Combat atmosphere rendering method based on virtual digital terrain

Abstract

This application relates to the field of virtual digitized terrain, and specifically to a method for rendering a combat atmosphere based on virtual digitized terrain. This method utilizes an intelligent deep learning-based algorithm to automatically extract and enhance the spatial distribution characteristics of three-dimensional virtual digitized terrain. Using the combat atmosphere rendering style representation vector as constraints and guidance information, this method performs cross-modal joint optimization of the spatial distribution characteristics of the three-dimensional virtual digitized terrain, thereby obtaining the three-dimensional virtual digitized terrain characteristics guided by the rendering style. Furthermore, a diffusion model is used to construct a style renderer, which automatically generates and renders the three-dimensional virtual digitized terrain based on the three-dimensional virtual digitized terrain characteristics guided by the rendering style, thereby enhancing the visual effect of the digitized terrain.

Description

Technical Field

[0001] The present application relates to the field of virtual digital terrain, and in particular to a method for rendering a combat atmosphere based on virtual digital terrain. Background Art

[0002] With the rapid development of virtual reality (VR) and augmented reality (AR) technologies, digital terrain is increasingly being used in military training, urban planning, environmental simulation, and other fields. To enhance the realism and immersion of these applications, digital terrain needs to be rendered to resemble a real-world battlefield environment, making it more realistic. Specifically, realistic combat atmosphere rendering involves using technical means to simulate visual effects and atmosphere close to those of a real battlefield environment, enhancing training effectiveness or providing a more realistic experience.

[0003] Traditional rendering methods often rely on manual design and adjustment, but this approach is highly dependent on the operator's skill level, knowledge level, and experience, and suffers from problems such as low efficiency and long time consumption. Therefore, an optimized method for rendering combat atmosphere based on virtual digital terrain is desired. Summary of the Invention

[0004] This application is made in consideration of the above problems. One purpose of this application is to provide a method for rendering a combat atmosphere based on virtual digital terrain, which can enhance the visual effect of the digital terrain.

[0005] The embodiment of the present application provides a method for rendering a combat atmosphere based on virtual digital terrain, which includes:

[0006] Import three-dimensional virtual digital terrain;

[0007] Import actual combat atmosphere rendering style representation vector;

[0008] Extracting terrain features from the three-dimensional virtual digitized terrain to obtain a three-dimensional virtual digitized terrain feature map;

[0009] Inputting the three-dimensional virtual digitized terrain feature map into a feature attention selection and enhancement module based on a compression-suppression structure to obtain an enhanced three-dimensional virtual digitized terrain feature map;

[0010] Cross-modally fusing the actual combat atmosphere rendering style representation vector and the enhanced three-dimensional virtual digitized terrain feature map to obtain a three-dimensional virtual digitized terrain feature map guided by the rendering style;

[0011] Based on the three-dimensional virtual digitized terrain feature map guided by the rendering style, a rendered three-dimensional virtual digitized terrain is generated.

[0012] Furthermore, according to the above-mentioned method for rendering a combat atmosphere based on virtual digital terrain, the step of extracting terrain features from the three-dimensional virtual digital terrain to obtain a three-dimensional virtual digital terrain feature map includes:

[0013] The three-dimensional virtual digitized terrain is input into a virtual terrain feature extractor based on a three-dimensional convolutional neural network model to obtain the three-dimensional virtual digitized terrain feature map.

[0014] Furthermore, according to the above-mentioned method for rendering a combat atmosphere based on virtual digital terrain, the step of inputting the three-dimensional virtual digital terrain feature map into a feature attention selection enhancement module based on a compression-suppression structure to obtain an enhanced three-dimensional virtual digital terrain feature map includes:

[0015] Calculating the global mean of each feature matrix along the channel dimension of the three-dimensional virtual digitized terrain feature map to obtain a three-dimensional virtual digitized terrain feature compression information representation vector;

[0016] Performing one-dimensional convolution coding on the three-dimensional virtual digitized terrain feature compression representation vector to obtain a feature vector representing the correlation between the three-dimensional virtual digitized terrain feature compression information;

[0017] Cascading the three-dimensional virtual digitized terrain feature compression information representation vector and the three-dimensional virtual digitized terrain feature compression information association representation feature vector to obtain a three-dimensional virtual digitized terrain feature compression information multi-scale representation vector;

[0018] Inputting the multi-scale representation vector of the compressed information of the three-dimensional virtual digitized terrain features into a compression information feature extraction module including a multi-layer perceptron and a SiLU activation function to obtain a multi-scale correlation feature vector of the compressed information of the three-dimensional virtual digitized terrain features;

[0019] Normalizing the multi-scale correlation feature vector of the three-dimensional virtual digitized terrain feature compression information to obtain a three-dimensional virtual digitized terrain weight feature vector;

[0020] Based on the three-dimensional virtual digitized terrain weight feature vector, feature amplification and suppression operations are performed on the three-dimensional virtual digitized terrain feature map to obtain the enhanced three-dimensional virtual digitized terrain feature map.

[0021] Furthermore, according to the above-mentioned method for rendering a combat atmosphere based on virtual digitized terrain, a normalization operation is performed on the multi-scale correlation feature vector of the three-dimensional virtual digitized terrain feature compression information to obtain a three-dimensional virtual digitized terrain weight feature vector, including:

[0022] A Sigmoid function is used to perform a normalization operation on the multi-scale correlation feature vector of the three-dimensional virtual digitized terrain feature compression information to obtain the three-dimensional virtual digitized terrain weight feature vector.

[0023] Furthermore, according to the above-mentioned method for rendering a combat atmosphere based on virtual digitized terrain, the method further comprises: performing feature amplification and suppression operations on the three-dimensional virtual digitized terrain feature map based on the three-dimensional virtual digitized terrain weight feature vector to obtain the enhanced three-dimensional virtual digitized terrain feature map, including:

[0024] The Kronecker product of the three-dimensional virtual digitized terrain weight feature vector and each feature matrix along the channel dimension of the three-dimensional virtual digitized terrain feature map is calculated to obtain the enhanced three-dimensional virtual digitized terrain feature map.

[0025] Furthermore, according to the above-mentioned method for rendering a combat atmosphere based on virtual digital terrain, cross-modal fusion of the combat atmosphere rendering style representation vector and the enhanced three-dimensional virtual digital terrain feature map is performed to obtain a three-dimensional virtual digital terrain feature map guided by the rendering style, including:

[0026] The actual combat atmosphere rendering style representation vector and the enhanced three-dimensional virtual digitized terrain feature map are input into a cross-modal joint encoder based on the MetaNet model to obtain the three-dimensional virtual digitized terrain feature map guided by the rendering style.

[0027] Furthermore, according to the above-mentioned method for rendering a combat atmosphere based on virtual digital terrain, generating and rendering a three-dimensional virtual digital terrain based on a three-dimensional virtual digital terrain feature map guided by the rendering style includes:

[0028] The three-dimensional virtual digitized terrain feature map guided by the rendering style is input into a style renderer based on a diffusion model to obtain the rendered three-dimensional virtual digitized terrain.

