Component-based three-dimensional model design method and system

Abstract

The invention discloses a modularization-based three-dimensional model design method and system, and relates to the field of modularization-based three-dimensional model design, and the method comprises the steps: building a three-dimensional model design platform based on three-dimensional design software; establishing a space rectangular coordinate system, and dividing the three-dimensional scene into a plurality of cubes; establishing an effective curved surface optimization model, and smoothing effective curved surfaces of scene elements contained in each cube; based on a Laplacian pyramid algorithm, carrying out further optimization on three-dimensional scene modularization multi-angle fusion; establishing a color fusion optimization model, and performing smooth optimization processing on the colors of the three-dimensional scene and the splicing ports of the components; and establishing a three-dimensional scene component fusion optimization model, and obtaining a high-quality and high-integrating-degree component fusion three-dimensional scene model. The method has the advantages that the fusion quality and integrating degree of the three-dimensional scene and the component are effectively improved, and the problems that the three-dimensional scene and the component are incompatible during fusion, and gaps, overlapping or color faults occur are solved.

Description

Technical Field

[0001] The present invention relates to the field of component-based three-dimensional model design, and in particular to a component-based three-dimensional model design method and system. Background Art

[0002] Component-based 3D model design is to decompose complex 3D models into multiple relatively simple components. These components can be designed, modified and tested independently, thereby simplifying the overall design process. Different components can be quickly combined and adjusted according to project requirements to achieve diversified design solutions. By reusing existing components, the design efficiency of 3D models can be effectively improved. By selecting templates, simply dragging components and configuring parameters, a large 3D scene screen can be quickly built and the 3D scene can be quickly released. It is of great significance to greatly improve the design efficiency of 3D models, enhance design flexibility, promote team collaboration, reduce maintenance costs, support standardization and normalization, improve simulation and visualization effects, and promote technological innovation and upgrading.

[0003] The existing component-based 3D model design technology has incomplete construction of databases, model libraries, material coding systems, etc. The geometric shapes, topological structures and sizes of different components may not match completely, and the fusion areas and colors between components may appear unnatural. It is easy for the 3D scene and components to be incompatible when fused, and there may be gaps, overlaps or color faults, resulting in poor 3D model design quality and low fit between the 3D scene and components, which is not conducive to the final rendering effect of the 3D model. Summary of the invention

[0004] In order to solve the above technical problems, a component-based 3D model design method and system are provided. The technical solution solves the problems that the database, model library, material coding system, etc. proposed in the above background technology are still not perfect, the geometric shapes, topological structures and sizes of different components may not match completely, the fusion area components and color fusion between components may appear unnatural, and the 3D scene and components are prone to incompatibility during fusion, gaps, overlaps or color faults, resulting in poor quality of 3D model design, low fit between 3D scenes and components, and unfavorable final rendering effect of the 3D model.

[0005] In order to achieve the above purpose, the technical solution adopted by the present invention is:

[0006] A component-based three-dimensional model design method, comprising:

[0007] Based on big data, we obtain componentized components of 3D scenes, formulate templates for various industries, and build a 3D model design platform based on 3D design software;

[0008] Establish a spatial rectangular coordinate system, perform detailed grid segmentation on the three-dimensional scene, and divide the three-dimensional scene into several cubes;

[0009] According to the effective surfaces of the scene elements contained in each cube, an effective surface optimization model is established to smooth the effective surfaces of the scene elements contained in each cube;

[0010] According to the smoothing results of the effective surface optimization model, the componentized multi-angle fusion of the three-dimensional scene is further optimized based on the Laplace pyramid algorithm;

[0011] According to the color changes of the 3D scene and component fusion, a color fusion optimization model is established to perform smooth optimization processing on the colors of the 3D scene and the interfaces of each component;

[0012] According to the surface optimization results and color optimization results between the three-dimensional scene and components, a three-dimensional scene component fusion optimization model is established based on the neural network algorithm to obtain a high-quality and highly compatible component-fused three-dimensional scene model.

[0013] Preferably, the method of obtaining componentized components of a three-dimensional scene based on big data, formulating various industry templates, and building a three-dimensional model design platform based on three-dimensional design software specifically includes:

[0014] Based on big data, componentized components of 3D scenes are obtained and various industry templates are formulated, among which industry templates serve as the basic templates of 3D scenes;

[0015] Based on 3D design software, a 3D model design platform is built, and the 3D scenes and componentized components are integrated and designed through the platform.

[0016] Preferably, the step of establishing a spatial rectangular coordinate system, performing detailed grid segmentation on the three-dimensional scene, and segmenting the three-dimensional scene into a plurality of cubes specifically includes:

[0017] Based on the 3D model design platform, a spatial rectangular coordinate system is established according to the 3D scene template and components, and the grid segmentation interval of the 3D scene is set;

[0018] The grid segmentation interval distance of the three-dimensional scene is used as the coordinate distance between two adjacent points on each coordinate axis in the spatial rectangular coordinate system;

[0019] According to the spatial rectangular coordinate system, the spatial coordinate information of several cubes after the three-dimensional scene is segmented is obtained.

