A multi-level LOD generation method for real scene three-dimensional building model data

Abstract

The application discloses a kind of real scene three-dimensional building model data multistage LOD generation method, it is related to three-dimensional model light weight technical field, comprising the following steps: S1: according to the level of building model and the lightweight degree of visual condition, establish multistage LOD respectively as LOD0, LOD1, LOD2, LOD3, define the grading standard of each LOD;S2: according to the OSGB data of building model, using the method based on object to extract the vector surface of target building, namely LOD0;S3: based on the vector surface data of building model, in combination with the building elevation data contained in OSGB, the method of the application is more highly automated, more simple operation, both can reduce the volume of three-dimensional model data, can also maintain the basic structure of model under different LOD levels, can complete the conversion between building models under limited data, construct different dimensions three-dimensional building model monomer, and apply it to different levels and scene, meet the needs of city space analysis and management.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of three-dimensional model lightweight, and particularly relates to a multi-level LOD generation method for real scene three-dimensional building model data. BACKGROUND

[0002] The real scene three-dimensional China construction is an important measure to implement the digital China and digital economic strategy, and is also a specific arrangement to implement the national new infrastructure construction, aiming to coordinate from the national level, adopt a unified three-dimensional space reference, and establish a high-precision, realistic and authoritative three-dimensional geographic space database covering the whole country to meet the demand of natural resource management and economic and social development in various fields for three-dimensional space public base. The real scene three-dimensional China construction transforms and upgrades the traditional basic surveying and mapping business, adopts special technical means such as stereoscopic reconstruction, entity modeling and realistic description, digitizes the relevant geographic space entities in the national space category, develops multi-scale, time-series and correlated real scene three-dimensional information products, constructs real scene three-dimensional digital space, and supports the construction and application of digital China and peaceful China.

[0003] With the continuous development of the real scene three-dimensional construction, the volume of three-dimensional building model data is increasing, which brings challenges to model storage and real-time loading display. In different application scenarios, the texture details of the building model are not necessary, and the texture details of the building model required by different application scenarios are different in level, so it is more inclined to faster data transmission, smooth real-time loading display and lower memory occupation. Therefore, for the real scene three-dimensional building model data, it is very important to construct multi-level LOD model. The traditional model LOD grading is based on CityGML (City Geographic Markup Language) to grade the level of detail, and this method is more used in building information model (BIM). The disadvantage of this method is that the construction of BIM model needs to consume a lot of manpower and material resources, and a large amount of prior knowledge is needed for the simplification of different LOD levels corresponding to different buildings, and with the continuous increase of three-dimensional model data, this method cannot meet the demand of subsequent batch processing of three-dimensional models. Therefore, a multi-level LOD generation method for real scene three-dimensional building model data is proposed. SUMMARY

[0004] The purpose of the present application is to solve the problems in the prior art, and a multi-level LOD generation method for real scene three-dimensional building model data is proposed.

[0005] A multi-level LOD generation method for real scene three-dimensional building model data, comprising the following steps:

[0006] S1: According to the level visible condition and lightweight degree of the building model, multi-level LOD is established, which is LOD0, LOD1, LOD2 and LOD3, and the grading standard of each level of LOD is defined.

[0007] S2: extracting a vector surface of the target building, i.e. LOD0, by using an object-oriented method based on the OSGB data of the building model;

[0008] S3: obtaining a block-shaped building, i.e. LOD1, by using a vector stretching method based on the vector surface data of the building model and the building elevation data contained in the OSGB;

[0009] S4: obtaining a building monomer model, i.e. LOD3, by using a range clipping method based on the vector surface data of the building model;

[0010] S5: simplifying the building monomer model by using a topological reconstruction method based on the building monomer model data, so that the building monomer model meets the LOD2 grading standard.