[0029] Furthermore, according to the above-mentioned method for rendering a combat atmosphere based on virtual digital terrain, the method further includes:

[0030] The virtual terrain feature extractor based on the three-dimensional convolutional neural network model, the feature attention selection and enhancement module based on the compression-suppression structure, the cross-modal joint encoder based on the MetaNet model, and the style renderer based on the diffusion model are trained.

[0031] Furthermore, according to the above-mentioned method for rendering a combat atmosphere based on virtual digital terrain, the training of the virtual terrain feature extractor based on the three-dimensional convolutional neural network model, the feature attention selection and enhancement module based on the compression-suppression structure, the cross-modal joint encoder based on the MetaNet model, and the style renderer based on the diffusion model includes:

[0032] Acquire training data, the training data including a training three-dimensional virtual digitized terrain, a training actual combat atmosphere rendering style representation vector, and a real value of the rendered three-dimensional virtual digitized terrain;

[0033] Inputting the training three-dimensional virtual digitized terrain into the virtual terrain feature extractor based on the three-dimensional convolutional neural network model to obtain a training three-dimensional virtual digitized terrain feature map;

[0034] Inputting the training three-dimensional virtual digitized terrain feature map into the feature attention selection and enhancement module based on the compression-suppression structure to obtain a training enhanced three-dimensional virtual digitized terrain feature map;

[0035] Inputting the training combat atmosphere rendering style representation vector and the training enhanced three-dimensional virtual digitized terrain feature map into the cross-modal joint encoder based on the MetaNet model to obtain a three-dimensional virtual digitized terrain feature map guided by the training rendering style;

[0036] Inputting the three-dimensional virtual digitized terrain feature map guided by the training rendering style into a style renderer based on a diffusion model to obtain a generation loss function value;

[0037] Calculating a predetermined loss function value of the training rendering style-guided three-dimensional virtual digitized terrain feature map to obtain a rendering style-guided three-dimensional virtual digitized terrain loss function value;

[0038] The weighted sum of the generation loss function value and the rendering style-guided three-dimensional virtual digitized terrain loss function value is used as the loss function value to train the virtual terrain feature extractor based on the three-dimensional convolutional neural network model, the feature attention selection and enhancement module based on the compression-suppression structure, the cross-modal joint encoder based on the MetaNet model, and the style renderer based on the diffusion model.

[0039] According to the embodiment of the present application, the method for rendering a combat atmosphere based on virtual digital terrain uses an intelligent algorithm based on deep learning to automatically extract and enhance the spatial distribution characteristics of the three-dimensional virtual digital terrain, and uses the combat atmosphere rendering style representation vector as constraint and guidance information to perform cross-modal joint optimization of the spatial distribution characteristics of the three-dimensional virtual digital terrain, thereby obtaining the three-dimensional virtual digital terrain characteristics guided by the rendering style, and uses the diffusion model to construct a style renderer to automatically generate and render the three-dimensional virtual digital terrain based on the three-dimensional virtual digital terrain characteristics guided by the rendering style. This rendering method is highly efficient and can enhance the visual effect of the digital terrain. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following briefly introduces the drawings of the embodiments of the present application. Obviously, the drawings described below only relate to some embodiments of the present application, and are not intended to limit the present application.

[0041] Figure 1 A schematic diagram of the application architecture of a method for rendering a combat atmosphere based on virtual digital terrain according to an embodiment of the present application is shown;

[0042] Figure 2 A flowchart of a method for rendering a combat atmosphere based on virtual digital terrain in an embodiment of the present application is shown;

[0043] Figure 3 A flowchart of sub-step S540 of the method for rendering a combat atmosphere based on virtual digital terrain in an embodiment of the present application is shown;

[0044] Figure 4 A schematic structural diagram of a combat atmosphere rendering system based on virtual digital terrain in an embodiment of the present application is shown;

[0045] Figure 5 A diagram showing an application scenario of a method for rendering a combat atmosphere based on virtual digital terrain according to an embodiment of the present application is shown;

[0046] Figure 6 A terrain white model according to an embodiment of the present application is shown. DETAILED DESCRIPTION

[0047] The following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the accompanying drawings. Obviously, the embodiments described are only some embodiments of the present application, not all embodiments. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without making creative efforts also fall within the scope of protection of this application.

[0048] The terms used in this specification are those commonly used in the art currently in consideration of the functions of the present application, but these terms may vary according to the intentions of those skilled in the art, precedents, or new technologies in the art. In addition, specific terms may be selected, and in such cases, their detailed meanings will be described in the detailed description of the present application. Therefore, the terms used in the specification should not be understood as simple names, but rather as the meaning of the terms and the overall description of the present application.

[0049] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules can be used and run on the user terminal and / or server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0050] Flowcharts are used throughout this application to illustrate the operations performed by the systems of the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed in exact order. Instead, the various steps may be processed in reverse order or simultaneously, as needed. Furthermore, other operations may be added to these processes, or one or more operations may be removed from these processes.

[0051] Figure 1 A schematic diagram of the application architecture of a method for rendering a combat atmosphere based on virtual digital terrain in an embodiment of the present application is shown, including a server 100 and a terminal device 200.

[0052] The terminal device 200 and the server 100 can be connected via the Internet to enable communication between them. Optionally, the Internet utilizes standard communication technologies and / or protocols. The Internet is typically the Internet, but may also be any network, including but not limited to a local area network (LAN), a metropolitan area network (MAN), a wide area network (WAN), a mobile, wired, or wireless network, a private network, or any combination of a virtual private network. In some embodiments, technologies and / or formats such as Hypertext Markup Language (HTML) and Extensible Markup Language (XML) are used to represent data exchanged over the network. Conventional encryption technologies such as Secure Sockets Layer (SSL), Transport Layer Security (TLS), Virtual Private Networks (VPN), and Internet Protocol Security (IPsec) may also be used to encrypt all or some links. In other embodiments, customized and / or dedicated data communication technologies may be used in place of or in addition to the aforementioned data communication technologies.

[0053] Server 100 can provide various network services to terminal device 200. Server 100 can be a single server, a server cluster consisting of multiple servers, or a cloud computing center. Specifically, server 100 may include a processor 110 (Center Processing Unit, CPU), memory 120, input devices 130, and output devices 140. Input devices 130 may include a keyboard, mouse, touch screen, etc. Output devices 140 may include display devices such as a liquid crystal display (LCD) or a cathode ray tube (CRT).

[0054] The memory 120 may include a read-only memory (ROM) and a random access memory (RAM), and provides program instructions and data stored in the memory 120 to the processor 110. In an embodiment of the present application, the memory 120 may be used to store the program of the method for rendering a real combat atmosphere based on virtual digitized terrain in an embodiment of the present application.

[0055] The processor 110 calls the program instructions stored in the memory 120, and the processor 110 is used to execute the steps of any one of the methods for rendering a real combat atmosphere based on virtual digitized terrain in the embodiments of the present application according to the obtained program instructions.

[0056] In addition, the application architecture diagram in the embodiment of the present application is intended to more clearly illustrate the technical solution in the embodiment of the present application, and does not constitute a limitation on the technical solution provided in the embodiment of the present application. Of course, for other application architectures and business applications, the technical solution provided in the embodiment of the present application is also applicable to similar problems.

