Component-based 3D model design method and system

Abstract

The present invention discloses a component-based three-dimensional model design method and system, which relates to the field of component-based three-dimensional model design and includes: building a three-dimensional model design platform based on three-dimensional design software; establishing a spatial rectangular coordinate system and dividing the three-dimensional scene into several cubes; establishing an effective surface optimization model to smooth the effective surfaces of the scene elements contained in each cube; further optimizing the multi-angle fusion of the three-dimensional scene components based on the Laplacian pyramid algorithm; establishing a color fusion optimization model to smoothly optimize the colors of the three-dimensional scene and the splicing interfaces of each component; and establishing a three-dimensional scene component fusion optimization model to obtain a three-dimensional scene model with high quality and high compatibility in component fusion. The advantages of the present invention are: effectively improving the quality and compatibility of the fusion between the three-dimensional scene and the components, and reducing the problems of incompatibility, gaps, overlaps or color breaks when the three-dimensional scene and the components are fused.

Description

Technical Field

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

[0002] Component-based three-dimensional model design is to decompose a complex three-dimensional model into multiple relatively simple components. These components can be designed, modified, and tested independently, thus simplifying the overall design process. And according to project requirements, different components can be quickly combined and adjusted to achieve diverse design schemes. By reusing existing components, the design efficiency of three-dimensional models can be effectively improved. Thus, by selecting templates, simply dragging components and performing parameter configuration, a three-dimensional scene large screen can be quickly constructed, and the rapid release of three-dimensional scenes can be completed, which is of great significance in greatly improving the design efficiency of three-dimensional models, enhancing design flexibility, promoting teamwork, reducing maintenance costs, supporting standardization and normalization, improving simulation and visualization effects, and promoting technological innovation and upgrading.

[0003] In the existing component-based three-dimensional model design technology, due to the imperfect construction of databases, model libraries, material coding systems, etc., the geometric shapes, topological structures, and sizes of different components may not be completely matched, and the component and color fusion in the fusion area between components may appear unnatural. It is easy to have problems such as incompatibility between the three-dimensional scene and components during fusion, such as gaps, overlaps, or color breaks, resulting in poor quality of three-dimensional model design, low compatibility between the three-dimensional scene and components, and being unfavorable to the final rendering effect of three-dimensional models. Summary of the Invention

[0004] To solve the above technical problems, a method and system for component-based three-dimensional model design are provided. This technical solution solves the problems proposed in the above background art, that is, the construction of databases, model libraries, material coding systems, etc. is not yet perfect, the geometric shapes, topological structures, and sizes of different components may not be completely matched, the component and color fusion in the fusion area between components may appear unnatural, it is easy to have problems such as incompatibility between the three-dimensional scene and components during fusion, such as gaps, overlaps, or color breaks, resulting in poor quality of three-dimensional model design, low compatibility between the three-dimensional scene and components, and being unfavorable to the final rendering effect of three-dimensional models.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

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

[0007] Based on big data, obtain componentized components of a three-dimensional scene, formulate multiple industry templates, and build a three-dimensional model design platform based on three-dimensional design software;

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

[0009] Based on the effective surfaces of the scene elements contained in each cube, establish an effective surface optimization model, and smooth the effective surfaces of the scene elements contained in each cube;

[0010] Based on the smoothing results of the effective surface optimization model, further optimize the multi-angle fusion of the three-dimensional scene components based on the Laplacian pyramid algorithm;

[0011] Based on the color changes of the three-dimensional scene and component fusion, establish a color fusion optimization model, and smooth and optimize the colors of the three-dimensional scene and the splicing interfaces of each component;

[0012] Based on the surface optimization results and color optimization results between the three-dimensional scene and the components, establish a three-dimensional scene component fusion optimization model based on the neural network algorithm to obtain a three-dimensional scene model with high quality and high compatibility in componentized fusion.