[0011] Preferably, in the step S1, the building structure corresponding to the building model of the LOD0 level is a two-dimensional plane, and the corresponding grading standard is a building vector surface; the building structure corresponding to the building model of the LOD1 level is a block-shaped body, and the corresponding grading standard is a minimum circumscribed body; the building structure corresponding to the building model of the LOD2 level is a differentiated roof and a block-shaped body, and the corresponding grading standard is the actual size of the building; the building structure corresponding to the building model of the LOD3 level contains a real roof, a building accessory structure and a building texture, and the corresponding grading standard is that the building model contains external detailed structures.

[0012] Preferably, in the step S2, the extraction of the vector surface of the target building comprises the following steps:

[0013] (1) generating digital surface model (DSM) data and digital orthophoto (DOM) data according to the OSGB data of the building model;

[0014] (2) inputting the digital DSM data and the DOM data;

[0015] (3) separating the building objects by using a threshold segmentation method based on the height information contained in the DSM data;

[0016] (4) separating the vegetation based on the building objects obtained in the step (3) to realize the re-optimization of the building extraction result;

[0017] (5) merging all the building classification objects;

[0018] (6) removing the non-building objects by using a threshold segmentation method based on the building size, so as to complete the classification of the building based on the building objects obtained in the step (5);

[0019] (7) Manually optimize the building extraction result.

[0020] Preferably, in the step S3, the method of vector stretching is used to generate the LOD1 level building monomer model, which includes the following steps:

[0021] (1) Based on the vector data extracted in step S2, the partition statistics is performed to obtain the elevation data of the corresponding building in the DSM data;

[0022] (2) The vector surface is assigned with the elevation information of the corresponding building to generate the block-shaped building.

[0023] Preferably, in the step S4, the range clipping is performed on the urban three-dimensional model data to obtain the LOD3 level building monomer model.

[0024] Preferably, in the step S5, the method of topological reconstruction includes the following steps:

[0025] (1) Set the acute angle threshold value and convert the angle threshold value to radians;

[0026] (2) Set the target edge length and store it;

[0027] (3) Set the target number of triangles and store it;

[0028] (4) Construct an axis-aligned bounding box tree, create an axis-aligned bounding box for each triangle, and construct a bounding box tree;

[0029] (5) Obtain the average edge length of the initial mesh;

[0030] (6) Reconstruct the mesh, perform the mesh reconstruction process according to the specified number of iterations, including the steps of splitting long edges, folding short edges, flipping edges, moving vertices, and projecting vertices;

[0031] (7) Split the long edge, traverse each edge of each face, and if the edge length exceeds the square of the specified threshold, perform the edge splitting operation;

[0032] (8) Fold the short edge, traverse each edge of each face, and if the edge length is less than the square of the minimum threshold and both endpoints of the edge are not feature points, perform the edge folding operation;

[0033] (9) Flip the edge, traverse each edge of each face, and if the edge has an opposite edge, try to perform the edge flipping operation;

[0034] (10) Move the vertex, update the degree of the vertex, the triangle normal and the vertex normal, and adjust the position of each vertex;

[0035] (11) Projecting vertex, for each non-feature vertex, according to its normal direction, ray projection is carried out along a bounding box, the triangle intersecting with it is found, and the vertex is projected on the nearest intersection point.

[0036] Compared with the prior art, the present application has the advantages of:

[0037] The method of the present application has higher automation degree and simpler operation, can reduce the volume of three-dimensional model data, maintain the basic structure of the model at different LOD levels, complete the conversion between building models in limited data, construct three-dimensional building model monomers of different dimensions, and apply them to different levels and scenes, and meet the needs of city space analysis and management. BRIEF DESCRIPTION OF DRAWINGS

[0038] Figure 1 The figure is a flowchart of the present application.

[0039] Figure 2 The figure is a hierarchical diagram of different LOD level models in the present application.

[0040] Figure 3 The figure is a diagram of LOD0 level model in the present application.

[0041] Figure 4 The figure is a diagram of LOD1 level model in the present application.

[0042] Figure 5 The figure is a diagram of LOD2 level model in the present application.

[0043] Figure 6 The figure is a diagram of LOD3 level model in the present application.