[0057] The following non-restrictive description of the actual combat atmosphere rendering method based on virtual digitized terrain provided according to at least one embodiment of the present application is given through several examples or embodiments. As described below, different features in these specific examples or embodiments can be combined with each other without conflicting with each other to obtain new examples or embodiments, and these new examples or embodiments also fall within the scope of protection of this application.

[0058] In response to the above technical problems, the technical concept of this application is: to use an intelligent algorithm based on deep learning to automatically extract and enhance the spatial distribution characteristics of three-dimensional virtual digitized terrain, and use the actual combat atmosphere rendering style representation vector as constraints and guiding information to perform cross-modal joint optimization of the spatial distribution characteristics of the three-dimensional virtual digitized terrain, so as to obtain the three-dimensional virtual digitized terrain characteristics under the guidance of the rendering style, and use the diffusion model to construct a style renderer to automatically generate and render the three-dimensional virtual digitized terrain based on the three-dimensional virtual digitized terrain characteristics guided by the rendering style.

[0059] Based on this, Figure 2 The flowchart of the method for rendering a combat atmosphere based on virtual digital terrain in an embodiment of the present application is shown. For example, the method for rendering a combat atmosphere based on virtual digital terrain can be executed by a server, which can be Figure 1 The server 100 shown in FIG. Figure 2 As shown, the method for rendering a combat atmosphere based on virtual digital terrain according to an embodiment of the present application includes the following steps:

[0060] S510, importing three-dimensional virtual digital terrain;

[0061] S520, import the actual combat atmosphere rendering style representation vector;

[0062] S530, extracting terrain features from the three-dimensional virtual digitized terrain to obtain a three-dimensional virtual digitized terrain feature map;

[0063] S540, inputting the three-dimensional virtual digitized terrain feature map into a feature attention selection and enhancement module based on a compression-suppression structure to obtain an enhanced three-dimensional virtual digitized terrain feature map;

[0064] S550, cross-modally fuses the actual combat atmosphere rendering style representation vector and the enhanced 3D virtual digitized terrain feature map to obtain a 3D virtual digitized terrain feature map guided by the rendering style;

[0065] S560: Generate a rendered three-dimensional virtual digitized terrain based on the three-dimensional virtual digitized terrain feature map guided by the rendering style.

[0066] Specifically, in the technical solution of the present application, first, a three-dimensional virtual digitized terrain is imported. Among them, the three-dimensional virtual digitized terrain is a three-dimensional terrain model created using computer graphics technology, which can represent the terrain's elevation, slope, vegetation cover and other geographical features in a digital form. At the same time, a combat atmosphere rendering style representation vector is imported. Here, importing the combat atmosphere rendering style representation vector can ensure that the rendering results meet the requirements of specific military training or simulation scenarios in the subsequent processing process, and provide a customized visual experience. In an embodiment of the present application, the process of importing the combat atmosphere rendering style representation vector includes: first, clarifying the type of combat atmosphere that needs to be simulated, such as attack explosion effects, damage status, combat sound effects, bombing, artillery attacks and other special effects related to real combat; then expressing the requirements of the combat atmosphere in text form, and converting it into a structured vector representation through natural language processing technology.

[0067] The 3D virtual digitized terrain is then fed into a virtual terrain feature extractor based on a 3D convolutional neural network model to produce a 3D virtual digitized terrain feature map. 3D convolutional neural networks (3D CNNs) are particularly adept at capturing implicit features in spatial data. In terrain data, these features may include terrain undulations, slope, and roughness.

[0068] Accordingly, in step S530, terrain features are extracted from the three-dimensional virtual digitized terrain to obtain a three-dimensional virtual digitized terrain feature map, including: inputting the three-dimensional virtual digitized terrain into a virtual terrain feature extractor based on a three-dimensional convolutional neural network model to obtain the three-dimensional virtual digitized terrain feature map.

[0069] Considering that the three-dimensional virtual digitized terrain may contain some redundant information and background information, in the technical solution of the present application, the three-dimensional virtual digitized terrain feature map is further input into the feature attention selection and enhancement module based on the compression-suppression structure to obtain an enhanced three-dimensional virtual digitized terrain feature map. In particular, the feature attention selection and enhancement module based on the compression-suppression structure introduces the SiLU function, which can assign higher weights to important information in the feature distribution, allowing important terrain features to be reused multiple times, thereby suppressing the flow of irrelevant background feature information. More specifically, based on the idea of the Squeeze and Excitation (SE) structure, the feature attention selection and enhancement module based on the compression-suppression structure uses global average pooling (Avg-Pool) to compress the information of the three-dimensional virtual digitized terrain feature map, and uses one-dimensional convolutional coding to capture the correlation information between the compressed information. Feature extraction is then performed using a multi-layer perceptron (MLP) and SiLU activation function. The weight values of different channels are then obtained after normalization using the Sigmoid function. Finally, the weight values are combined to perform feature amplification and suppression operations on the three-dimensional virtual digitized terrain feature map, thereby improving the network's ability to select three-dimensional virtual digitized terrain features.

[0070] Accordingly, in step S540, Figure 3 As shown, the three-dimensional virtual digitized terrain feature map is input into the feature attention selection and enhancement module based on the compression-suppression structure to obtain an enhanced three-dimensional virtual digitized terrain feature map, including:

[0071] S541, calculating the global mean of each feature matrix along the channel dimension of the three-dimensional virtual digitized terrain feature map to obtain a three-dimensional virtual digitized terrain feature compression information representation vector;

[0072] S542, performing one-dimensional convolution coding on the three-dimensional virtual digitized terrain feature compression representation vector to obtain a feature vector representing the correlation between the three-dimensional virtual digitized terrain feature compression information;

[0073] S543, cascading the three-dimensional virtual digitized terrain feature compression information representation vector and the three-dimensional virtual digitized terrain feature compression information correlation representation feature vector to obtain a three-dimensional virtual digitized terrain feature compression information multi-scale representation vector;

[0074] S544, inputting the multi-scale representation vector of the compressed information of the three-dimensional virtual digitized terrain features into a compression information feature extraction module including a multi-layer perceptron and a SiLU activation function to obtain a multi-scale correlation feature vector of the compressed information of the three-dimensional virtual digitized terrain features;

[0075] S545, performing a normalization operation on the multi-scale correlation feature vector of the three-dimensional virtual digitized terrain feature compression information to obtain a three-dimensional virtual digitized terrain weight feature vector;

[0076] S546 , based on the three-dimensional virtual digitized terrain weight feature vector, perform feature amplification and suppression operations on the three-dimensional virtual digitized terrain feature map to obtain an enhanced three-dimensional virtual digitized terrain feature map.

[0077] In step S545, a normalization operation is performed on the multi-scale correlation feature vector of the three-dimensional virtual digitized terrain feature compression information to obtain a three-dimensional virtual digitized terrain weight feature vector, including: using a Sigmoid function to normalize the multi-scale correlation feature vector of the three-dimensional virtual digitized terrain feature compression information to obtain a three-dimensional virtual digitized terrain weight feature vector.

