idea of ​​image semantic segmentation, and the point cloud of the key facade structure, the ceiling structure and the floor structure surface in step S3 is removed to obtain the internal furniture surface point cloud data, and a set of furniture parameters is generated for each furniture in the internal furniture surface point cloud data by adjusting the shape and position parameters of the pre-created furniture template; Step S5: According to the height parameters of the ceiling structure, the height parameters of the floor structure, the key facade structure parameters and the furniture parameters extracted in steps S2, S3 and S4, a three-dimensional model of the indoor scene is constructed in combination with a triangulation algorithm, and an electromagnetic simulation model is obtained after matching the electromagnetic parameters of the three-dimensional model, and a ray tracing algorithm is used to simulate the propagation process of the electromagnetic signal in the indoor scene to simulate the channel characteristics of the receiving point.

[0007] In a possible implementation manner of the first aspect, step S2 specifically includes:

[0008] Step S201, cropping the collected three-dimensional point cloud data in three-dimensional space using a bounding box to remove point cloud data outside the room;

[0009] Step S202: De-noising the three-dimensional point cloud data obtained in step S201 according to an outlier removal algorithm, and down-sampling the three-dimensional point cloud data using a random sampling algorithm to compress the data size.

[0010] Optionally, in another possible implementation of the first aspect, the step S3 specifically includes:

[0011] Step S301: Reduce the dimension of the three-dimensional point cloud data and project it onto a two-dimensional plane, and divide the grid based on a preset grid size to obtain a grid with a height of N. h , width N w Point cloud density map

[0012] Step S302: extract key facade structures from the point cloud density map obtained in step S301 by combining Hough transform and random sampling consistency algorithm, and remove point cloud data on the surface of the key facade structure to update the point cloud density map;

[0013] Step S303: for the key facade structures extracted in step S302, the best connection mode between the key facade structures is fitted according to the residual point cloud of the connection area between the endpoints of the key facade structures, and the topological relationship between the key facade structures is reconstructed;

[0014] Step S304: for the point cloud density map updated in step S302, the height parameters of the ceiling structure and the height parameters of the floor structure are screened out based on the statistical histogram on the height axis.

[0015] Optionally, in another possible implementation of the first aspect, the step S4 specifically includes:

[0016] Step S401, projecting the key facade structure extracted in step S3 onto the vertical facade corresponding to the facade structure to construct a wall image;

[0017] Step S402: for the wall image in step S401, semantic segmentation is performed based on the zero-sample image semantic segmentation model SegmentAnything 2 to extract the door and window structure on the wall image, and the opening structure of the wall corresponding to the door and window structure is obtained according to the door and window structure on the wall image;

[0018] Step S403: For the ceiling, floor and key facade structures extracted in step S3, the point cloud data of the corresponding structure surface is removed from the point cloud data. For the common furniture in the indoor scene, a furniture template defined by a set of shape parameters and position parameters is pre-built, and the furniture is fitted from the remaining point cloud data by adjusting the shape parameters and position parameters to achieve 3D reconstruction of the furniture.

[0019] Optionally, in another possible implementation of the first aspect, the step S5 specifically includes:

[0020] Step S501: when performing three-dimensional modeling on the key facade structure, sequential three-dimensional modeling is performed according to the key facade structure parameters extracted in step S3 in the direction of wall connection, and each wall section is further segmented based on the left and right boundaries and the upper and lower boundaries of the opening structure of the wall extracted in step S4 to complete the segmentation of the door and window area;

[0021] Step S502: For the polygons of the key facade structure projected on the two-dimensional plane extracted in step S3 and the height parameters of the ceiling structure and the height parameters of the floor structure, the polygons are segmented into triangular facets based on the ear clipping algorithm to complete the three-dimensional modeling of the ceiling structure and the floor structure;

[0022] Step S503, determining the facets to which different structures belong, and performing electromagnetic parameter matching on different facets based on electromagnetic material parameters recommended by ITU to construct an electromagnetic simulation model;

[0023] Step S504: import the electromagnetic simulation model obtained in step S503 into the ray tracing channel simulation software Wireless Insite, configure the parameters of the transmitting and receiving antennas, the frequency and bandwidth of the transmission signal to ensure that the simulation conditions match the actual communication environment, and calculate the propagation path of the electromagnetic wave in the indoor communication scenario through the ray tracing algorithm, and record the reflection and diffraction electromagnetic propagation phenomena of the electromagnetic wave in the propagation path and the surrounding environment.