[0044] Figure 7 The figure is a diagram of different LOD level models in the present application. DETAILED DESCRIPTION

[0045] In order to make the technical means, creative features, purposes and effects of the present application easy to understand, the present application is further described below in combination with specific embodiments.

[0046] Referring to Figures 1-7 The figure is a diagram of different LOD level models in the present application.

[0047] S1: According to the hierarchical visibility condition and lightweight degree of the building model, multi-level LODs are established, which are LOD0, LOD1, LOD2 and LOD3, and the grading standards of each LOD are defined;

[0048] S2: According to the OSGB data of the building model, the vector surface of the target building, i.e. LOD0, is extracted by using an object-oriented method;

[0049] S3: Based on the vector surface data of the building model, combined with the building elevation data contained in the OSGB, the block-shaped building, i.e. LOD1, is obtained by vector stretching;

[0050] S4: Based on the vector surface data of the building model, the building monomer model, i.e. LOD3, is obtained by range clipping;

[0051] S5: Based on the building monomer model data, the building monomer model is simplified by a topological reconstruction method so as to meet the LOD2 grading standard.

[0052] In the step S1, the building structure corresponding to the building model of the LOD0 level is a two-dimensional plane, and the corresponding grading standard is a building vector surface; the building structure corresponding to the building model of the LOD1 level is a block-shaped body, and the corresponding grading standard is a minimum circumscribed body; the building structure corresponding to the building model of the LOD2 level is a differentiated roof and a block-shaped body, and the corresponding grading standard is the actual size of the building; the building structure corresponding to the building model of the LOD3 level contains a real roof, a building accessory structure and a building texture, and the corresponding grading standard is that the building model contains external detailed structures.

[0053] In the step S2, the extraction of the vector surface of the target building includes the following steps:

[0054] (1) According to the OSGB data of the building model, digital surface model (DSM) data and digital orthophoto (DOM) data are generated;

[0055] (2) The digital DSM data and DOM data are inputted;

[0056] (3) Based on the height information contained in the DSM data, the separation of the building object is realized by threshold segmentation;

[0057] (4) On the basis of the building object obtained in step (3), the spectral characteristics of the vegetation are utilized to separate the same, so as to realize the re-optimization of the building extraction result;

[0058] (5) All building classification objects are merged;

[0059] (6) On the basis of the building object obtained in step (5), the non-building object is removed by threshold segmentation according to the size of the building, so as to complete the classification of the building;

[0060] (7) The building extraction result is manually optimized.

[0061] The step S3, the method of vector stretching to generate the LOD1 level building single model includes the following steps:

[0062] (1) Based on the vector data extracted in step S2, the elevation data of the corresponding building in the DSM data is obtained through partition statistics;

[0063] (2) The building corresponding to the vector surface is given the elevation information to generate the block building.

[0064] In step S4, the range of the city three-dimensional model data is cut to obtain the LOD3 level building single model.

[0065] In step S5, a topological reconstruction method includes the following steps:

[0066] (1) Set the acute angle threshold, and convert the angle threshold to radian;

[0067] (2) Set the target edge length and store;

[0068] (3) Set the target number of triangles and store;

[0069] (4) Construct an axis-aligned bounding box tree, create an axis-aligned bounding box for each triangle, and construct a bounding box tree;

[0070] (5) Get the average edge length of the initial mesh;

[0071] (6) Reconstruct the mesh, and perform the mesh reconstruction process according to the specified number of iterations, including the steps of splitting long edges, folding short edges, flipping edges, moving vertices, and projecting vertices;

[0072] (7) Split the long edge, traverse each edge of each face, and if the edge length exceeds the square of the specified threshold, perform the edge splitting operation;

[0073] (8) Fold the short edge, traverse each edge of each face, and if the edge length is less than the square of the minimum threshold and both endpoints of the edge are not feature points, then perform the edge folding operation;

[0074] (9) Flip the edge, traverse each edge of each face, and if the edge has an opposite edge, try to perform the edge flipping operation;

[0075] (10) Move the vertex, update the degree of the vertex, the triangle normal and the vertex normal, and adjust the position of each vertex;

[0076] (11) Project the vertex, for each non-feature vertex, according to the normal direction, perform ray projection along a bounding box, find the triangle intersected by it, and project the vertex to the nearest intersection point.