[0078] In step S546, based on the three-dimensional virtual digitized terrain weight feature vector, feature amplification and suppression operations are performed on the three-dimensional virtual digitized terrain feature map to obtain an enhanced three-dimensional virtual digitized terrain feature map, including: calculating the Kronecker product of the three-dimensional virtual digitized terrain weight feature vector and each feature matrix of the three-dimensional virtual digitized terrain feature map along the channel dimension to obtain the enhanced three-dimensional virtual digitized terrain feature map.

[0079] In a specific example, the three-dimensional virtual digitized terrain feature map is input into a feature attention selection and enhancement module based on a compression-suppression structure to obtain an enhanced three-dimensional virtual digitized terrain feature map, including: inputting the three-dimensional virtual digitized terrain feature map into the feature attention selection and enhancement module based on a compression-suppression structure and processing it with the following enhancement formula to obtain an enhanced three-dimensional virtual digitized terrain feature map; wherein the enhancement formula is:

[0080]

[0081]

[0082]

[0083] in, It is a 3D virtual digital terrain feature map. The coordinates in the channel are The eigenvalues of and are the height and width of the three-dimensional virtual digital terrain feature map, It is the three-dimensional virtual digital terrain feature compression information representation vector eigenvalues, It is the three-dimensional virtual digital terrain feature compression information representation vector, represents one-dimensional convolutional coding, Indicates cascade, It is a multi-scale representation vector of compressed information of three-dimensional virtual digital terrain features. represents a multilayer perceptron, represents the sigmoid function, It is a three-dimensional virtual digital terrain feature map. Represents the Kronecker product of the calculation vector and each feature matrix along the channel dimension of the feature map, It is an enhanced three-dimensional virtual digital terrain feature map.

[0084] Next, the combat atmosphere rendering style representation vector and the enhanced 3D virtual digitized terrain feature map are fed into a MetaNet-based cross-modal joint encoder to generate a rendering style-guided 3D virtual digitized terrain feature map. This MetaNet-based cross-modal joint encoder combines the combat atmosphere style features with the 3D virtual terrain features to form a terrain feature distribution constrained by the combat atmosphere rendering style. This provides richer input for rendering, helping the system understand the rendering intent and desired effect.

[0085] Accordingly, in step S550, the actual combat atmosphere rendering style representation vector and the enhanced three-dimensional virtual digitized terrain feature map are cross-modally fused to obtain a three-dimensional virtual digitized terrain feature map guided by the rendering style, including: inputting the actual combat atmosphere rendering style representation vector and the enhanced three-dimensional virtual digitized terrain feature map into a cross-modal joint encoder based on the MetaNet model to obtain a three-dimensional virtual digitized terrain feature map guided by the rendering style.

[0086] The rendering style-guided 3D virtual digitized terrain feature map is then fed into a style renderer based on a diffusion model to produce a rendered 3D virtual digitized terrain. In this way, the diffusion model can apply a specific visual style to the 3D virtual digitized terrain based on the given input, achieving automated rendering.

[0087] Accordingly, in step S560, based on the rendering style-guided three-dimensional virtual digitized terrain feature map, a rendered three-dimensional virtual digitized terrain is generated, including: inputting the rendering style-guided three-dimensional virtual digitized terrain feature map into a style renderer based on a diffusion model to obtain a rendered three-dimensional virtual digitized terrain.

[0088] In a preferred example, the three-dimensional virtual digitized terrain feature map is enhanced to express the three-dimensional local image semantic association features of the three-dimensional virtual digitized terrain based on attention selection enhancement. Therefore, after the actual combat atmosphere rendering style representation vector and the enhanced three-dimensional virtual digitized terrain feature map are input into the cross-modal joint encoder based on the MetaNet model, the three-dimensional virtual digitized terrain feature map obtained under the guidance of the rendering style will also have cross-modal and feature heterogeneous structure interaction differences based on feature channels. Therefore, it is expected to improve the expression effect of cross-modal semantic joint perception based on cross-modal channel local feature interaction differences of the three-dimensional virtual digitized terrain feature map guided by the rendering style.

[0089] In this preferred embodiment, the rendering style-guided three-dimensional virtual digitized terrain feature map is input into a style renderer based on a diffusion model to obtain a rendered three-dimensional virtual digitized terrain, including:

[0090] Calculating the sum of the absolute values of each eigenvalue of the three-dimensional virtual digitized terrain feature map under the guidance of the rendering style to obtain the three-dimensional virtual digitized terrain and the modulation value under the guidance of the first rendering style, and calculating the square root of the sum of the squares of each eigenvalue of the three-dimensional virtual digitized terrain feature map under the guidance of the rendering style to obtain the three-dimensional virtual digitized terrain and the modulation value under the guidance of the second rendering style;

[0091] After performing dot-wise subtraction on the rendering style-guided three-dimensional virtual digitized terrain feature map and the second rendering style-guided three-dimensional virtual digitized terrain and modulation value, the three-dimensional virtual digitized terrain feature map is dot-wise multiplied by the number of eigenvalues of the rendering style-guided three-dimensional virtual digitized terrain feature map and the reciprocal of the first rendering style-guided three-dimensional virtual digitized terrain and modulation value, and the reciprocal of each eigenvalue is taken to obtain the first rendering style-guided three-dimensional virtual digitized terrain phase transition feature map;

[0092] After performing dot-wise subtraction on the rendering style-guided three-dimensional virtual digitized terrain feature map and the first rendering style-guided three-dimensional virtual digitized terrain and modulation value, the three-dimensional virtual digitized terrain feature map is dot-wise multiplied with the square root of the number of eigenvalues of the rendering style-guided three-dimensional virtual digitized terrain feature map and the reciprocal of the second rendering style-guided three-dimensional virtual digitized terrain and modulation value, and the reciprocal of each eigenvalue is taken to obtain the second rendering style-guided three-dimensional virtual digitized terrain phase transition feature map;

[0093] The optimized 3D virtual digitized terrain feature map guided by the rendering style is obtained by point-wise subtracting the 3D virtual digitized terrain phase transition feature map guided by the first rendering style from the 3D virtual digitized terrain phase transition feature map guided by the second rendering style and the dot-wise multiplication feature map of the weighted hyperparameter.

[0094] The three-dimensional virtual digitized terrain feature map guided by the optimized rendering style is passed through a style renderer based on a diffusion model to obtain a rendered three-dimensional virtual digitized terrain.

[0095] Here, the 3D virtual digital terrain feature map guided by the rendering style The optimization expression is:

[0096]

[0097]

[0098]

[0099]

[0100]

[0101]

[0102]

[0103] in, It is a three-dimensional virtual digital terrain feature map guided by the rendering style. represents the set of real numbers, 、 、 They represent the width, length and height of the three-dimensional virtual digital terrain feature map guided by the rendering style, The first part represents the 3D virtual digital terrain feature map guided by the rendering style. The eigenvalues at the positions, Indicates the number of eigenvalues of the 3D virtual digitized terrain feature map guided by the rendering style, Indicates the three-dimensional virtual digital terrain and modulation value guided by the first rendering style, Indicates the three-dimensional virtual digitized terrain and modulation value guided by the second rendering style, Indicates point multiplication by position, is matrix multiplication, Indicates subtraction by position, Indicates the calculation of the inverse of each eigenvalue of the feature map, It represents the phase transition feature map of the three-dimensional virtual digitized terrain guided by the first rendering style. It represents the phase transition feature map of the three-dimensional virtual digitized terrain guided by the second rendering style. represents the weighted hyperparameter, Represents a three-dimensional virtual digitized terrain feature map guided by an optimized rendering style.