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

[0014] Based on big data, obtain the componentized components of the three-dimensional scene, and formulate multiple industry templates, where the industry templates serve as the basic templates of the three-dimensional scene;

[0015] Based on three-dimensional design software, build a three-dimensional model design platform, and through the platform, fuse the three-dimensional scene and the componentized components and conduct scene design.

[0016] Preferably, the establishment of the spatial rectangular coordinate system, conducting a detailed grid segmentation of the three-dimensional scene, and dividing the three-dimensional scene into several cubes specifically includes:

[0017] Based on the three-dimensional model design platform, establish a spatial rectangular coordinate system according to the three-dimensional scene template and components, and set the grid segmentation interval of the three-dimensional scene;

[0018] Take the grid segmentation interval distance of the three-dimensional scene 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, obtain the spatial coordinate information of several cubes after the three-dimensional scene is segmented.

[0020] Preferably, the establishment of the effective surface optimization model based on 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:

[0021] According to the spatial coordinate information of each cube, obtain the distribution maps of scene elements on different planes of each cube, and establish an atlas of distribution maps of scene elements on different planes of the cube;

[0022] Perform grayscale processing on the pictures in the atlas of distribution maps of scene elements on different planes of the cube to obtain an atlas of distribution maps of scene elements on different planes of the grayscaled cube;

[0023] Through a filtering algorithm, perform filtering and noise reduction processing on the atlas of distribution maps of scene elements on different planes of the grayscaled cube to optimize the picture quality;

[0024] Set the grayscale value of the pixel points without scene elements to the invalid point value, and according to the distribution of the invalid point values, obtain the segmentation curves of the scene elements on each face of the cube;

[0025] According to the number of vertexes of the segmentation curves on different faces of the cube containing scene elements, obtain the spatial coordinate values of the face points on each face of the cube;

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

[0027] According to the spatial coordinate values of the face points and edge points on each face of the cube, obtain the updated vertex spatial coordinate values on each face of the cube;

[0028] According to the spatial coordinate values of the face points, edge points and updated vertexes on each face of the cube, obtain a new cube segmented by the segmentation curves, and use the segmented cube containing scene elements as the base point for scene connection;

[0029] Based on the steps and methods of obtaining the cube segmented by the new segmentation curves, establish an effective surface optimization model, and through the update and iteration of the steps of obtaining the cube segmented by the new segmentation curves, smooth the effective surfaces of the scene elements contained in each cube to obtain the optimal base point for scene connection;

[0030] The expression for obtaining the spatial coordinate values of the face points on each face of the cube is:

[0031]

[0032] In the formula, is the spatial coordinate value of the face point on each face of the cube, is the number of vertexes of the segmentation curves on different faces of the cube containing scene elements, is the th vertex spatial coordinate value of a single face of the cube;

[0033] The expression for obtaining the spatial coordinate values of the edge points on each face of the cube is:

[0034]

[0035] Wherein, is the spatial coordinate value of the edge points of each face of the cube, , are respectively the spatial coordinate values of the two end points of the th side line, , are respectively the spatial coordinate values of the edge points of the two faces having the same th side line;

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

[0037]

[0038] Wherein, is the updated vertex spatial coordinate value of the th original vertex of the cube, is the spatial coordinate value of the face point adjacent to the th original vertex, is the spatial coordinate value of the edge point of the th side line connected to the th original vertex, is the number of vertices of the segmentation curve of different faces of the cube containing scene elements, is the number of side lines connecting the vertices of the segmentation curve of different faces of the cube containing scene elements.