[0020] Preferably, establishing an effective surface optimization model according to the effective surfaces of the scene elements contained in each cube, and smoothing the effective surfaces of the scene elements contained in each cube specifically comprises:

[0021] According to the spatial coordinate information of each cube, a distribution map of different plane scene elements of each cube is obtained, and a distribution map set of different plane scene elements of the cube is established;

[0022] Grayscale the images in the distribution atlas of different plane scene elements of the cube to obtain the grayscale distribution atlas of different plane scene elements of the cube;

[0023] Through the filtering algorithm, the distribution atlas of different plane scene elements of the grayscale cube is filtered and denoised to optimize the image quality;

[0024] Set the grayscale value of the pixel point that does not contain scene elements to an invalid point value, and obtain the segmentation curve of each face of the cube that contains scene elements according to the distribution of the invalid point values;

[0025] According to the number of segmentation curve vertices of different faces of the cube containing scene elements, the spatial coordinate value of the face point of each face of the cube is obtained;

[0026] According to the spatial coordinate values ​​of the surface points of each face of the cube, obtain the spatial coordinate values ​​of the edge points of each face of the cube;

[0027] According to the spatial coordinate values ​​of the face points and the spatial coordinate values ​​of the edge points of each face of the cube, the updated spatial coordinate values ​​of the vertices of each face of the cube are obtained;

[0028] According to the spatial coordinate values ​​of the face points, the spatial coordinate values ​​of the edge points and the updated spatial coordinate values ​​of the vertices of each face of the cube, a new cube segmented by the segmentation curve is obtained, and the segmented cube containing the scene elements is used as the base point for scene connection;

[0029] Based on the steps and methods of obtaining a new cube segmented by the segmentation curve, an effective surface optimization model is established, and by updating and iterating the steps of obtaining a new cube segmented by the segmentation curve, the effective surface of the scene elements contained in each cube is smoothed to obtain the optimal base point of the scene connection;

[0030] The expression for obtaining the spatial coordinate value of each face point of the cube is:

[0031]

[0032] In the formula, is the spatial coordinate value of the point on each face of the cube, is the number of split curve vertices of the cube containing different faces of scene elements, For a single face of a cube Vertex space coordinate values;

[0033] The expression for obtaining the spatial coordinate value of the edge point of each face of the cube is:

[0034]

[0035] In the formula, is the spatial coordinate value of the edge point of each face of the cube, , Respectively The spatial coordinates of the two endpoints of the edge line, , have the same The spatial coordinate values ​​of the edge points of the two faces of the edge line;

[0036] The expression for obtaining the updated vertex space coordinate value of each face of the cube is:

[0037]

[0038] In the formula, For the cube The updated vertex space coordinate values ​​of the original vertices, For the The spatial coordinate values ​​of the surface points adjacent to the original vertices, For the The original vertices are connected to The spatial coordinates of the edge points of the edge line, is the number of vertices of the splitting curves of different faces of the cube containing scene elements, The number of edges connecting the vertices of the split curves of different faces of the cube containing scene elements.

[0039] Preferably, further optimizing the componentized multi-angle fusion of the three-dimensional scene based on the smoothing result of the effective surface optimization model and the Laplace pyramid algorithm specifically includes:

[0040] According to the smoothing result of the effective surface optimization model, the optimal scene connection base point of each cube is obtained, and the scene connection base point set of the three-dimensional scene template and component is established;

[0041] According to the spatial coordinate system at the connection point between the 3D scene template and the component, the weight parameters of the 3D scene template and the component are set through the platform to determine the adaptive connection between the 3D scene template and the component;

[0042] The adaptive connection method of the three-dimensional scene template and the component specifically includes:

[0043] The weight parameters of the 3D scene template and components are set according to the platform to determine the parts to be eliminated when the 3D scene template and components are fused.

[0044] If the weight parameter of the 3D scene template is set to be greater than the weight parameter of the component, it means that when the 3D scene template and the component are merged, the spatial coordinates of the component connection are adjusted to remove the component parts with overlapping spatial coordinates;

[0045] If the weight parameter of the 3D scene template is set to be smaller than the weight parameter of the component, it means that when the 3D scene template and the component are merged, the spatial coordinates of the connection of the 3D scene template are adjusted to remove the part of the 3D scene template with overlapping spatial coordinates;

[0046] Obtain multi-angle pictures after the 3D scene template and components are integrated, and establish a multi-angle atlas after the 3D scene template and components are integrated;

[0047] According to the multi-angle atlas after the fusion of 3D scene templates and components, the componentized multi-angle fusion of 3D scenes is further optimized based on the Laplace pyramid algorithm.

[0048] Preferably, the step of establishing a color fusion optimization model according to the color changes of the fusion of the three-dimensional scene and the components, and performing smooth optimization processing on the colors of the three-dimensional scene and the interface of each component specifically includes:

[0049] Obtain multi-angle images after component-based multi-angle fusion optimization of the 3D scene, and establish a component-based multi-angle fusion optimization atlas of the 3D scene;

[0050] According to the three-dimensional scene componentized multi-angle fusion optimization atlas, the source color value of the three-dimensional scene componentized multi-angle fusion area is obtained, and the source color value input matrix of the three-dimensional scene componentized multi-angle fusion area is established;

[0051] Based on big data, the target color value of the componentized multi-angle fusion area of ​​the 3D scene is obtained, and the standard input matrix of the target color value of the componentized multi-angle fusion area of ​​the 3D scene is established;

[0052] According to the source color value input matrix and the target color value standard input matrix, obtain the deviation value between the fused color value and the source color value and the target color value;

[0053] According to the source color value input matrix and the target color value standard input matrix, obtain the smoothed value of the fused color value and the source color value and the target color value;

[0054] According to the deviation value and smoothing value of the fused color value and the source color value and the target color value, a loss function of the fused color value and the source color value and the target color value is established;

[0055] Based on the gradient descent algorithm, determine the update function of the color value after the fusion of the componentized multi-angle fusion area of ​​the three-dimensional scene;

[0056] Based on the above steps, a color fusion optimization model is established to perform smooth optimization processing on the colors of the three-dimensional scene and the interfaces of each component;

[0057] The expression for obtaining the deviation value between the fused color value and the source color value and the target color value is:

[0058]

[0059] In the formula, is the deviation value between the fused color value and the source color value and the target color value. is the color value set of the fusion area, For pixels The blended color value at the point, For pixels The source color value at the point, For pixels The target color value at the point;

[0060] The expression for obtaining the smoothing value of the fused color value, the source color value, and the target color value is:

[0061]

[0062] In the formula, is the smoothing value of the fused color value with the source color value and the target color value. Pixel Point neighborhood, For pixels The blended color value at the point;

[0063] The loss function expression of the fused color value, the source color value and the target color value is:

[0064]

[0065] In the formula, is the loss value of the fused color value with the source color value and the target color value, is the smoothing weight balance value;

[0066] The gradient descent algorithm expression is:

[0067]

[0068] In the formula, For pixels The updated value of the fused color value at the point, is the learning rate.