[0024] Optionally, in another possible implementation of the first aspect, the algorithm combining the random sampling consistency algorithm with the Hough transform to extract key facade structure parameters from the point cloud density map includes:

[0025] The key facade structure is modeled as a straight line on the plane projection, represented by a set of parameters Ψ vplane ={θ,rho} describes the density weighted parameter subspace of the point cloud density map calculated based on the Hough transform

[0026]

[0027] in is the parameter Ψ vplane = {θ, rho} The subset of density grid points S through which the straight line described by θ,rho The corresponding grid clustering density, the clustering algorithm used is DBSCAN;

[0028] Density-weighted parameter subspace Select the best sampling set S optimal :

[0029]

[0030] According to the random sampling consensus algorithm, the best sampling set S optimal The wall structure parameters are extracted from the random sampling consensus algorithm, where the sampling threshold ε corresponding to the optimal sampling set S optimal The width w obtained by pre-clustering is calculated as:

[0031]

[0032] Among them, k1.k2 is used to describe the influence of different wall widths on threshold selection, ε max ,ε min are the maximum and minimum values ​​selected by the random sampling consensus algorithm threshold.

[0033] The present application provides a method for modeling indoor scene wireless channels based on laser radar perception reconstruction. First, a laser radar is used to collect dense three-dimensional point cloud data of the indoor communication scene. The three-dimensional point cloud data includes three-dimensional point cloud data of the ceiling, floor, wall, column, door, window and furniture surface. Then, according to the three-dimensional point cloud data in step S1, regional cropping, filtering, denoising and point cloud downsampling preprocessing are performed. The three-dimensional point cloud data processed in step S2 is projected onto a two-dimensional plane and meshed to obtain a point cloud density map. An algorithm combining a random sampling consistency algorithm and a Hough transform is used to extract key facade structure parameters from the point cloud density map. The key facade structure parameters include wall structure parameters and column structure parameters. The height parameters of the ceiling structure and the height parameters of the floor structure are detected and extracted from the point cloud density map by calculating the statistical histogram on the height axis. Then, according to The key facade structure parameters in step S3 project the key facade structure onto a two-dimensional plane to obtain a wall image, and extract the door and window structure on the wall image based on the idea of ​​image semantic segmentation, remove the point cloud of the key facade structure, ceiling structure and floor structure surface in step S3, obtain the internal furniture surface point cloud data, and generate a set of furniture parameters for each furniture in the internal furniture surface point cloud data by adjusting the shape and position parameters of the pre-created furniture template. Finally, according to the height parameters of the ceiling structure, the height parameters of the floor structure, the key facade structure parameters and the furniture parameters extracted in steps S2, S3 and S4, a three-dimensional model of the indoor scene is constructed in combination with a triangulation algorithm. After matching the electromagnetic parameters of the three-dimensional model, an electromagnetic simulation model is obtained, and a ray tracing algorithm is used to simulate the propagation process of electromagnetic signals in the indoor scene to simulate the channel characteristics of the receiving point. The indoor communication scene reconstruction method proposed by the present invention divides the complex scene into structural elements and internal elements, and designs corresponding algorithms for different elements to extract key geometric features, thereby improving the reconstruction accuracy and robustness. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0035] Figure 1 A schematic diagram of a method for indoor communication scene reconstruction and channel modeling based on laser radar perception assistance provided in an implementation of the present application;

[0036] Figure 2 A schematic diagram of a wall structure extracted based on the RANSAC algorithm and the Hough transform algorithm provided in an embodiment of the present application;