[0039] Preferably, the further optimization of the multi-angle fusion of the three-dimensional scene components based on the Laplacian pyramid algorithm according to the smoothing result of the effective surface optimization model specifically includes:

[0040] According to the smoothing result of the effective surface optimization model, obtain the optimal scene connection base point of each cube, and establish a scene connection base point set for the three-dimensional scene template and components;

[0041] According to the spatial coordinate system at the connection of the three-dimensional scene template and components, set the weight parameters of the three-dimensional scene template and components through the platform, and determine the adaptive connection of the three-dimensional scene template and components;

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

[0043] According to the weight parameters of the three-dimensional scene template and components set by the platform, determine the removed part during the fusion of the three-dimensional scene template and components, wherein,

[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 fusing the 3D scene template and the component, by adjusting the spatial coordinates at the connection of the components, the overlapping component parts in the spatial coordinates are removed;

[0045] If the weight parameter of the 3D scene template is set to be less than the weight parameter of the component, it means that when fusing the 3D scene template and the component, by adjusting the spatial coordinates at the connection of the 3D scene template, the overlapping 3D scene template parts in the spatial coordinates are removed;

[0046] Obtain multi-angle images after fusing the 3D scene template and the component, and establish a multi-angle image set after fusing the 3D scene template and the component;

[0047] Based on the multi-angle image set after fusing the 3D scene template and the component, further optimize the multi-angle fusion of 3D scene components based on the Laplacian pyramid algorithm.

[0048] Preferably, the specific steps of establishing a color fusion optimization model according to the color change of the 3D scene and component fusion and performing smooth optimization processing on the colors of the 3D scene and each component splicing interface include:

[0049] Obtain multi-angle images after optimizing the multi-angle fusion of 3D scene components, and establish a multi-angle image set after optimizing the multi-angle fusion of 3D scene components;

[0050] According to the multi-angle image set after optimizing the multi-angle fusion of 3D scene components, obtain the source color values of the multi-angle fusion area of 3D scene components, and establish a source color value input matrix for the multi-angle fusion area of 3D scene components;

[0051] Based on big data, obtain the target color values of the multi-angle fusion area of 3D scene components, and establish a standard input matrix for the target color values of the multi-angle fusion area of 3D scene components;

[0052] According to the source color value input matrix and the standard input matrix of the target color values, obtain the deviation values of the fusion color values from the source color values and the target color values;

[0053] According to the source color value input matrix and the standard input matrix of the target color values, obtain the smooth values of the fusion color values from the source color values and the target color values;

[0054] According to the deviation values and smooth values of the fusion color values from the source color values and the target color values, establish a loss function of the fusion color values from the source color values and the target color values;

[0055] Based on the gradient descent algorithm, determine the update function of the fused color values in the multi-angle fusion area of 3D scene components;

[0056] Based on the above steps, a color fusion optimization model is established to perform smooth optimization on the colors of the three-dimensional scene and the splicing 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 set of color values in the fusion area, is at pixel the fused color value at point, is at pixel the source color value at point, is at pixel the target color value at point;

[0060] The expression for obtaining the smooth value between the fused color value and the source color value and the target color value is:

[0061]

[0062] In the formula, is the smooth value between the fused color value and the source color value and the target color value, is the neighborhood of pixel point, is at pixel the fused color value at point;

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

[0064]

[0065] In the formula, is the loss value between the fused color value and the source color value and the target color value, is the smooth weight balance value;

[0066] The expression for the gradient descent algorithm is:

[0067]

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

[0069] Preferably, based on the surface optimization result and color optimization result 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 obtaining a three-dimensional scene model with high-quality and high-fit componentization fusion specifically includes:

[0070] Based on big data, obtain the expected data of multi-angle fusion of three-dimensional scene components as the target data, and establish the output target matrix of the neural network;

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

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

[0073] Based on the neural network algorithm, establish a three-dimensional scene component fusion optimization model, comprehensively adjust the three-dimensional scene model, and obtain a three-dimensional scene model with high-quality and high-fit componentization fusion.