[0104] Accordingly, in a preferred example, the difference in the difference and modulation representation of the feature value of the three-dimensional virtual digitized terrain feature map under the guidance of the rendering style relative to the overall feature set of the three-dimensional virtual digitized terrain feature map under the guidance of the rendering style is used as the semantic change intensity information, and a phase-like transformation corresponding to the position-based intensity modulation is performed through different and modulation representations, so that the aggregation enhancement of the semantic change phase perception can improve the axial aggregation receptive field along the feature aggregation direction by performing a spatial translation operation based on alternating stacking under the set scale balance of the three-dimensional virtual digitized terrain feature map under the guidance of the rendering style, thereby improving the perception effect of the aggregation semantics of the three-dimensional virtual digitized terrain feature map under the guidance of the rendering style for the detailed semantic changes, so as to improve the expression effect of the three-dimensional virtual digitized terrain feature map under the guidance of the rendering style, and improve the rendering quality of the rendered three-dimensional virtual digitized terrain obtained by the style renderer based on the diffusion model.

[0105] Furthermore, in the technical solution of the present application, the method for rendering a combat atmosphere based on virtual digital terrain also includes a training step: training a virtual terrain feature extractor based on a three-dimensional convolutional neural network model, a feature attention selection enhancement module based on a compression-suppression structure, a cross-modal joint encoder based on a MetaNet model, and a style renderer based on a diffusion model.

[0106] The training steps include: obtaining training data, which includes training three-dimensional virtual digitized terrain, training actual combat atmosphere rendering style representation vector and real value of rendered three-dimensional virtual digitized terrain; inputting the training three-dimensional virtual digitized terrain into a virtual terrain feature extractor based on a three-dimensional convolutional neural network model to obtain a training three-dimensional virtual digitized terrain feature map; inputting the training three-dimensional virtual digitized terrain feature map into a feature attention selection enhancement module based on a compression-suppression structure to obtain a training enhanced three-dimensional virtual digitized terrain feature map; inputting the training actual combat atmosphere rendering style representation vector and the training enhanced three-dimensional virtual digitized terrain feature map into a cross-modal joint encoder based on a MetaNet model to obtain a training rendering style extractor. The method comprises the following steps: guiding a three-dimensional virtual digitized terrain feature map under training rendering style; inputting the three-dimensional virtual digitized terrain feature map under training rendering style into a style renderer based on a diffusion model to obtain a generation loss function value; calculating a predetermined loss function value of the three-dimensional virtual digitized terrain feature map under training rendering style to obtain a three-dimensional virtual digitized terrain loss function value guided by rendering style; taking the weighted sum of the generation loss function value and the three-dimensional virtual digitized terrain loss function value guided by rendering style as the loss function value, and training a virtual terrain feature extractor based on a three-dimensional convolutional neural network model, a feature attention selection enhancement module based on a compression-suppression structure, a cross-modal joint encoder based on a MetaNet model, and a style renderer based on a diffusion model.

[0107] In a preferred example, the present application further introduces a generation loss function in the model training process, such as a predetermined loss function other than the difference loss function between the actual generation result and the predicted generation result, that is, training the model through gradient back propagation based on the loss function includes the following steps: expanding the three-dimensional virtual digitized terrain feature map guided by the training rendering style into a three-dimensional virtual digitized terrain feature vector guided by the training rendering style; calculating a first training rendering style guided three-dimensional virtual digitized terrain weight matrix and a second training rendering style guided three-dimensional virtual digitized terrain weight matrix based on the training rendering style guided three-dimensional virtual digitized terrain feature vector, wherein the first and second training rendering style guided three-dimensional virtual digitized terrain weight matrices are The eigenvalues of the positions are the first and second eigenvalues of the 3D virtual digitized terrain feature vector guided by the training rendering style. Eigenvalues and The mean and half of the absolute value of the difference of the eigenvalues; perform query matrix multiplication on the eigenvalue vector of the three-dimensional virtual digitized terrain guided by the training rendering style and the weight matrix of the three-dimensional virtual digitized terrain guided by the first training rendering style and the weight matrix of the three-dimensional virtual digitized terrain guided by the second training rendering style respectively to obtain the intermediate vector of the three-dimensional virtual digitized terrain guided by the first training rendering style and the intermediate vector of the three-dimensional virtual digitized terrain guided by the second training rendering style; calculate the vector inner product of the intermediate vector of the three-dimensional virtual digitized terrain guided by the first training rendering style and the intermediate vector of the three-dimensional virtual digitized terrain guided by the second training rendering style to obtain the loss term of the three-dimensional virtual digitized terrain guided by the first training rendering style; perform matrix multiplication on the weight matrix of the three-dimensional virtual digitized terrain guided by the first training rendering style and the weight matrix of the three-dimensional virtual digitized terrain guided by the second training rendering style, and calculate the result matrix norm to obtain the three-dimensional virtual digitized terrain loss term under the guidance of the second training rendering style; the three-dimensional virtual digitized terrain loss term under the guidance of the first training rendering style is subtracted from the product of the predetermined weight hyperparameter and the three-dimensional virtual digitized terrain loss term guided by the second training rendering style to obtain the three-dimensional virtual digitized terrain loss function value guided by the rendering style; based on the weighted sum of the three-dimensional virtual digitized terrain loss function value guided by the rendering style and the generation loss function value, the model parameters are optimized by gradient back propagation.

[0108] The loss function value of the three-dimensional virtual digital terrain guided by the rendering style is specifically calculated using the following loss calculation formula, which is expressed as:

[0109]

[0110]

[0111]

[0112] in, is the loss function value of the three-dimensional virtual digital terrain guided by the rendering style, It is the feature vector of 3D virtual digitized terrain guided by training rendering style. and They are the first and second feature vectors of 3D virtual digitized terrain guided by the training rendering style. Hedi eigenvalues, It is the weight matrix of the three-dimensional virtual digitized terrain guided by the first training rendering style. is the weight matrix of the three-dimensional virtual digitized terrain guided by the second training rendering style, and The coordinates of the three-dimensional virtual digitized terrain weight matrix under the guidance of the first training rendering style and the three-dimensional virtual digitized terrain weight matrix under the guidance of the second training rendering style are respectively The eigenvalues of is a predetermined weight hyperparameter, express norm, represents matrix multiplication, Indicates taking the transpose, Indicates taking the absolute value.