[0069] Preferably, the process of establishing a 3D scene component fusion optimization model based on a neural network algorithm according to the surface optimization results and color optimization results between the 3D scene and the components to obtain a high-quality, high-compatibility component-fused 3D scene model specifically includes:

[0070] Based on big data, the expected data of 3D scene componentization and multi-angle fusion is obtained as target data, and the output target matrix of the neural network is established;

[0071] According to the surface optimization results and color optimization results between the three-dimensional scene and the components, the optimization results are used as the input matrix of the neural network;

[0072] Normalize the input matrix and output target matrix data of the neural network to eliminate the dimension effect of the data;

[0073] Based on the neural network algorithm, a 3D scene component fusion optimization model is established, and the 3D scene model is comprehensively adjusted to obtain a high-quality and highly compatible component-based fusion 3D scene model.

[0074] Furthermore, the present solution proposes a component-based 3D model design system for implementing the component-based 3D model design method as described above, including:

[0075] A platform building module, which is used to obtain componentized components of a three-dimensional scene based on big data, formulate multiple industry templates, and build a three-dimensional model design platform based on three-dimensional design software;

[0076] A surface fusion optimization module, which is used to establish a spatial rectangular coordinate system, perform detailed grid segmentation on the three-dimensional scene, and divide the three-dimensional scene into a number of cubes; according to the effective surfaces of the scene elements contained in each cube, an effective surface optimization model is established, and the effective surfaces of the scene elements contained in each cube are smoothed; according to the smoothing result of the effective surface optimization model, based on the Laplace pyramid algorithm, the componentized multi-angle fusion of the three-dimensional scene is further optimized;

[0077] A color fusion optimization module, which is used to establish a color fusion optimization model according to the color changes of the three-dimensional scene and the component fusion, and to perform smooth optimization processing on the colors of the three-dimensional scene and the interface of each component;

[0078] The comprehensive optimization module is used to establish a three-dimensional scene component fusion optimization model based on the surface optimization results and color optimization results between the three-dimensional scene and the components, based on the neural network algorithm, to obtain a high-quality, high-fit component-fused three-dimensional scene model.

[0079] Preferably, the surface fusion optimization module includes:

[0080] A scene segmentation unit, the scene segmentation unit is used to establish a spatial rectangular coordinate system, perform detailed grid segmentation on the three-dimensional scene, and divide the three-dimensional scene into a plurality of cubes;

[0081] A surface optimization unit, wherein the surface optimization unit is used to establish an effective surface optimization model according to the effective surfaces of the scene elements contained in each cube, and to smooth the effective surfaces of the scene elements contained in each cube;

[0082] The surface fusion optimization unit is used to further optimize the componentized multi-angle fusion of the three-dimensional scene based on the smoothing result of the effective surface optimization model and the Laplace pyramid algorithm.

[0083] Compared with the prior art, the present invention has the following beneficial effects:

[0084] By establishing a spatial rectangular coordinate system, setting the grid segmentation interval of the three-dimensional scene, and taking the grid segmentation interval distance of the three-dimensional scene as the coordinate distance between two adjacent points of each coordinate axis in the spatial rectangular coordinate system, the spatial coordinate information of several cubes after the three-dimensional scene is segmented is obtained. Secondly, according to the spatial coordinate information of each cube, by obtaining the surface point spatial coordinate values, edge point spatial coordinate values ​​and updated vertex spatial coordinate values ​​of the effective surface of the scene elements contained in each cube, a new cube segmented by the segmentation curve is obtained, and an effective surface optimization model is established. By updating and iterating the steps of obtaining the new cube segmented by the segmentation curve, the effective surface of the scene elements contained in each cube is smoothed, and the segmented cube containing the scene elements is used as the base point for scene connection, so as to obtain the optimal base point for scene connection, and according to the effective curve The smoothing result of the surface optimization model is further optimized for the componentized multi-angle fusion of the three-dimensional scene based on the Laplace pyramid algorithm. Furthermore, according to the color changes of the fusion of the three-dimensional scene and the components, by analyzing the loss function relationship between the fusion color value and the source color value and the target color value, a color fusion optimization model is established based on the gradient descent algorithm to smooth and optimize the colors of the three-dimensional scene and the interfaces of each component. Finally, according to the surface optimization results and color optimization results between the three-dimensional scene and the components, a three-dimensional scene component fusion optimization model is established based on the neural network algorithm, and the three-dimensional scene model is comprehensively adjusted to obtain a high-quality and high-fit componentized fusion three-dimensional scene model, thereby effectively improving the quality and fit of the fusion of the three-dimensional scene and the components, and reducing the incompatibility between the three-dimensional scene and the components during fusion, and the problems of gaps, overlaps or color faults. BRIEF DESCRIPTION OF THE DRAWINGS

[0085] Figure 1 A flow chart of a component-based three-dimensional model design method of the present invention;

[0086] Figure 2 The present invention is a flow chart of establishing an effective surface optimization model according to the effective surfaces of the scene elements contained in each cube, and performing smoothing processing on the effective surfaces of the scene elements contained in each cube;

[0087] Figure 3 This is a flow chart of further optimizing the componentized multi-angle fusion of three-dimensional scenes based on the smoothing result of the effective surface optimization model of the present invention based on the Laplace pyramid algorithm;

[0088] Figure 4 The present invention establishes a color fusion optimization model according to the color changes of the three-dimensional scene and the component fusion, and performs a smooth optimization processing flow chart on the colors of the three-dimensional scene and the splicing interfaces of each component. DETAILED DESCRIPTION

[0089] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are only examples, and those skilled in the art may think of other obvious variations.