[0074] Furthermore, this solution proposes a three-dimensional model design system based on componentization for implementing the three-dimensional model design method based on componentization as described above, including:

[0075] The platform building module is used to obtain the componentized components of the 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] The surface fusion optimization module is used to establish a spatial rectangular coordinate system, conduct a detailed grid segmentation of the three-dimensional scene, and divide the three-dimensional scene into several cubes; according to the effective surface of the scene elements contained in each cube, establish an effective surface optimization model, and smooth the effective surface of the scene elements contained in each cube; based on the smoothing result of the effective surface optimization model, further optimize the multi-angle fusion of the three-dimensional scene components based on the Laplacian pyramid algorithm;

[0077] The color fusion optimization module is used to establish a color fusion optimization model according to the color change of the three-dimensional scene and component fusion, and perform a smoothing optimization process on the colors of the three-dimensional scene and the splicing interfaces 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 result and color optimization result between the three-dimensional scene and the components, and obtain a three-dimensional scene model with high-quality and high-fit componentization fusion.

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

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

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

[0082] A surface fusion optimization unit, which is used to further optimize the multi-angle fusion of three-dimensional scene components based on the Laplacian pyramid algorithm according to the smoothing result of the effective surface optimization model.

[0083] Compared with the prior art, the beneficial effects of the present invention are as follows:

[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 on each coordinate axis in the spatial rectangular coordinate system, the spatial coordinate information of a number of cubes after the three-dimensional scene segmentation 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 divided by the segmented curve is obtained, and an effective surface optimization model is established. Through the update and iteration of the steps of obtaining the cube divided by the new segmented curve, the effective surface of the scene elements contained in each cube is smoothed. By taking the segmented cube containing the scene elements as the base point of scene connection, the optimal base point of scene connection is obtained, and according to the smoothing result of the effective surface optimization model, the multi-angle fusion of three-dimensional scene components is further optimized based on the Laplacian pyramid algorithm. Furthermore, according to the color change of the three-dimensional scene and component fusion, by analyzing the loss function relationship between the fusion color value and the source color value and target color value, a color fusion optimization model is established based on the gradient descent algorithm to smooth and optimize the color of the three-dimensional scene and the splicing interface of each component. Finally, according to the surface optimization result and color optimization result between the three-dimensional scene and the component, a three-dimensional scene component fusion optimization model is established based on the neural network algorithm to comprehensively adjust the three-dimensional scene model, and a three-dimensional scene model with high quality and high fit degree of componentized fusion is obtained, thereby effectively improving the quality and fit degree of the fusion of the three-dimensional scene and the component, and reducing the problems of incompatibility, gaps, overlaps, or color breaks when the three-dimensional scene and the component are fused. Description of the Drawings

[0085] Figure 1 It is a flowchart of a component-based three-dimensional model design method of the present invention;

[0086] Figure 2 Flow chart for establishing an effective surface optimization model based on the effective surface of the scene elements contained in each cube of the present invention and smoothing the effective surface of the scene elements contained in each cube

[0087] Figure 3 Flow chart for further optimizing the multi-angle fusion of three-dimensional scene components based on the Laplacian pyramid algorithm according to the smoothing result of the effective surface optimization model of the present invention

[0088] Figure 4 Flow chart for establishing a color fusion optimization model based on the color change of the three-dimensional scene and component fusion of the present invention and smoothing and optimizing the color of the interfaces between the three-dimensional scene and each component Detailed implementation manners

[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 in the following description are only examples, and those skilled in the art can think of other obvious variants.

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

[0091] Based on big data, obtain the componentized components of the three-dimensional scene, formulate multiple industry templates, and build a three-dimensional model design platform based on three-dimensional design software;

[0092] Establish a space rectangular coordinate system, conduct a detailed grid division of the three-dimensional scene, and divide the three-dimensional scene into several cubes;

[0093] Based on the effective surface of the scene elements contained in each cube, establish an effective surface optimization model and smooth the effective surface of the scene elements contained in each cube;

[0094] According to the smoothing result of the effective surface optimization model, further optimize the multi-angle fusion of the three-dimensional scene components based on the Laplacian pyramid algorithm;

[0095] Based on the color change of the three-dimensional scene and component fusion, establish a color fusion optimization model and smooth and optimize the color of the interfaces between the three-dimensional scene and each component;

[0096] Based on the surface optimization result and color optimization result between the three-dimensional scene and the components, establish a three-dimensional scene component fusion optimization model based on the neural network algorithm to obtain a high-quality and highly compatible componentized fusion three-dimensional scene model.