[0037] Figure 3 A schematic diagram of the discrete wall connection effect provided in one embodiment of the present application;

[0038] Figure 4 A schematic diagram of point cloud semantic segmentation provided in an embodiment of the present application;

[0039] Figure 5 A schematic diagram of an L-shaped workstation table furniture template provided in one embodiment of the present application;

[0040] Figure 6 A schematic diagram of furniture geometric model reconstruction based on a furniture template provided in an embodiment of the present application;

[0041] Figure 7 A schematic diagram of door and window opening structure extraction based on a semantic segmentation algorithm provided in an embodiment of the present application;

[0042] Figure 8 A schematic diagram of a ray tracing channel simulation provided in an embodiment of the present application;

[0043] Fig. 9 This is a ray tracing channel simulation path loss simulation result provided in one embodiment of the present application. DETAILED DESCRIPTION

[0044] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present application. However, it should be clear to those skilled in the art that the present application may also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present application.

[0045] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0046] It should also be understood that the term “and / or” used in the specification and appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0047] As used in the specification and appended claims of this application, the term "if" can be interpreted as "when" or "uponce" or "in response to determining" or "in response to detecting", depending on the context. Similarly, the phrase "if it is determined" or "if [described condition or event] is detected" can be interpreted as meaning "uponce it is determined" or "in response to determining" or "uponce [described condition or event] is detected" or "in response to detecting [described condition or event]", depending on the context.

[0048] In addition, in the description of the present application specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0049] References to "one embodiment" or "some embodiments" etc. described in the specification of this application mean that one or more embodiments of the present application include specific features, structures or characteristics described in conjunction with the embodiment. Therefore, the statements "in one embodiment", "in some embodiments", "in some other embodiments", "in some other embodiments", etc. that appear in different places in this specification do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0050] Example 1.

[0051] See also Figure 1-Figure 9 ,like Figure 1 As shown, the indoor scene wireless channel modeling method based on laser radar perception reconstruction includes the following steps:

[0052] Step S1, using a laser radar to collect dense three-dimensional point cloud data of an indoor communication scene, the three-dimensional point cloud data including three-dimensional point cloud data of ceilings, floors, walls, columns, doors, windows and furniture surfaces;

[0053] Step S2: performing region cropping, filtering, denoising and point cloud downsampling preprocessing according to the three-dimensional point cloud data in step S1;

[0054] Step S3, projecting the three-dimensional point cloud data processed in step S2 onto a two-dimensional plane and performing grid segmentation to obtain a point cloud density map, extracting key facade structure parameters from the point cloud density map using an algorithm combining a random sampling consistency algorithm with a Hough transform, the key facade structure parameters including wall structure parameters and cylindrical structure parameters, and detecting and extracting height parameters of the ceiling structure and height parameters of the floor structure from the point cloud density map by calculating a statistical histogram on the height axis;

[0055] Step S4: projecting the key facade structure onto a two-dimensional plane to obtain a wall image according to the key facade structure parameters in step S3, extracting the door and window structure on the wall image based on the idea of ​​image semantic segmentation, removing the point cloud of the key facade structure, ceiling structure and floor structure surface in step S3, obtaining the internal furniture surface point cloud data, and generating a set of furniture parameters for each piece of furniture in the internal furniture surface point cloud data by adjusting the shape and position parameters of the pre-created furniture template;

[0056] Step S5: construct a three-dimensional model of the indoor scene based on the height parameters of the ceiling structure, the height parameters of the floor structure, the key facade structure parameters and the furniture parameters extracted in steps S2, S3 and S4, and combine the triangulation algorithm to obtain an electromagnetic simulation model after matching the electromagnetic parameters of the three-dimensional model. Use a ray tracing algorithm to simulate the propagation process of the electromagnetic signal in the indoor scene to simulate the channel characteristics of the receiving point.

[0057] Furthermore, in a possible implementation of the embodiment of the present application, the above step S2 may include:

[0058] Step S201, cropping the collected three-dimensional point cloud data in three-dimensional space using a bounding box to remove point cloud data outside the room;

[0059] Step S202: De-noising the three-dimensional point cloud data obtained in step S201 according to an outlier removal algorithm, and down-sampling the three-dimensional point cloud data using a random sampling algorithm to compress the data size.