[0113] That is, in the above preferred example, the rendering style-guided three-dimensional virtual digitized terrain loss function value is trained by training the query composition of the detail inner product space within the rendering style-guided three-dimensional virtual digitized terrain feature map through the structured feature representation of the short-range and long-range cross-scale detail links of the rendering style-guided three-dimensional virtual digitized terrain feature map, so as to approximate the low-rank independent observable composition of the link detail composition provided by the structured detail interaction of the training rendering style-guided three-dimensional virtual digitized terrain feature map. In this way, by training with the rendering style-guided three-dimensional virtual digitized terrain loss function value, the detail group decomposition can be performed on the basis of detail complexity by training the distributed detail group of the rendering style-guided three-dimensional virtual digitized terrain feature map, so as to promote the diffusion generation regression decomposition recognition of the complex feature structure of the training rendering style-guided three-dimensional virtual digitized terrain feature map, and improve the training efficiency of the style renderer based on the diffusion model.

[0114] Based on the above embodiments, see Figure 4The figure shows a schematic diagram of the structure of a virtual digital terrain-based combat atmosphere rendering system 800 according to an embodiment of the present application. The virtual digital terrain-based combat atmosphere rendering system 800 includes: a terrain import module 810 for importing a three-dimensional virtual digital terrain; a vector import module 820 for importing a combat atmosphere rendering style representation vector; a terrain feature extraction module 830 for extracting terrain features from the three-dimensional virtual digital terrain to obtain a three-dimensional virtual digital terrain feature map; an enhancement module 840 for inputting the three-dimensional virtual digital terrain feature map into a feature attention selection enhancement module based on a compression-suppression structure to obtain an enhanced three-dimensional virtual digital terrain feature map; a fusion module 850 for cross-modally fusing the combat atmosphere rendering style representation vector and the enhanced three-dimensional virtual digital terrain feature map to obtain a rendering style-guided three-dimensional virtual digital terrain feature map; and a generation module 860 for generating a rendered three-dimensional virtual digital terrain based on the rendering style-guided three-dimensional virtual digital terrain feature map.

[0115] Here, those skilled in the art will appreciate that the specific functions and operations of the various modules in the above-mentioned combat atmosphere rendering system 800 based on virtual digital terrain have been described in detail above. Figures 2 to 3 It has been introduced in detail in the description of the actual combat atmosphere rendering method based on virtual digitized terrain, and therefore, its repeated description will be omitted.

[0116] Figure 5 FIG is an application scenario diagram of a method for rendering a combat atmosphere based on virtual digital terrain according to an embodiment of the present application. Figure 5 As shown, in this application scenario, first, a three-dimensional virtual digitized terrain is imported (for example, Figure 5 D1 as shown in the figure) and the actual combat atmosphere rendering style representation vector (for example, Figure 5 Then, the three-dimensional virtual digitized terrain and the actual combat atmosphere rendering style representation vector are input to a server (for example, Figure 5 In S) as shown in , the server can use the actual combat atmosphere rendering algorithm based on virtual digitized terrain to process the three-dimensional virtual digitized terrain and the actual combat atmosphere rendering style representation vector to generate a rendered three-dimensional virtual digitized terrain.

[0117] As you can understand, the main terrain types to be constructed based on demand are coastal wetlands, plateau mountains, plains and hills, and snowy and frigid regions, thus forming a virtual digital terrain. The virtual maintenance training system places high demands on the virtual scene, simulating a specific area at a specific longitude and latitude in the real world for training. A terrain scene approximately 5km by 5km in size, centered at that longitude and latitude, is constructed. Considering the complex terrain, such as mountains and water bodies, conventional methods are slow to produce. Therefore, existing map data can be combined with terrain generation tools to generate elevation maps for terrain construction.

[0118] First, you need to make a scene terrain elevation map to restore terrain features, such as mountains, slopes, cliff faults, etc. After determining the longitude and latitude and after circling the size range on the map, set the output information parameters of the terrain in the terrain generation tool and confirm the output terrain. The tool will make a terrain elevation map based on the terrain information of the circled range. After the production is completed, it will form terrain data that can be used by the Unity3D engine. After importing the Unity3D engine, you can see the corresponding terrain white model. In the Unity3D engine, use the terrain tool to make simple adjustments to complete the production of the complete terrain white model for the area. After completing the production of the terrain white model, use the model map to make the terrain map material based on the generated landform picture, and combine the material and the white model terrain to obtain the final terrain effect. The terrain scene is not just the terrain, but also requires ground buildings, trees and other landforms, which are also indispensable and important components. Ground objects, numbers, etc. are produced through three-dimensional model construction technology. After production is completed, they are laid out according to specific positions. Among them, the terrain white model is such as Figure 6 shown.

[0119] The complex and unique environment is composed of three major components: 3D virtual digitized terrain, virtual weather conditions, and realistic combat atmosphere rendering. These three components combine to form a complete and vibrant virtual combat environment and situation. This allows for virtual combat to incorporate terrain conditions, weather conditions, different combat objectives and target conditions, and firepower deployment and strike conditions in diverse geographical environments (coastal wetlands, mountainous plateaus, hilly plains, and snowy and cold regions). These three components can be freely configured and combined to create a variety of variable combat situations, lending the virtual combat's variability and simulation capabilities vitality.

[0120] The construction of complex and special environments includes the construction of complex terrain and the rendering of environmental effects. Complex terrain is to produce scenes with landform characteristics of designated areas such as mountains, rivers, vegetation, etc. according to requirements. Environmental effects are mainly to render the effects of the scenes, including the atmosphere rendering of actual combat atmosphere, rendering of weather effects such as sunshine, snow, rain, fog, and rendering of lights, etc.

[0121] Regarding virtual environment rendering, virtual scenes require the combination of terrain and environment to present realistic effects and achieve the most realistic immersive feeling. To complete the construction of complex terrain, it is necessary to render the overall environment effect. Environment rendering includes scene lighting rendering and weather effect rendering.

[0122] Lighting illuminates and creates a scene's atmosphere. Its setup is crucial to creating a scene's ambiance. Without lighting, there would be no lightmaps, no shadows, no distinct regions, weather conditions, or combat atmosphere, and the entire scene would lack a sense of three-dimensionality. In a three-dimensional scene, lighting serves more than just illuminating objects; it also conveys additional information to trainees through its effects. To achieve a realistic final scene, many different lights are required. In the real world, light sources vary, such as sunlight, candlelight, and fluorescent lights, and objects appear differently under different light sources. In the Unity3D engine, four common light sources are directional, point, spot, and area.

[0123] Among them, parallel light usually exists as the main light source of the scene, used to simulate the effect of sunlight. Parallel light can set general properties such as color, rendering mode, and light intensity. It can also set the type of shadow of objects affected by the light. Point light source is similar to a spherical light source. It only affects the lighting effect of objects within the scene range and has little effect on peripheral lighting. It is usually used to create effects such as street lights and light boxes. In the virtual maintenance training system, all indicator lights of a certain vehicle are implemented using point light sources. For example, the indicator lights on the control console are implemented using point light sources to simulate the effect of indicator lights. The spotlight effect is similar to the spotlight effect. The illumination range is conical and can be used to simulate the light effect of a flashlight. The spotlight property panel can be set to adjust the lighting range, light color, light intensity, etc. One of the disassembly and assembly scenes of a certain type of vehicle components in the virtual maintenance training system simulates an indoor environment and uses a spotlight for rendering. Range lights require baking the scene to take effect. Baking is the process of rendering static objects and static light sources in the scene in the form of baked maps. Since real-time lighting consumes a lot of hardware performance, rendering can save hardware performance consumption and greatly improve the smoothness of the camera's rendered image. In addition to adding common types of lights to scene lighting effects, you can also use environment reflection balls, light searchlights, etc. to set special lighting effects. In addition to the light source, the light and shadow effects produced by objects are also greatly affected by the object's own shader, material, etc. For example, in the shader, you can set whether to receive light illumination, whether to reflect light sources, and other conditions. Therefore, rendering complex scene lighting effects requires the coordinated use of the light source itself and the object shader.