[0097] It can be explained that in this solution, by establishing a spatial rectangular coordinate system, setting the grid segmentation interval of the three-dimensional scene, and using the grid segmentation interval distance of the three-dimensional scene as the coordinate distance between two adjacent points on each coordinate axis in the spatial rectangular coordinate system, the spatial coordinate information of several cube spaces after the three-dimensional scene is segmented is obtained. Secondly, according to the spatial coordinate information of each cube, by obtaining the spatial coordinate values of the surface points, edge points, and updated vertex points of the effective surface of the scene elements contained in each cube, the cube after being segmented by the new segmented curve is obtained, and an effective surface optimization model is established. Through the update and iteration of the steps of obtaining the cube after being segmented by the new segmented curve, the effective surface of the scene elements contained in each cube is smoothed. By using the segmented cube containing the scene elements as the base point for scene connection, the optimal base point for scene connection is obtained. And according to the smoothing result of the effective surface optimization model, based on the Laplacian pyramid algorithm, the multi-angle fusion of the three-dimensional scene components is further optimized. Furthermore, according to the color change of the three-dimensional scene and component fusion, by analyzing the loss function relationship between the fusion color value and the source color value and the target color value, based on the gradient descent algorithm, a color fusion optimization model is established to smooth and optimize the color of the three-dimensional scene and the splicing interface of each component. Finally, according to the surface optimization result and color optimization result between the three-dimensional scene and the component, based on the neural network algorithm, a three-dimensional scene component fusion optimization model is established to comprehensively adjust the three-dimensional scene model, and a three-dimensional scene model with high quality and high compatibility of component fusion is obtained, thereby effectively improving the quality and compatibility of the three-dimensional scene and component fusion, and reducing the problems of incompatibility, gaps, overlaps, or color breaks when the three-dimensional scene and the component are fused.

[0098] Referring to Figure 2 As shown, the establishment of the effective surface optimization model according to the effective surface of the scene elements contained in each cube and the smoothing process of the effective surface of the scene elements contained in each cube specifically include:

[0099] According to the spatial coordinate information of each cube, obtain the distribution map of the scene elements on different planes of each cube, and establish an atlas of the distribution maps of the scene elements on different planes of the cube;

[0100] Grayscale the pictures in the atlas of the distribution maps of the scene elements on different planes of the cube to obtain the grayscaled atlas of the distribution maps of the scene elements on different planes of the cube;

[0101] Through the filtering algorithm, perform filtering and noise reduction processing on the grayscaled atlas of the distribution maps of the scene elements on different planes of the cube to optimize the picture quality;

[0102] Set the grayscale value of the pixel points without scene elements as the invalid position value, and according to the distribution of the invalid position values, obtain the segmented curve of the scene elements contained in each face of the cube;

[0103] Obtain the spatial coordinate values of the surface points of each face of the cube according to the number of vertexes of the segmentation curve on different faces of the cube containing scene elements;

[0104] Obtain the spatial coordinate values of the edge points of each face of the cube according to the spatial coordinate values of the surface points of each face of the cube;

[0105] Obtain the updated spatial coordinate values of the vertexes of each face of the cube according to the spatial coordinate values of the surface points and the spatial coordinate values of the edge points of each face of the cube;

[0106] Obtain a new cube segmented by the segmentation curve according to the spatial coordinate values of the surface points, the spatial coordinate values of the edge points, and the updated spatial coordinate values of the vertexes of each face of the cube, and use the segmented cube containing scene elements as the base point for scene connection;

[0107] Based on the steps and methods of obtaining the new cube segmented by the segmentation curve, establish an effective surface optimization model. Through the update and iteration of the steps of obtaining the new cube segmented by the segmentation curve, smooth the effective surface of the scene elements contained in each cube to obtain the optimal base point for scene connection;

[0108] The expression for obtaining the spatial coordinate values of the surface points of each face of the cube is:

[0109]

[0110] In the formula, is the spatial coordinate value of the surface point of each face of the cube, is the number of vertexes of the segmentation curve on different faces of the cube containing scene elements, h is the th vertex spatial coordinate value of a single face of the cube;

[0111] The expression for obtaining the spatial coordinate values of the edge points 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, , are respectively the spatial coordinate values of the two end points of the th side line, , are respectively the spatial coordinate values of the edge points of the two faces with the same th side line;

[0114] The expression for obtaining the updated spatial coordinate values of the vertexes of each face of the cube is:

[0115]

[0116] In the formula, is the vertex space coordinate value after updating the th original vertex of the cube, is the face point space coordinate value adjacent to the th original vertex, is the edge point space coordinate value of the th original vertex connected to the th edge line, is the number of vertexes of the segmentation curve of different faces of the cube containing scene elements, is the number of edge lines connecting the vertexes of the segmentation curve of different faces of the cube containing scene elements.

[0117] It can be explained that when performing the fusion of the 3D scene and components, due to the possible incomplete matching of the geometric shapes, topological structures, and sizes of the 3D scene and different components, problems such as incompatibility, gaps, and overlaps are likely to occur during the fusion of the 3D scene and components, thus affecting the design quality of the 3D model. In this solution, by establishing a spatial rectangular coordinate system, setting the grid segmentation interval of the 3D scene, and using the grid segmentation interval distance of the 3D scene as the coordinate distance between two adjacent points on each coordinate axis in the spatial rectangular coordinate system, several cube spatial coordinate information after the segmentation of the 3D scene is obtained. And according to each cube spatial coordinate information, by obtaining the face point space coordinate value, edge point space coordinate value, and updated vertex space coordinate value 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. Through the update and iteration of the steps of obtaining the cube segmented by the new segmentation curve, the effective surface of the scene elements contained in each cube is smoothed. By using the segmented cube containing scene elements as the base point for scene connection, the optimal base point for scene connection is obtained.

[0118] Referring to Figure 3 shown, further optimization of the multi-angle fusion of the 3D scene components based on the Laplace pyramid algorithm according to the smoothing result of the effective surface optimization model specifically includes:

[0119] According to the smoothing result of the effective surface optimization model, obtain the optimal base point for scene connection of each cube, and establish a scene connection base point set for the 3D scene template and components;

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

[0121] The specific adaptive connection method of the 3D scene template and components includes:

[0122] Set the weight parameters of the 3D scene template and components according to the platform, and determine the parts to be removed when the 3D scene template and components are fused. Among them,

[0123] If the weight parameter of the 3D scene template is set to be greater than that of the component, it means that when the 3D scene template and components are fused, by adjusting the spatial coordinates at the connection of the components, the overlapping component parts in spatial coordinates are removed;

[0124] If the weight parameter of the 3D scene template is set to be less than that of the component, it means that when the 3D scene template and components are fused, by adjusting the spatial coordinates at the connection of the 3D scene template, the overlapping 3D scene template parts in spatial coordinates are removed;

[0125] Obtain multi-angle pictures after the fusion of the 3D scene template and components, and establish a multi-angle picture set after the fusion of the 3D scene template and components;

[0126] According to the multi-angle picture set after the fusion of the 3D scene template and components, based on the Laplacian pyramid algorithm, further optimize the multi-angle fusion of 3D scene components.

[0127] It can be explained that according to the effective surface optimization model, the spatial coordinates of the optimal base points for scene connection can be effectively obtained. By setting the weight parameters of the 3D scene template and components on the platform, the adaptive connection of the 3D scene template and components is determined. Thus, by judging whether the weight parameter of the 3D scene template is greater than that of the component, if so, it means that when the 3D scene template and components are fused, by adjusting the spatial coordinates at the connection of the components, the overlapping component parts in spatial coordinates are removed; if not, it means that when the 3D scene template and components are fused, by adjusting the spatial coordinates at the connection of the 3D scene template, the overlapping 3D scene template parts in spatial coordinates are removed. Thereby, the fitting degree of the fusion of the 3D scene template and components is effectively improved. To further improve the smoothness of multiple angles when the 3D scene template and components are fused and reduce the influence of singular points in the fusion area, this solution further optimizes the multi-angle fusion of 3D scene components through the Laplacian pyramid algorithm;