[0060] Furthermore, in a possible implementation of the embodiment of the present application, the above step S3 may include:

[0061] Step S301: Reduce the dimension of the three-dimensional point cloud data and project it onto a two-dimensional plane, and divide the grid based on a preset grid size to obtain a grid with a height of N. h , width N w Point cloud density map

[0062] Step S302: extract key facade structures from the point cloud density map obtained in step S301 by combining Hough transform and random sampling consistency algorithm, and remove point cloud data on the surface of the key facade structure to update the point cloud density map;

[0063] Step S303: for the key facade structures extracted in step S302, the best connection mode between the key facade structures is fitted according to the residual point cloud of the connection area between the endpoints of the key facade structures, and the topological relationship between the key facade structures is reconstructed;

[0064] Step S304: for the point cloud density map updated in step S302, the height parameters of the ceiling structure and the height parameters of the floor structure are screened out based on the statistical histogram on the height axis.

[0065] Furthermore, in a possible implementation of the embodiment of the present application, the above step S4 may include:

[0066] Step S401, projecting the key facade structure extracted in step S3 onto the vertical facade corresponding to the facade structure to construct a wall image;

[0067] Step S402: for the wall image in step S401, semantic segmentation is performed based on the zero-sample image semantic segmentation model SegmentAnything 2 to extract the door and window structure on the wall image, and the opening structure of the wall corresponding to the door and window structure is obtained according to the door and window structure on the wall image;

[0068] Step S403: For the ceiling, floor and key facade structures extracted in step S3, the point cloud data of the corresponding structure surface is removed from the point cloud data. For the common furniture in the indoor scene, a furniture template defined by a set of shape parameters and position parameters is pre-built, and the furniture is fitted from the remaining point cloud data by adjusting the shape parameters and position parameters to achieve 3D reconstruction of the furniture. Figure 5 As shown, the template created in this embodiment is described by ten shape parameters, including the left desktop length l A , right desktop length l B 、Left desktop width w A 、Right desktop width w B , Left partition length l DA , left partition width w DA , right partition length l DB , right partition width w DB , table top height h and partition height h D .

[0069] Furthermore, in a possible implementation of the embodiment of the present application, the above step S5 may include:

[0070] Step S501: when performing three-dimensional modeling on the key facade structure, sequential three-dimensional modeling is performed according to the key facade structure parameters extracted in step S3 in the direction of wall connection, and each wall section is further segmented based on the left and right boundaries and the upper and lower boundaries of the opening structure of the wall extracted in step S4 to complete the segmentation of the door and window area;

[0071] Step S502: For the polygons of the key facade structure projected on the two-dimensional plane extracted in step S3 and the height parameters of the ceiling structure and the height parameters of the floor structure, the polygons are segmented into triangular facets based on the ear clipping algorithm to complete the three-dimensional modeling of the ceiling structure and the floor structure;

[0072] Step S503, determining the facets to which different structures belong, and performing electromagnetic parameter matching on different facets based on electromagnetic material parameters recommended by ITU to construct an electromagnetic simulation model;

[0073] Step S504: import the electromagnetic simulation model obtained in step S503 into the ray tracing channel simulation software Wireless Insite, configure the parameters of the transmitting and receiving antennas, the frequency and bandwidth of the transmission signal to ensure that the simulation conditions match the actual communication environment, and calculate the propagation path of the electromagnetic wave in the indoor communication scenario through the ray tracing algorithm, and record the reflection and diffraction electromagnetic propagation phenomena of the electromagnetic wave in the propagation path and the surrounding environment.