[0077] The working process and principles of the present application are as follows:

[0078] A1: LOD grading is performed on the building model, and LOD0, LOD1, LOD2 and LOD3 are obtained.

[0079] A2: Prepare the oblique three-dimensional model (OSGB) data, load the tiles, select the tiles according to the range, and output the DOM and DSM data;

[0080] A3: Import the DOM and DSM data generated by the oblique photography model;

[0081] A4: Based on the RGB three-channel data of the DSM and the DOM, multi-scale segmentation is performed on the pixel layer, the scale is set to 95, and the shape and compactness weights are set to 0.4 and 0.5 respectively;

[0082] A5: The mean elevation range is set to be greater than or equal to 30, the building objects are classified by the algorithm, that is, the threshold condition is set to Mean DSM>=30, and the classification algorithm is set to the building category;

[0083] A6: The classified objects are optimized, the spectral characteristics of the vegetation target are used to separate them, the green channel ratio feature is constructed, that is, green_ratio=mean ofgreen / (mean ofred+mean ofgreen+mean ofblue), the characteristics of the vegetation in the green channel are amplified, and the rule "mean of green / (mean ofred+mean ofgreen+mean ofblue)>=0.36" is used to remove the vegetation objects in the building classification objects that meet the characteristics;

[0084] A7: The building objects obtained by the segmentation classification are merged by the algorithm;

[0085] A8: The building objects obtained after step A7 are compared, and the best threshold value is set to be greater than or equal to 5200 pixels according to the size of the building area to execute the classification algorithm, and the non-building objects with small area are separated from the building category;

[0086] A9: Some classified incomplete building objects and some non-building objects mistakenly classified as building objects are manually edited, and finally the LOD0 level building vector surface is obtained;

[0087] A10: Import the building vector data and the DSM raster data;

[0088] A11: Open the partition statistics tool, input the building vector face, summary field and DSM raster data, execute to get the partition statistics table, associate the partition statistics result to the building vector face, and add the average elevation field to the building vector face;

[0089] A12: Import the building vector face data with the added average elevation field, perform the stretching operation, and generate the LOD1 level building block;

[0090] A13: According to the building vector face data obtained in A9, perform range clipping on the OSGB oblique photography data to obtain the LOD3 level building monomer model data;

[0091] A14: Calculate the included angle of the edge;

[0092] A15: Set the scale factor for converting the angle to radian;

[0093] A16: Convert the angle value to radian for subsequent processing, judgment and calculation;

[0094] A17: Set the target triangle side length for subsequent use in the mesh reconstruction process;

[0095] A18: Set the target triangle number for subsequent use in the mesh reconstruction process;

[0096] A19: Create a vector storing the axis-aligned bounding box, which has the same size as the target triangle number, and each element is an object representing the axis-aligned bounding box;

[0097] A20: Traverse each triangle, add the triangle data to the corresponding axis-aligned bounding box, and update the center point of the bounding box;

[0098] A21: Create a vector storing the face index, add the index of each triangle to the vector, which has the same size as the target triangle number;

[0099] A22: Create a total axis-aligned bounding box for storing the total axis-aligned bounding box after combining all triangles;

[0100] A23: Update the center point of the axis-aligned bounding box, delete the old axis-aligned bounding box tree, create a new axis-aligned bounding box tree, and pass in the axis-aligned bounding box storing each triangle, the triangle index vector and the total axis-aligned bounding box as parameters;

[0101] A24: Define the constructor, accept two parameters, including the pointer to the vertex vector and the pointer to the triangle index;

[0102] A25: Initialize the pointer to the vertex vector to the parameter value of the constructor;

[0103] A26: Initialize the pointer to the triangle index vector to the parameter value of the constructor;

[0104] A27: Use the two parameters in A25 and A26 to initialize the mesh instance, the two parameters point to the actual data of the vertex and triangle index respectively;