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

[0129] According to the multi-angle picture set after the fusion of the 3D scene template and components, separate the largest image and the smallest image, and construct an image pyramid model;

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

[0131] Based on the Gaussian pyramid model and the Laplacian pyramid algorithm, the multi-angle atlas after the fusion of the three-dimensional scene template and components after replacement and reconstruction is optimized to further improve the quality of the multi-angle fusion of the three-dimensional scene components.

[0132] Referring to Figure 4 As shown, the establishment of a color fusion optimization model according to the color changes in the fusion of the three-dimensional scene and components, and the smooth optimization of the colors at the joints of the three-dimensional scene and each component specifically include:

[0133] Obtain the multi-angle pictures after the optimization of the multi-angle fusion of the three-dimensional scene components, and establish an optimized atlas for the multi-angle fusion of the three-dimensional scene components;

[0134] According to the optimized atlas for the multi-angle fusion of the three-dimensional scene components, obtain the source color values in the multi-angle fusion area of the three-dimensional scene components, and establish a source color value input matrix for the multi-angle fusion area of the three-dimensional scene components;

[0135] Based on big data, obtain the target color values in the multi-angle fusion area of the three-dimensional scene components, and establish a standard input matrix for the target color values in the multi-angle fusion area of the three-dimensional scene components;

[0136] According to the source color value input matrix and the standard input matrix of the target color values, obtain the deviation values of the fusion color values from the source color values and the target color values;

[0137] According to the source color value input matrix and the standard input matrix of the target color values, obtain the smooth values of the fusion color values from the source color values and the target color values;

[0138] According to the deviation values and smooth values of the fusion color values from the source color values and the target color values, establish a loss function for the fusion color values from the source color values and the target color values;

[0139] Based on the gradient descent algorithm, determine the update function of the fused color values in the multi-angle fusion area of the three-dimensional scene components;

[0140] Based on the above steps, establish a color fusion optimization model to smoothly optimize the colors at the joints of the three-dimensional scene and each component;

[0141] The expression for obtaining the deviation value of the fusion color value from the source color value and the target color value is:

[0142]

[0143] In the formula, is the deviation value of the fusion color value from the source color value and the target color value, is the set of color values in the fusion area, is at the pixel point, the fusion color value, is the source color value at the pixel point, is the target color value at the pixel point;

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

[0145]

[0146] In the formula, is the smoothed value of the fused color value with the source color value and the target color value, is the neighborhood of the pixel point, is the fused color value at the pixel point;

[0147] The expression for the loss function of the fused color value with 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 smoothed weight balance value;

[0150] The expression for the gradient descent algorithm is:

[0151]

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

[0153] It can be explained that when performing the fusion of a 3D scene and components, the color fusion fitness between different components and the 3D scene is also particularly important. The color difference in the fusion area can be easily felt intuitively by observation. Therefore, improving the color fusion during the fusion of a 3D scene and components plays a very important role in improving the 3D model design. According to the color changes in the fusion of a 3D scene and components, this solution obtains the fused color value, the source color value, and the target color value in the fusion area, establishes the loss function relationship between the fused color value and the source color value and the target color value, and based on the gradient descent algorithm, establishes a color fusion optimization model. By reducing the loss value of the fused color value with the source color value and the target color value, the color of the splicing interface between the 3D scene and each component is smoothed and optimized, and the color fusion quality during the fusion of the 3D scene and components is improved.