[0090] Reference Figure 1 As shown, a component-based three-dimensional model design method includes:

[0091] Based on big data, we obtain componentized components of 3D scenes, formulate templates for various industries, and build a 3D model design platform based on 3D design software;

[0092] Establish a spatial rectangular coordinate system, perform detailed grid segmentation on the three-dimensional scene, and divide the three-dimensional scene into several cubes;

[0093] According to the effective surfaces of the scene elements contained in each cube, an effective surface optimization model is established to smooth the effective surfaces of the scene elements contained in each cube;

[0094] According to the smoothing results of the effective surface optimization model, the componentized multi-angle fusion of the three-dimensional scene is further optimized based on the Laplace pyramid algorithm;

[0095] According to the color changes of the 3D scene and component fusion, a color fusion optimization model is established to perform smooth optimization processing on the colors of the 3D scene and the interfaces of each component;

[0096] According to the surface optimization results and color optimization results between the three-dimensional scene and components, a three-dimensional scene component fusion optimization model is established based on the neural network algorithm to obtain a high-quality and highly compatible component-fused three-dimensional scene model.

[0097] It can be explained that this scheme establishes a spatial rectangular coordinate system, sets the grid segmentation interval of the three-dimensional scene, and uses the grid segmentation interval distance of the three-dimensional scene as the coordinate distance between two adjacent points of each coordinate axis in the spatial rectangular coordinate system to obtain the spatial coordinate information of several cubes after the three-dimensional scene is segmented. Secondly, according to the spatial coordinate information of each cube, by obtaining the surface point spatial coordinate values, edge point spatial coordinate values ​​and updated vertex spatial coordinate values ​​of the effective surface of the scene elements contained in each cube, a new cube segmented by the segmentation curve is obtained, and an effective surface optimization model is established. By updating and iterating the steps of obtaining the new cube segmented by the segmentation curve, the effective surface of the scene elements contained in each cube is smoothed, and the optimal base point for scene connection is obtained by using the segmented cube containing the scene elements as the base point for scene connection, and According to the smoothing results of the effective surface optimization model, the componentized multi-angle fusion of the three-dimensional scene is further optimized based on the Laplace pyramid algorithm. Furthermore, according to the color changes of the fusion of the three-dimensional scene and the components, by analyzing the loss function relationship between the fusion color value and the source color value and the target color value, a color fusion optimization model is established based on the gradient descent algorithm, and the colors of the three-dimensional scene and the interfaces of each component are smoothed and optimized. Finally, according to the surface optimization results and color optimization results between the three-dimensional scene and the components, a three-dimensional scene component fusion optimization model is established based on the neural network algorithm, and the three-dimensional scene model is comprehensively adjusted to obtain a high-quality and high-fit componentized fusion three-dimensional scene model, thereby effectively improving the quality and fit of the fusion of the three-dimensional scene and the components, and reducing the incompatibility of the three-dimensional scene and the components during fusion, and the problems of gaps, overlaps or color faults.

[0098] Reference Figure 2 As shown, the effective surface optimization model is established according to the effective surface of the scene element contained in each cube, and the effective surface of the scene element contained in each cube is smoothed, which specifically includes:

[0099] According to the spatial coordinate information of each cube, a distribution map of different plane scene elements of each cube is obtained, and a distribution map set of different plane scene elements of the cube is established;

[0100] Grayscale the images in the distribution atlas of different plane scene elements of the cube to obtain the grayscale distribution atlas of different plane scene elements of the cube;

[0101] Through the filtering algorithm, the distribution atlas of different plane scene elements of the grayscale cube is filtered and denoised to optimize the image quality;

[0102] Set the grayscale value of the pixel point that does not contain scene elements to an invalid point value, and obtain the segmentation curve of each face of the cube that contains scene elements according to the distribution of the invalid point values;

[0103] According to the number of segmentation curve vertices of different faces of the cube containing scene elements, the spatial coordinate value of the face point of each face of the cube is obtained;

[0104] According to the spatial coordinate values ​​of the surface points of each face of the cube, obtain the spatial coordinate values ​​of the edge points of each face of the cube;

[0105] According to the spatial coordinate values ​​of the face points and the spatial coordinate values ​​of the edge points of each face of the cube, the updated spatial coordinate values ​​of the vertices of each face of the cube are obtained;

[0106] According to the spatial coordinate values ​​of the face points, the spatial coordinate values ​​of the edge points and the updated spatial coordinate values ​​of the vertices of each face of the cube, a new cube segmented by the segmentation curve is obtained, and the segmented cube containing the scene elements is used as the base point for scene connection;

[0107] Based on the steps and methods of obtaining a new cube segmented by the segmentation curve, an effective surface optimization model is established, and by updating and iterating the steps of obtaining a new cube segmented by the segmentation curve, the effective surface of the scene elements contained in each cube is smoothed to obtain the optimal base point of the scene connection;

[0108] The expression for obtaining the spatial coordinate value of each face point of the cube is:

[0109]

[0110] In the formula, is the spatial coordinate value of the point on each face of the cube, is the number of vertices of the segmentation curves of different faces of the cube containing scene elements, h For a single face of a cube Vertex space coordinate values;

[0111] The expression for obtaining the spatial coordinate value of the edge point of each face of the cube is:

[0112]

[0113] In the formula, is the spatial coordinate value of the edge point of each face of the cube, , Respectively The spatial coordinates of the two endpoints of the edge line, , have the same The spatial coordinate values ​​of the edge points of the two faces of the edge line;

[0114] The expression for obtaining the updated vertex space coordinate value of each face of the cube is:

[0115]

[0116] In the formula, For the cube The updated vertex space coordinate values ​​of the original vertices, For the The spatial coordinate values ​​of the surface points adjacent to the original vertices, For the The original vertices are connected to The spatial coordinates of the edge points of the edge line, is the number of split curve vertices of the cube containing different faces of scene elements, The number of edges connecting the vertices of the split curves of different faces of the cube containing scene elements.

[0117] It can be explained that when the three-dimensional scene and the component are merged, the geometric shapes, topological structures and sizes of the three-dimensional scene and different components may not completely match, which leads to incompatibility, gaps and overlaps between the three-dimensional scene and the components when they are merged, thus affecting the design quality of the three-dimensional model. This solution establishes a spatial rectangular coordinate system, sets the grid segmentation interval of the three-dimensional scene, and uses the grid segmentation interval distance of the three-dimensional scene as the coordinate distance between two adjacent points of each coordinate axis in the spatial rectangular coordinate system to obtain the spatial coordinate information of several cubes after the three-dimensional scene is segmented. According to the spatial coordinate information of each cube, the surface point spatial coordinate values, edge point spatial coordinate values ​​and updated vertex spatial coordinate values ​​of the effective surface of the scene elements contained in each cube are obtained to obtain a new cube segmented by the segmentation curve, establish an effective surface optimization model, and smooth the effective surface of the scene elements contained in each cube by updating and iterating the steps of obtaining the new cube segmented by the segmentation curve. The optimal base point for scene connection is obtained by using the segmented cube containing the scene elements as the base point for scene connection.

[0118] Reference Figure 3 As shown, the smoothing result of the effective surface optimization model is further optimized based on the Laplace pyramid algorithm for componentized multi-angle fusion of three-dimensional scenes, specifically including:

[0119] According to the smoothing result of the effective surface optimization model, the optimal scene connection base point of each cube is obtained, and the scene connection base point set of the three-dimensional scene template and components is established;

[0120] According to the spatial coordinate system at the connection point between the 3D scene template and the component, the weight parameters of the 3D scene template and the component are set through the platform to determine the adaptive connection between the 3D scene template and the component;

[0121] The adaptive connection method of the three-dimensional scene template and the component specifically includes:

[0122] The weight parameters of the 3D scene template and components are set according to the platform to determine the parts to be eliminated when the 3D scene template and components are fused.

[0123] If the weight parameter of the 3D scene template is set to be greater than the weight parameter of the component, it means that when the 3D scene template and the component are merged, the spatial coordinates of the component connection are adjusted to remove the component parts with overlapping spatial coordinates;

[0124] If the weight parameter of the 3D scene template is set to be smaller than the weight parameter of the component, it means that when the 3D scene template and the component are merged, the spatial coordinates of the connection of the 3D scene template are adjusted to remove the part of the 3D scene template with overlapping spatial coordinates;

[0125] Obtain multi-angle pictures after the 3D scene template and components are integrated, and establish a multi-angle atlas after the 3D scene template and components are integrated;

[0126] According to the multi-angle atlas after the fusion of 3D scene templates and components, the componentized multi-angle fusion of 3D scenes is further optimized based on the Laplace pyramid algorithm.

[0127] It can be explained that, according to the effective surface optimization model, the spatial coordinates of the base point of the optimal scene connection can be effectively obtained, and the weight parameters of the three-dimensional scene template and the component can be set through the platform to determine the adaptive connection of the three-dimensional scene template and the component, thereby judging whether the weight parameter of the three-dimensional scene template is greater than the weight parameter of the component. If so, it means that when the three-dimensional scene template and the component are fused, the spatial coordinates of the component connection are adjusted to eliminate the component part with overlapping spatial coordinates. If not, it means that when the three-dimensional scene template and the component are fused, the spatial coordinates of the three-dimensional scene template connection are adjusted to eliminate the three-dimensional scene template part with overlapping spatial coordinates, thereby effectively improving the fit of the three-dimensional scene template and the component fusion. In order to further improve the multi-angle smoothness when the three-dimensional scene template and the component are fused and reduce the influence of singular points in the fusion area, this solution further optimizes the multi-angle fusion of three-dimensional scene components through the Laplace pyramid algorithm.

[0128] The specific steps of the Laplace pyramid algorithm are as follows:

[0129] According to the multi-angle atlas after the fusion of the 3D scene template and components, the maximum image and the minimum image are separated to construct an image pyramid model;

[0130] Perform Gaussian blur on the image of the next layer of the image pyramid model, and delete the even rows and columns of the blurred image, and repeat this process to obtain a Gaussian pyramid model;

[0131] According to the Gaussian pyramid model and based on the Laplace pyramid algorithm, the replaced and reconstructed 3D scene templates and the multi-angle atlas after component fusion are optimized to further improve the quality of componentized multi-angle fusion of 3D scenes.