[0105] A28: Calculate and store the initial average edge length for subsequent use in the mesh reconstruction process;

[0106] A29: Get the average edge length of the initial mesh;

[0107] A30: Delete the old triangle normal vector data, create new triangle normal vector data storage space, calculate the normal vector of each triangle and store it;

[0108] A31: If the target triangle number is set, calculate the total area of the current mesh, calculate the area of each target triangle, and calculate the target edge length according to the target triangle area;

[0109] A32: Calculate the minimum and maximum target edge length, as well as the square value of the minimum and maximum target edge length;

[0110] A33: Build an axis-aligned bounding box tree for the current vertex and triangle;

[0111] A34: If the acute angle edge threshold is set, split the long edge, update the triangle normal vector, and mark the acute angle edge, otherwise split the field edge and mark the boundary feature;

[0112] A35: According to the specified number of iterations, loop A36-A40 steps

[0113] A36: Split the overlong edge and merge the too short edge;

[0114] A37: Flip the edge that may improve the mesh quality;

[0115] A38: Adjust the vertex position;

[0116] A39: Project the vertex to the inside of the axis-aligned bounding box;

[0117] A40: Traverse each face in the mesh;

[0118] A41: Get the starting half-edge of the current face;

[0119] A42: Traverse each half-edge of the current face;

[0120] A43: Get the next half-edge of the current half-edge;

[0121] A44: Calculate the square of the length of the current half-edge and the next half-edge using the position information of the vertexes;

[0122] A45: If the square of the length exceeds the square of the maximum allowed edge length, perform a split operation on the edge where the current half-edge is located;

[0123] A46: After splitting, jump out of the current loop and continue to process the next half-edge;

[0124] A47: Obtain the reconstructed half-edge mesh;

[0125] A48: Traverse each face in the mesh;

[0126] A49: If the face has been marked for removal, skip processing;

[0127] A50: Obtain the index of the current face and the starting half-edge of the face;

[0128] A51: Traverse all half-edges of the current face starting from the starting half-edge;

[0129] A52: Obtain the next half-edge and calculate the square of the distance between the starting vertex of the current half-edge and the starting vertex of the next half-edge;

[0130] A53: If the length of the current half-edge and the next half-edge is less than the square of the minimum allowed length, enter the conditional judgment;

[0131] A54: Determine whether the starting vertex of the current half-edge and the starting vertex of the next half-edge are not feature vertices;

[0132] A55: If the condition of A54 is met, attempt to perform a folding operation on the current half-edge, and if successful, exit the current loop;

[0133] A56: Continue processing the next half-edge, and loop until all half-edges of the current face are processed, and when the starting half-edge is reached, end the loop;

[0134] A57: Move to the next face;

[0135] A58: Traverse each face in the mesh and obtain the starting half-edge of the current face;

[0136] A59: Traverse all half-edges of the current face starting from the starting half-edge;

[0137] A60: Obtain the next half-edge of the current half-edge;

[0138] A61: Determine whether the current half-edge can be flipped, i.e., whether there is an opposite half-edge;

[0139] A62: Check if a flip operation can be performed, and if the flip is successful, exit the half-edge traversal of the current face;

[0140] A63: Move to the next half-edge, loop until all half-edges of the current face are checked;

[0141] A64: Move to the next face;

[0142] A65: Update the vertex's degree information, update the triangle's normal vector, update the vertex's normal vector;

[0143] A66: Loop condition, check if the vertex pointer is not empty, adjust and optimize the current vertex's position;

[0144] A67: End of iteration, get the pointer of the next vertex, loop until all vertices of the current face are processed;

[0145] A68: Traverse each vertex in the mesh;

[0146] A69: If the vertex is a feature point, skip, get the first half-edge of the vertex, if the vertex has no half-edge, skip;

[0147] A70: Create a bounding box and update its position to the current vertex's position;

[0148] A71: Traverse the surrounding half-edges from the current vertex's first half-edge, update the bounding box's position to contain all related vertices;