[0074] In the embodiment of the present application, the algorithm combining the random sampling consistency algorithm with the Hough transform is used to extract key facade structural parameters from the point cloud density map, specifically including:

[0075] The key facade structure is modeled as a straight line on the plane projection, represented by a set of parameters Ψ vplane ={θ,rho} describes the density weighted parameter subspace of the point cloud density map calculated based on the Hough transform

[0076]

[0077] in is the parameter Ψ vplane = {θ, rho} The subset of density grid points S through which the straight line described by θ,tho The corresponding grid clustering density, the clustering algorithm used is DBSCAN;

[0078] Density-weighted parameter subspace Select the best sampling set S optimal :

[0079]

[0080] According to the random sampling consensus algorithm, the best sampling set S optimal The wall structure parameters are extracted from the random sampling consensus algorithm, where the sampling threshold ε corresponding to the optimal sampling set S optimal The width w obtained by pre-clustering is calculated as:

[0081]

[0082] Among them, k1.k2 is used to describe the influence of different wall widths on threshold selection, ε max ,ε min are the maximum and minimum values ​​selected by the random sampling consensus algorithm threshold.

[0083] The wireless channel modeling method for indoor scenes based on lidar perception reconstruction provided in the present application is a method for extracting geometric features of key objects in indoor scenes by utilizing the high-precision perception characteristics of lidar, extracting key wall structures by utilizing random sampling consistency algorithm and Hough transform algorithm, and then realizing high-precision furniture reconstruction and door and window structure extraction by utilizing the ideas of furniture template fitting and image semantic segmentation. Finally, the channel characteristics of indoor scenes are simulated by ray tracing simulation software, and the method is suitable for simulating high-precision channel characteristics in indoor scenes. Complex scenes are divided into structural elements and internal elements, and corresponding algorithms are designed for different elements to extract key geometric features, thereby improving reconstruction accuracy and robustness.

[0084] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0085] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A method for modeling indoor scene wireless channels based on laser radar perception reconstruction, characterized in that: The method comprises: Step S1, using a laser radar to collect dense three-dimensional point cloud data of an indoor communication scene, wherein the three-dimensional point cloud data includes three-dimensional point cloud data of ceilings, floors, walls, columns, doors, windows, and furniture surfaces; Step S2: performing region cropping, filtering, denoising and point cloud downsampling preprocessing according to the three-dimensional point cloud data in step S1; Step S3, projecting the three-dimensional point cloud data processed in step S2 onto a two-dimensional plane and performing grid segmentation to obtain a point cloud density map, extracting key facade structure parameters from the point cloud density map using an algorithm combining a random sampling consistency algorithm with a Hough transform, wherein the key facade structure parameters include wall structure parameters and cylindrical structure parameters, and detecting and extracting height parameters of a ceiling structure and a floor structure from the point cloud density map by calculating a statistical histogram on a height axis; Step S4, projecting the key facade structure onto a two-dimensional plane to obtain a wall image according to the key facade structure parameters in step S3, extracting the door and window structures on the wall image based on the idea of ​​image semantic segmentation, removing the point clouds of the key facade structure, the ceiling structure and the floor structure surface in step S3, obtaining internal furniture surface point cloud data, and generating a set of furniture parameters for each piece of furniture in the internal furniture surface point cloud data by adjusting the shape and position parameters of the pre-created furniture template; Step S5, based on the height parameters of the ceiling structure, the height parameters of the floor structure, the key facade structure parameters and the furniture parameters extracted in steps S2, S3 and S4, a three-dimensional model of the indoor scene is constructed in combination with a triangulation algorithm, and an electromagnetic simulation model is obtained after electromagnetic parameter matching of the three-dimensional model, and a ray tracing algorithm is used to simulate the propagation process of the electromagnetic signal in the indoor scene to simulate the channel characteristics of the receiving point.

2. The method according to claim 1, characterized in that The step S2 specifically includes: Step S201, cropping the collected three-dimensional point cloud data using a bounding box in three-dimensional space to remove point cloud data outside the room; Step S202: De-noising the three-dimensional point cloud data obtained in step S201 according to an outlier removal algorithm, and down-sampling the three-dimensional point cloud data using a random sampling algorithm to compress the data size.