[0132] Reference Figure 4 As shown, the color fusion optimization model is established according to the color changes of the 3D scene and the component fusion, and the color of the 3D scene and the interface of each component is smoothly optimized, which specifically includes:

[0133] Obtain multi-angle images after component-based multi-angle fusion optimization of the 3D scene, and establish a component-based multi-angle fusion optimization atlas of the 3D scene;

[0134] According to the three-dimensional scene componentized multi-angle fusion optimization atlas, the source color value of the three-dimensional scene componentized multi-angle fusion area is obtained, and the source color value input matrix of the three-dimensional scene componentized multi-angle fusion area is established;

[0135] Based on big data, the target color value of the componentized multi-angle fusion area of ​​the 3D scene is obtained, and the standard input matrix of the target color value of the componentized multi-angle fusion area of ​​the 3D scene is established;

[0136] According to the source color value input matrix and the target color value standard input matrix, obtain the deviation value between the fused color value and the source color value and the target color value;

[0137] According to the source color value input matrix and the target color value standard input matrix, obtain the smoothed value of the fused color value and the source color value and the target color value;

[0138] According to the deviation value and smoothing value of the fused color value and the source color value and the target color value, a loss function of the fused color value and the source color value and the target color value is established;

[0139] Based on the gradient descent algorithm, determine the update function of the color value after the fusion of the componentized multi-angle fusion area of ​​the three-dimensional scene;

[0140] Based on the above steps, a color fusion optimization model is established to perform smooth optimization processing on the colors of the three-dimensional scene and the interfaces of each component;

[0141] The expression for obtaining the deviation value between the fused color value and the source color value and the target color value is:

[0142]

[0143] In the formula, is the deviation value between the fused color value and the source color value and the target color value. is the color value set of the fusion area, For pixels The blended color value at the point, For pixels The source color value at the point, For pixels The target color value at the point;

[0144] The expression for obtaining the smoothing value of the fused color value, the source color value, and the target color value is:

[0145]

[0146] In the formula, is the smoothing value of the fused color value with the source color value and the target color value. Pixel Point neighborhood, For pixels The blended color value at the point;

[0147] The loss function expression of the fused color value, the source color value and the target color value is:

[0148]

[0149] In the formula, is the loss value of the fused color value with the source color value and the target color value, is the smoothing weight balance value;

[0150] The gradient descent algorithm expression is:

[0151]

[0152] In the formula, For pixels The updated value of the fused color value at the point, is the learning rate.

[0153] It can be explained that when fusing three-dimensional scenes and components, the color fusion match between different components and three-dimensional scenes is also particularly important. The color difference of the fusion area can be easily and intuitively felt through observation. Therefore, improving the color fusion when fusing three-dimensional scenes and components plays a very important role in improving three-dimensional model design. This scheme establishes a loss function relationship between the fusion color value and the source color value and the target color value by obtaining the fusion color value of the fusion area according to the color changes of the three-dimensional scene and component fusion, and establishes a color fusion optimization model based on the gradient descent algorithm. By reducing the loss value of the fusion color value and the source color value and the target color value, the color of the three-dimensional scene and the interface of each component can be smoothed and optimized, thereby improving the color fusion quality when the three-dimensional scene and the component are merged.

[0154] Further, based on the same inventive concept as the above-mentioned component-based 3D model design method, this solution proposes a component-based 3D model design system, including:

[0155] A platform building module, which is used to obtain componentized components of a three-dimensional scene based on big data, formulate multiple industry templates, and build a three-dimensional model design platform based on three-dimensional design software;

[0156] A surface fusion optimization module, which is used to establish a spatial rectangular coordinate system, perform detailed grid segmentation on the three-dimensional scene, and divide the three-dimensional scene into a number of cubes; according to the effective surfaces of the scene elements contained in each cube, an effective surface optimization model is established, and the effective surfaces of the scene elements contained in each cube are smoothed; according to the smoothing result of the effective surface optimization model, based on the Laplace pyramid algorithm, the componentized multi-angle fusion of the three-dimensional scene is further optimized;

[0157] A color fusion optimization module, which is used to establish a color fusion optimization model according to the color changes of the three-dimensional scene and the component fusion, and to perform smooth optimization processing on the colors of the three-dimensional scene and the interface of each component;

[0158] A comprehensive optimization module, which is used to establish a 3D scene component fusion optimization model based on a neural network algorithm according to the surface optimization results and color optimization results between the 3D scene and the components, and obtain a high-quality, high-fit component-fused 3D scene model;

[0159] The surface fusion optimization module includes:

[0160] A scene segmentation unit, the scene segmentation unit is used to establish a spatial rectangular coordinate system, perform detailed grid segmentation on the three-dimensional scene, and divide the three-dimensional scene into a plurality of cubes;

[0161] A surface optimization unit, wherein the surface optimization unit is used to establish an effective surface optimization model according to the effective surfaces of the scene elements contained in each cube, and to smooth the effective surfaces of the scene elements contained in each cube;

[0162] The surface fusion optimization unit is used to further optimize the componentized multi-angle fusion of the three-dimensional scene based on the smoothing result of the effective surface optimization model and the Laplace pyramid algorithm.

[0163] In summary, the advantages of the present invention are: effectively improving the quality and compatibility of the fusion of the three-dimensional scene and the components, and reducing the incompatibility between the three-dimensional scene and the components during fusion, and reducing the problems of gaps, overlaps or color faults.

[0164] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions only describe the principles of the present invention. The present invention may be subject to various changes and improvements without departing from the spirit and scope of the present invention. These changes and improvements fall within the scope of the present invention. The scope of protection claimed by the present invention is defined by the attached claims and their equivalents.