[0124] Regarding weather simulation, weather simulation includes simulation of weather effects such as sunshine, snow, rain, fog, and simulation of day and night rotation. It also includes the time passage function. The passage of time will affect the changes in weather effects and light and shadow effects.

[0125] Weather simulations, such as sunny, snowy, rainy, and foggy, are primarily created using shaders and particle effects within the Unity3D engine. First, different shaders are selected based on the desired weather. Parameters such as light position, light reception, cloud thickness, color, and density are set. These attributes are visualized within the Unity3D engine's component palette. Weather effects can be previewed by adjusting the corresponding parameter ranges within the palette. Parameter attribute values can also be controlled through code, allowing for dynamic simulation of various weather conditions.

[0126] Special weather conditions such as rain and snow need to be achieved through particle effects. To create snowy weather effects, you need to use a particle system, which controls the occurrence and evolution of various particle effects, as well as setting various snowflake parameters, including the duration of the snowflake, whether it appears in a loop, the range of snowflakes, the speed of snowflakes, and other properties. The style of the snowflakes is rendered by generating a material ball through the material map of the snowflake shader. The snow effect on the ground is also controlled by modifying the parameters of the ground shader. In snowy weather, the sky appears overcast, and snowflakes can be seen falling in the field of vision. At the same time, objects such as the ground will gradually have the effect of snow turning white.

[0127] In rainy weather, the sky becomes overcast, and raindrops can be seen falling. At the same time, the ground gradually becomes wet or stagnant. In foggy weather, there will be bursts of fog in the scene, and the distant visual effect will appear gray and hazy, and the field of vision will be reduced. In addition to the four weather types of clear, snowy, rainy, and foggy, we have also created sandstorms, thunderstorms, cloudy, overcast, sunny and rainy weather effects, each with its own distinct characteristics.

[0128] Furthermore, the weather management panel allows for selective switching, with transitional buffering and preview effects when switching from one weather type to another. You can also adjust the volume of ambient sound effects. Weather buffering is handled by the weather management module. For example, when switching from cloudy to clear weather, the clouds will gradually dissipate. When switching from rainy or snowy weather to clear weather, the rain and snow will gradually disappear, and accumulated water and snow will also gradually disappear, transforming into a clear ground effect.

[0129] Regarding day and night simulation, the day and night function simulates the passage of time. The time can be set to real time or virtual time. Real time is synchronized with the real environment time. Time passes slowly, and the changes in the scene are not obvious. Virtual time passes quickly, and the changes in scene effects can be clearly perceived. Changes include the position and brightness of the sun during the day and night, the position and phase of the moon at night, and the changes in the stars. You can quickly control the passage of virtual time by dragging the time slider in the weather switch panel to more intuitively experience the various changes. This effect is achieved by controlling the rotation of the sky box of the entire scene to show the switching of the sun and moon. The sky box operates through a set of time systems, and the brightness and darkness of the lights rely on the timeline. Therefore, controlling the changes in time can control the changes in day and night. The phases of the moon also change with the flow of time, but their changes are handled by a separate set of rules.

[0130] Virtual combat atmosphere rendering focuses on the development of vehicles and objects in three-dimensional virtual combat, such as various weapons, targets, military facilities and fortifications, and personnel roles. Virtual combat atmosphere primarily involves the development of various special effects related to real combat, such as attack explosion effects, damage status, combat sound effects, bombing, and artillery attacks. These effects are independently and set to be loaded into the virtual combat environment to create a realistic and brutal war environment, allowing trainees to gain a true and immersive combat experience in the virtual combat environment. Trainees can feel the pressure of actual combat in the virtual combat environment, overcome their fear of actual combat, and complete various tasks at the same time. This provides targeted psychological training for personnel, cultivating their courage to face danger, training their ability to adapt to intense, cruel, and dangerous situations, building their confidence in defeating the enemy, and improving their combat capabilities.

[0131] The creation of a realistic combat atmosphere is primarily based on sound, light, and electricity. Various models and special effects that embody real-world combat characteristics are added for rendering. These models include equipment, vehicle wreckage, and artillery shells. Special effects are created using the Unity3D engine's particle system, including explosion and combustion effects, shell trails, and smoke effects. 3D sound effects are used to simulate the sounds of explosions, flying shells, and bullets firing, moving from far to near. Sound, light, and electricity effects are typically used as static assets. To create a realistic and vivid combat atmosphere, they must be combined and their timing controlled. The Unity3D engine's Timeline component allows for the effective integration and control of these assets. A Timeline is like a timeline with multiple tracks, and different assets are controlled on different tracks. When different tracks reach the same time segment, the assets on each track are controlled to display their corresponding effects. By integrating and utilizing these combat assets using Timeline, a realistic combat environment can be simulated.

[0132] Based on the above embodiment, the present application also provides another exemplary embodiment of an electronic device. In some possible implementations, the electronic device in the present application may include a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, can implement the steps of the method for rendering a combat atmosphere based on virtual digitized terrain in the above embodiment.

[0133] For example, in the case of electronic equipment Figure 1 Taking the server 100 in the example for explanation, the processor in the electronic device is the processor 110 in the server 100, and the memory in the electronic device is the memory 120 in the server 100.

[0134] Embodiments of the present application also provide a computer-readable storage medium having computer-executable instructions stored thereon. When the computer-executable instructions are executed by a processor, the method for rendering a combat atmosphere based on virtual digitized terrain according to the embodiments of the present application described with reference to the above figures can be executed. The computer-readable storage medium includes, but is not limited to, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and / or cache memory. Non-volatile memory may include, for example, read-only memory (ROM), a hard disk, flash memory, etc.

[0135] Embodiments of the present application also provide a computer program product or computer program, which includes computer-executable instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer-executable instructions from the computer-readable storage medium and executes the computer-executable instructions, causing the computer device to perform the method for rendering a combat atmosphere based on virtual digitized terrain according to an embodiment of the present application.

[0136] Those skilled in the art will appreciate that the contents disclosed in this application may be subject to various modifications and improvements. For example, the various devices or components described above may be implemented through hardware, software, firmware, or a combination of some or all of the three.

[0137] Furthermore, although the present application makes various references to certain units in the system according to embodiments of the present application, any number of different units may be used and run on the client and / or server. The units are illustrative only, and different aspects of the system and method may use different units.