[0154] Furthermore, 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 3D scene based on big data, formulate various industry templates, and build a 3D model design platform based on 3D design software;

[0156] A surface fusion optimization module, which is used to establish a spatial rectangular coordinate system, conduct a detailed grid segmentation of the 3D scene, and divide the 3D scene into several cubes; establish an effective surface optimization model according to the effective surfaces of the scene elements contained in each cube, and smooth the effective surfaces of the scene elements contained in each cube; based on the smoothing results of the effective surface optimization model, further optimize the multi-angle fusion of the 3D scene components based on the Laplacian pyramid algorithm;

[0157] A color fusion optimization module, which is used to establish a color fusion optimization model according to the color changes of the 3D scene and component fusion, and conduct a smoothing optimization process on the colors of the 3D scene and the splicing interfaces of each component;

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

[0159] The surface fusion optimization module includes:

[0160] A scene segmentation unit, which is used to establish a spatial rectangular coordinate system, conduct a detailed grid segmentation of the 3D scene, and divide the 3D scene into several cubes;

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

[0162] A surface fusion optimization unit, which is used to further optimize the multi-angle fusion of the 3D scene components based on the Laplacian pyramid algorithm according to the smoothing results of the effective surface optimization model.

[0163] In summary, the advantages of the present invention are as follows: effectively improving the quality and compatibility of the fusion between the 3D scene and the components, and reducing the problems of incompatibility, gaps, overlaps, or color breaks when the 3D scene and the components are fused.

[0164] The basic principles, main features and advantages of the present invention have been shown and described above. Those skilled in the art should understand that the present invention is not limited by the above embodiments. What is described in the above embodiments and the specification is only the principle of the present invention. Without departing from the spirit and scope of the present invention, various changes and improvements will occur to the present invention, and these changes and improvements all fall within the scope of the present invention claimed. The scope of protection required by the present invention is defined by the appended 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 3D scene and components, a 3D scene component fusion optimization model is established based on a neural network algorithm to obtain a high-quality, high-fit component-fused 3D scene model; 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; 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; 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, E R 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, R(p) is the fusion color value at pixel p, R y (p) is the source color value at pixel p, R m (p) is the target color value at pixel p; 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, E q is the smoothed value of the fused color value, the source color value and the target color value, N(p) is the neighborhood of pixel p, and R(q) is the fused color value at pixel q; The loss function expression of the fused color value, the source color value and the target color value is: E=E R +δE q Where E is the loss value of the fused color value, the source color value, and the target color value, and δ is the smoothing weight balance value; The gradient descent algorithm expression is: In the formula, is the updated value of the fused color value at pixel p, and γ is the learning rate.

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: Where f(x, y, z) is the spatial coordinate value of the point on each face of the cube, n is the number of vertices of the segmentation curve of the cube containing different faces of the scene elements, and h(x i ,y i ,z i ) is the spatial coordinate value of the i-th vertex of a single face of the cube; The expression for obtaining the spatial coordinate value of the edge point of each face of the cube is: In the formula, g(x,y,z) is the spatial coordinate value of the edge point of each face of the cube, v 1-i (x,y,z),v 2-i (x, y, z) are the spatial coordinates of the two endpoints of the i-th edge line, respectively. 1-i (x,y,z),f 2-i (x, y, z) are the spatial coordinates of the edge points of the two faces with the same i-th edge line; The expression for obtaining the updated vertex space coordinate value of each face of the cube is: In the formula, is the updated vertex space coordinate value of the i-th original vertex of the cube, f i (x, y, z) is the spatial coordinate value of the surface point adjacent to the i-th original vertex, g j (x, y, z) is the spatial coordinate value of the edge point of the jth edge line connected to the i-th original vertex, n is the number of segmentation curve vertices of the cube containing different faces of scene elements, and m is the number of edge lines connecting the segmentation curve vertices of the cube containing different faces of scene elements.

5. A component-based 3D model design method according to claim 4, characterized in that: 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 between the 3D scene templates are adjusted to remove the part of the 3D scene template with overlapping spatial coordinates.

6. A component-based 3D model design method according to claim 5, 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.

7. 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 6, 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.

8. The component-based three-dimensional model design system according to claim 7, 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