Claims

1. A component-based three-dimensional model design method, characterized in that: include: Based on big data, we obtain componentized components of 3D scenes, formulate templates for various industries, and build a 3D model design platform based on 3D design software; Establish a spatial rectangular coordinate system, perform detailed grid segmentation on the three-dimensional scene, and divide the three-dimensional scene into several cubes; According to the effective surfaces of the scene elements contained in each cube, an effective surface optimization model is established to smooth the effective surfaces of the scene elements contained in each cube; According to the smoothing results of the effective surface optimization model, the componentized multi-angle fusion of the three-dimensional scene is further optimized based on the Laplace pyramid algorithm; According to the color changes of the 3D scene and component fusion, a color fusion optimization model is established to perform smooth optimization processing on the colors of the 3D scene and the interfaces of each component; According to the surface optimization results and color optimization results between the three-dimensional scene and components, a three-dimensional scene component fusion optimization model is established based on the neural network algorithm to obtain a high-quality and highly compatible component-fused three-dimensional scene model.

2. A component-based 3D model design method according to claim 1, characterized in that: The method of obtaining componentized components of 3D scenes based on big data, formulating various industry templates, and building a 3D model design platform based on 3D design software specifically includes: Based on big data, componentized components of 3D scenes are obtained and various industry templates are formulated, among which industry templates serve as the basic templates of 3D scenes; Based on 3D design software, a 3D model design platform is built, and the 3D scenes and componentized components are integrated and designed through the platform.

3. A component-based 3D model design method according to claim 2, characterized in that: The step of establishing a spatial rectangular coordinate system, performing detailed grid segmentation on the three-dimensional scene, and segmenting the three-dimensional scene into a plurality of cubes specifically includes: Based on the 3D model design platform, a spatial rectangular coordinate system is established according to the 3D scene template and components, and the grid segmentation interval of the 3D scene is set; The grid segmentation interval distance of the three-dimensional scene is used as the coordinate distance between two adjacent points on each coordinate axis in the spatial rectangular coordinate system; According to the spatial rectangular coordinate system, the spatial coordinate information of several cubes after the three-dimensional scene is segmented is obtained.

4. A component-based 3D model design method according to claim 3, characterized in that: The step of establishing an effective surface optimization model according to the effective surfaces of the scene elements contained in each cube and smoothing the effective surfaces of the scene elements contained in each cube specifically includes: According to the spatial coordinate information of each cube, a distribution map of different plane scene elements of each cube is obtained, and a distribution map set of different plane scene elements of the cube is established; Grayscale the images in the distribution atlas of different plane scene elements of the cube to obtain the grayscale distribution atlas of different plane scene elements of the cube; Through the filtering algorithm, the distribution atlas of different plane scene elements of the grayscale cube is filtered and denoised to optimize the image quality; Set the grayscale value of the pixel point that does not contain scene elements to an invalid point value, and obtain the segmentation curve of each face of the cube that contains scene elements according to the distribution of the invalid point values; According to the number of segmentation curve vertices of different faces of the cube containing scene elements, the spatial coordinate value of the face point of each face of the cube is obtained; According to the spatial coordinate values ​​of the surface points of each face of the cube, obtain the spatial coordinate values ​​of the edge points of each face of the cube; According to the spatial coordinate values ​​of the face points and the spatial coordinate values ​​of the edge points of each face of the cube, the updated spatial coordinate values ​​of the vertices of each face of the cube are obtained; According to the spatial coordinate values ​​of the face points, the spatial coordinate values ​​of the edge points and the updated spatial coordinate values ​​of the vertices of each face of the cube, a new cube segmented by the segmentation curve is obtained, and the segmented cube containing the scene elements is used as the base point for scene connection; Based on the steps and methods of obtaining a new cube segmented by the segmentation curve, an effective surface optimization model is established, and by updating and iterating the steps of obtaining a new cube segmented by the segmentation curve, the effective surface of the scene elements contained in each cube is smoothed to obtain the optimal base point of the scene connection; The expression for obtaining the spatial coordinate value of each face point of the cube is: In the formula, is the spatial coordinate value of the point on each face of the cube, is the number of vertices of the segmentation curves of different faces of the cube containing scene elements, h For a single face of a cube Vertex space coordinate values; The expression for obtaining the spatial coordinate value of the edge point of each face of the cube is: In the formula, is the spatial coordinate value of the edge point of each face of the cube, , Respectively The spatial coordinate values ​​of the two endpoints of the edge line, , have the same The spatial coordinate values ​​of the edge points of the two faces of the edge line; The expression for obtaining the updated vertex space coordinate value of each face of the cube is: In the formula, For the cube The updated vertex space coordinate values ​​of the original vertices, For the The spatial coordinate values ​​of the surface points adjacent to the original vertices, For the The original vertices are connected to The spatial coordinates of the edge points of the edge line, is the number of split curve vertices of the cube containing different faces of scene elements, The number of edges connecting the vertices of the split curves of different faces of the cube containing scene elements.

5. A component-based 3D model design method according to claim 4, characterized in that: The further optimization of the componentized multi-angle fusion of the three-dimensional scene based on the smoothing result of the effective surface optimization model and the Laplace pyramid algorithm specifically includes: According to the smoothing result of the effective surface optimization model, the optimal scene connection base point of each cube is obtained, and the scene connection base point set of the three-dimensional scene template and components is established; According to the spatial coordinate system at the connection point between the 3D scene template and the component, the weight parameters of the 3D scene template and the component are set through the platform to determine the adaptive connection between the 3D scene template and the component; The adaptive connection method of the three-dimensional scene template and the component specifically includes: The weight parameters of the 3D scene template and components are set according to the platform to determine the parts to be eliminated when the 3D scene template and components are fused. If the weight parameter of the 3D scene template is set to be greater than the weight parameter of the component, it means that when the 3D scene template and the component are merged, the spatial coordinates of the component connection are adjusted to remove the component parts with overlapping spatial coordinates; If the weight parameter of the 3D scene template is set to be smaller than the weight parameter of the component, it means that when the 3D scene template and the component are merged, the spatial coordinates of the connection of the 3D scene template are adjusted to remove the part of the 3D scene template with overlapping spatial coordinates; Obtain multi-angle pictures after the 3D scene template and components are integrated, and establish a multi-angle atlas after the 3D scene template and components are integrated; According to the multi-angle atlas after the fusion of 3D scene templates and components, the componentized multi-angle fusion of 3D scenes is further optimized based on the Laplace pyramid algorithm.