[0149] A72: Create a bounding box tree;

[0150] A73: Test the current vertex's bounding box with the bounding boxes in the entire bounding box tree for intersection, and record the intersected triangle index pairs;

[0151] A74: Store the intersection points and their distance information;

[0152] A75: Calculate the ray segment length of the vertex's movement;

[0153] A76: Traverse each pair of intersected ray bounding boxes and triangle bounding boxes;

[0154] A77: Get the vertex position of the current triangle;

[0155] A78: Calculate the intersection of the ray and the triangle plane;

[0156] A79: Calculate the normal vector at the intersection point;

[0157] A80: Check the direction of the normal vector, select the intersection point that meets the conditions;

[0158] A81: If there is an intersection point that meets the conditions, select the intersection point with the smallest distance as the new position of the vertex;

[0159] A82: the loop ends, and the next vertex is processed.

[0160] In summary, the method of the present application has higher automation degree and simpler operation compared with the LOD grading of BIM model, solves the defects of excessive consumption of human and material resources, complex implementation process, long time consumption and difficult batch processing in the prior art, can reduce the volume of three-dimensional model data, maintain the basic structure of the model under different LOD levels, complete the conversion between building models with limited data, construct three-dimensional building model monomers of different dimensions, and apply them to different levels and scenes, meet the needs of city space analysis and management,

[0161] From the common technical knowledge, the present application can be realized by other embodiments without departing from the spirit or essential characteristics thereof. Therefore, the above disclosed embodiments are only illustrative in all aspects, and are not the only ones. All changes within the scope of the present application or within the scope equivalent to the present application are included in the present application.

Claims

1. A method for generating multi-level LOD (Level of Detail) data of a real-world 3D building model, characterized in that: The method comprises the following steps: S1: according to the hierarchical visual conditions and the lightweight degree of the building model, a plurality of LODs of LOD0, LOD1, LOD2 and LOD3 are established, and the grading standards of each LOD are defined; S2: according to the OSGB data of the building model, the vector surface of the target building is extracted by using an object-oriented method, namely LOD0; S3: based on the vector surface data of the building model, the building block is obtained by the vector stretching method combined with the building elevation data contained in the OSGB, namely LOD1; S4: based on the vector surface data of the building model, the building monomer model is obtained by range clipping of the OSGB oblique photography data, namely LOD3; S5: based on the building monomer model data, the building monomer model is simplified by using the topological reconstruction method, so that the building monomer model meets the LOD2 grading standard; In the step S1, the building structure corresponding to the LOD0 level building model is a two-dimensional plane, and the corresponding grading standard is a building vector surface; the building structure corresponding to the LOD1 level building model is a block, and the corresponding grading standard is a minimum circumscribed body; the building structure corresponding to the LOD2 level building model is a differentiated roof and a block, and the corresponding grading standard is the actual size of the building; the building structure corresponding to the LOD3 level building model contains a real roof, a building accessory structure and a building texture, and the corresponding grading standard is that the building model contains external detailed structures; In the step S2, the extraction of the vector surface of the target building comprises the following steps: (1) according to the OSGB data of the building model, the digital surface model (DSM) data and the digital orthographic image (DOM) data are generated; (2) input the digital DSM data and the DOM data; (3) based on the height information contained in the DSM data, the building objects are separated by threshold segmentation; (4) based on the building objects obtained in step (3), the spectral characteristics of vegetation are used to separate them, and the building extraction result is further optimized; (5) all building classification objects are merged; (6) based on the building objects obtained in step (5), non-building objects are removed by threshold segmentation according to the size of the building, and the classification of the building is completed; (7) the building extraction result is manually optimized.

2. The multi-level LOD generation method for real scene three-dimensional building model data according to claim 1, characterized in that: In the step S3, the LOD1 level building monomer model is generated by the vector stretching method, which comprises the following steps: (1) based on the vector data extracted in step S2, the elevation data of the corresponding building in the DSM data is obtained by partitioning statistics; (2) the building elevation information corresponding to the vector surface is given, and the block building is generated.

Images