3. The method according to claim 2, characterized in that The step S3 specifically includes: Step S301: Reduce the dimension of the three-dimensional point cloud data and project it onto a two-dimensional plane, and divide the grid based on a preset grid size to obtain a grid with a height of N. h , width N w The point cloud density map Step S302: extracting key facade structures from the point cloud density map obtained in step S301 by combining Hough transform and random sampling consistency algorithm, and removing point cloud data on the surface of the key facade structure to update the point cloud density map; Step S303: for the key facade structures extracted in step S302, the best connection mode between the key facade structures is fitted according to the residual point cloud of the connection area between the endpoints of the key facade structures, and the topological relationship between the key facade structures is reconstructed; Step S304: for the point cloud density map updated in step S302, the height parameter of the ceiling structure and the height parameter of the floor structure are screened out based on the statistical histogram on the height axis.

4. The method according to claim 3, characterized in that The step S4 specifically includes: Step S401, projecting the key facade structure extracted in step S3 onto the vertical facade corresponding to the facade structure to construct the wall image; Step S402: for the wall image in step S401, semantic segmentation is performed based on the zero-sample image semantic segmentation model SegmentAnything 2 to extract the door and window structure on the wall image, and the opening structure of the wall corresponding to the door and window structure is obtained according to the door and window structure on the wall image; Step S403: for the ceiling, floor and key facade structures extracted in step S3, the point cloud data of the corresponding structure surfaces are removed from the point cloud data, and a furniture template defined by a set of shape parameters and position parameters is pre-built for common furniture in indoor scenes. The furniture is fitted from the remaining point cloud data by adjusting the shape parameters and position parameters to achieve three-dimensional reconstruction of the furniture.

5. The method according to claim 4, characterized in that The step S5 specifically includes: Step S501: when performing three-dimensional modeling on the key facade structure, sequential three-dimensional modeling is performed according to the key facade structure parameters extracted in step S3 in the direction of wall connection, and each wall section is further segmented based on the left and right boundaries and the upper and lower boundaries of the opening structure of the wall extracted in step S4 to complete the segmentation of the door and window area; Step S502: For the polygons of the key facade structure projected on the two-dimensional plane extracted in step S3 and the height parameters of the ceiling structure and the height parameters of the floor structure, the polygons are segmented into triangular facets based on the ear clipping algorithm to complete the three-dimensional modeling of the ceiling structure and the floor structure; Step S503, determining the facets to which different structures belong, and performing electromagnetic parameter matching on different facets based on electromagnetic material parameters recommended by ITU to construct an electromagnetic simulation model; Step S504: import the electromagnetic simulation model obtained in step S503 into the ray tracing channel simulation software Wireless Insite, configure the parameters of the transmitting and receiving antennas, the frequency and bandwidth of the transmission signal to ensure that the simulation conditions match the actual communication environment, and calculate the propagation path of the electromagnetic wave in the indoor communication scenario through the ray tracing algorithm, and record the reflection and diffraction electromagnetic propagation phenomena of the electromagnetic wave in the propagation path and the surrounding environment.

6. The method according to claim 3, characterized in that The algorithm combining random sampling consistency algorithm and Hough transform is used to extract key facade structural parameters from the point cloud density map, including: The key facade structure is modeled as a straight line on a plane projection, represented by a set of parameters Ψ vplane ={θ, rho} describes the density weighted parameter subspace of the point cloud density map calculated based on the Hough transform in is the parameter Ψ vplane = {θ, rho} The subset of density grid points S through which the straight line described by θ,rho The corresponding grid clustering density, the clustering algorithm used is DBSCAN; Density-weighted parameter subspace Select the best sampling set S optimal : According to the random sampling consensus algorithm, the best sampling set S optimal The wall structure parameters are extracted from the random sampling consensus algorithm, where the sampling threshold ε corresponding to the optimal sampling set S optimal The width w obtained by pre-clustering is calculated as: Among them, k1.k2 is used to describe the influence of different wall widths on threshold selection, ε max , ε min is the maximum and minimum of the threshold selected by the random sampling consensus algorithm.

Images