6. A component-based 3D model design method according to claim 5, characterized in that: The method of establishing a color fusion optimization model according to the color changes of the 3D scene and the component fusion, and performing smooth optimization processing on the colors of the 3D scene and the interface of each component specifically includes: Obtain multi-angle images after component-based multi-angle fusion optimization of the three-dimensional scene, and establish a component-based multi-angle fusion optimization atlas of the three-dimensional scene; According to the three-dimensional scene componentized multi-angle fusion optimization atlas, the source color value of the three-dimensional scene componentized multi-angle fusion area is obtained, and the source color value input matrix of the three-dimensional scene componentized multi-angle fusion area is established; Based on big data, the target color value of the componentized multi-angle fusion area of ​​the 3D scene is obtained, and the standard input matrix of the target color value of the componentized multi-angle fusion area of ​​the 3D scene is established; According to the source color value input matrix and the target color value standard input matrix, obtain the deviation value between the fused color value and the source color value and the target color value; According to the source color value input matrix and the target color value standard input matrix, obtain the smoothed value of the fused color value and the source color value and the target color value; According to the deviation value and smoothing value of the fused color value and the source color value and the target color value, a loss function of the fused color value and the source color value and the target color value is established; Based on the gradient descent algorithm, determine the update function of the color value after the fusion of the componentized multi-angle fusion area of ​​the three-dimensional scene; Based on the above steps, a color fusion optimization model is established to perform smooth optimization processing on the colors of the three-dimensional scene and the interfaces of each component; The expression for obtaining the deviation value between the fused color value and the source color value and the target color value is: In the formula, is the deviation value between the fused color value and the source color value and the target color value. is the color value set of the fusion area, For pixels The blended color value at the point, For pixels The source color value at the point, For pixels The target color value at the point; The expression for obtaining the smoothing value of the fused color value, the source color value, and the target color value is: In the formula, is the smoothing value of the fused color value with the source color value and the target color value. Pixel Point neighborhood, For pixels The blended color value at the point; The loss function expression of the fused color value, the source color value and the target color value is: In the formula, is the loss value of the fused color value with the source color value and the target color value, is the smoothing weight balance value; The gradient descent algorithm expression is: In the formula, For pixels The updated value of the fused color value at the point, is the learning rate.

7. A component-based 3D model design method according to claim 6, characterized in that: The method of establishing a 3D scene component fusion optimization model based on a neural network algorithm according to the surface optimization results and color optimization results between the 3D scene and the components to obtain a high-quality, high-fit component-fused 3D scene model specifically includes: Based on big data, the expected data of 3D scene componentization and multi-angle fusion is obtained as target data, and the output target matrix of the neural network is established; According to the surface optimization results and color optimization results between the three-dimensional scene and the components, the optimization results are used as the input matrix of the neural network; Normalize the input matrix and output target matrix data of the neural network to eliminate the dimension effect of the data; Based on the neural network algorithm, a 3D scene component fusion optimization model is established, and the 3D scene model is comprehensively adjusted to obtain a high-quality and highly compatible component-based fusion 3D scene model.

8. A component-based three-dimensional model design system, characterized in that: A component-based three-dimensional model design method for implementing any one of claims 1 to 7, comprising: A platform building module, which is used to obtain componentized components of a three-dimensional scene based on big data, formulate multiple industry templates, and build a three-dimensional model design platform based on three-dimensional design software; A surface fusion optimization module, which is used to establish a spatial rectangular coordinate system, perform detailed grid segmentation on the three-dimensional scene, and divide the three-dimensional scene into a number of cubes; according to the effective surfaces of the scene elements contained in each cube, an effective surface optimization model is established, and the effective surfaces of the scene elements contained in each cube are smoothed; according to the smoothing result of the effective surface optimization model, based on the Laplace pyramid algorithm, the componentized multi-angle fusion of the three-dimensional scene is further optimized; A color fusion optimization module, which is used to establish a color fusion optimization model according to the color changes of the three-dimensional scene and the component fusion, and to perform smooth optimization processing on the colors of the three-dimensional scene and the interface of each component; The comprehensive optimization module is used to establish a three-dimensional scene component fusion optimization model based on the surface optimization results and color optimization results between the three-dimensional scene and the components, based on the neural network algorithm, to obtain a high-quality, high-fit component-fused three-dimensional scene model.

9. The component-based three-dimensional model design system according to claim 8, characterized in that: The surface fusion optimization module includes: A scene segmentation unit, the scene segmentation unit is used to establish a spatial rectangular coordinate system, perform detailed grid segmentation on the three-dimensional scene, and divide the three-dimensional scene into a plurality of cubes; A surface optimization unit, wherein the surface optimization unit is used to establish an effective surface optimization model according to the effective surfaces of the scene elements contained in each cube, and to smooth the effective surfaces of the scene elements contained in each cube; The surface fusion optimization unit is used to further optimize the componentized multi-angle fusion of the three-dimensional scene based on the smoothing result of the effective surface optimization model and the Laplace pyramid algorithm.

Images