[0138] Those skilled in the art will appreciate that all or part of the steps in the above method can be performed by instructing the relevant hardware through a program, and the program can be stored in a computer-readable storage medium, such as a read-only memory, a magnetic disk, or an optical disk. Alternatively, all or part of the steps in the above embodiment can also be implemented using one or more integrated circuits. Accordingly, each module / unit in the above embodiment can be implemented in the form of hardware or in the form of software functional modules. This application is not limited to any particular form of combination of hardware and software.

[0139] Unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. It should also be understood that terms such as those defined in common dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology and should not be interpreted in an idealized or highly formal sense, unless expressly defined as such herein.

[0140] The above is an explanation of the present application and should not be considered as limiting thereof. Although several exemplary embodiments of the present application have been described, those skilled in the art will readily appreciate that many modifications may be made to the exemplary embodiments without departing from the novel teachings and advantages of the present application.

Claims

1. A method for rendering combat atmosphere based on virtual digital terrain, characterized in that: include: Import three-dimensional virtual digital terrain; Import actual combat atmosphere rendering style representation vector; Extracting terrain features from the three-dimensional virtual digitized terrain to obtain a three-dimensional virtual digitized terrain feature map; Inputting the three-dimensional virtual digitized terrain feature map into a feature attention selection and enhancement module based on a compression-suppression structure to obtain an enhanced three-dimensional virtual digitized terrain feature map; The process of obtaining the enhanced three-dimensional virtual digitized terrain feature map includes: calculating the global mean of each feature matrix along the channel dimension of the three-dimensional virtual digitized terrain feature map to obtain a three-dimensional virtual digitized terrain feature compression information representation vector; performing one-dimensional convolution coding on the three-dimensional virtual digitized terrain feature compression representation vector to obtain a three-dimensional virtual digitized terrain feature compression information inter-correlation representation feature vector; cascading the three-dimensional virtual digitized terrain feature compression information representation vector and the three-dimensional virtual digitized terrain feature compression information inter-correlation representation feature vector to obtain a three-dimensional virtual digitized terrain feature compression information multi-scale representation vector; inputting the three-dimensional virtual digitized terrain feature compression information multi-scale representation vector into a compression information feature extraction module including a multi-layer perceptron and a SiLU activation function to obtain a three-dimensional virtual digitized terrain feature compression information multi-scale correlation feature vector; performing a normalization operation on the three-dimensional virtual digitized terrain feature compression information multi-scale correlation feature vector to obtain a three-dimensional virtual digitized terrain weight feature vector; and performing feature amplification and suppression operations on the three-dimensional virtual digitized terrain feature map based on the three-dimensional virtual digitized terrain weight feature vector to obtain the enhanced three-dimensional virtual digitized terrain feature map. Cross-modally fusing the actual combat atmosphere rendering style representation vector and the enhanced three-dimensional virtual digitized terrain feature map to obtain a three-dimensional virtual digitized terrain feature map guided by the rendering style; Based on the three-dimensional virtual digitized terrain feature map guided by the rendering style, a rendered three-dimensional virtual digitized terrain is generated; the process of generating the rendered three-dimensional virtual digitized terrain includes: inputting the three-dimensional virtual digitized terrain feature map guided by the rendering style into a style renderer based on a diffusion model to obtain the rendered three-dimensional virtual digitized terrain.

2. The method for rendering combat atmosphere based on virtual digital terrain according to claim 1, characterized in that: The extracting terrain features of the three-dimensional virtual digitized terrain to obtain a three-dimensional virtual digitized terrain feature map includes: The three-dimensional virtual digitized terrain is input into a virtual terrain feature extractor based on a three-dimensional convolutional neural network model to obtain the three-dimensional virtual digitized terrain feature map.

3. The method for rendering combat atmosphere based on virtual digital terrain according to claim 1, characterized in that: The normalizing operation of the multi-scale correlation feature vector of the three-dimensional virtual digitized terrain feature compression information to obtain the three-dimensional virtual digitized terrain weight feature vector includes: A Sigmoid function is used to perform a normalization operation on the multi-scale correlation feature vector of the three-dimensional virtual digitized terrain feature compression information to obtain the three-dimensional virtual digitized terrain weight feature vector.

4. The method for rendering combat atmosphere based on virtual digital terrain according to claim 3, characterized in that: The step of performing feature amplification and suppression operations on the three-dimensional virtual digitized terrain feature map based on the three-dimensional virtual digitized terrain weight feature vector to obtain the enhanced three-dimensional virtual digitized terrain feature map includes: The Kronecker product of the three-dimensional virtual digitized terrain weight feature vector and each feature matrix along the channel dimension of the three-dimensional virtual digitized terrain feature map is calculated to obtain the enhanced three-dimensional virtual digitized terrain feature map.

5. The method for rendering combat atmosphere based on virtual digital terrain according to claim 4, characterized in that: The cross-modal fusion of the actual combat atmosphere rendering style representation vector and the enhanced three-dimensional virtual digitized terrain feature map to obtain the three-dimensional virtual digitized terrain feature map guided by the rendering style includes: The actual combat atmosphere rendering style representation vector and the enhanced three-dimensional virtual digitized terrain feature map are input into a cross-modal joint encoder based on the MetaNet model to obtain the three-dimensional virtual digitized terrain feature map guided by the rendering style.

6. The method for rendering combat atmosphere based on virtual digital terrain according to claim 5, characterized in that: The method further comprises: The virtual terrain feature extractor based on the three-dimensional convolutional neural network model, the feature attention selection enhancement module based on the compression-suppression structure, the cross-modal joint encoder based on the MetaNet model, and the style renderer based on the diffusion model are trained.

7. The method for rendering combat atmosphere based on virtual digital terrain according to claim 6, characterized in that: The training of a virtual terrain feature extractor based on a three-dimensional convolutional neural network model, a feature attention selection and enhancement module based on a compression-suppression structure, a cross-modal joint encoder based on a MetaNet model, and a style renderer based on a diffusion model includes: Acquire training data, the training data including a training three-dimensional virtual digitized terrain, a training actual combat atmosphere rendering style representation vector, and a real value of the rendered three-dimensional virtual digitized terrain; Inputting the training three-dimensional virtual digitized terrain into the virtual terrain feature extractor based on the three-dimensional convolutional neural network model to obtain a training three-dimensional virtual digitized terrain feature map; Inputting the training three-dimensional virtual digitized terrain feature map into the feature attention selection and enhancement module based on the compression-suppression structure to obtain a training enhanced three-dimensional virtual digitized terrain feature map; Inputting the training combat atmosphere rendering style representation vector and the training enhanced three-dimensional virtual digitized terrain feature map into the cross-modal joint encoder based on the MetaNet model to obtain a three-dimensional virtual digitized terrain feature map guided by the training rendering style; Inputting the three-dimensional virtual digitized terrain feature map guided by the training rendering style into a style renderer based on a diffusion model to obtain a generation loss function value; Calculating a predetermined loss function value of the training rendering style-guided three-dimensional virtual digitized terrain feature map to obtain a rendering style-guided three-dimensional virtual digitized terrain loss function value; The weighted sum of the generation loss function value and the rendering style-guided three-dimensional virtual digitized terrain loss function value is used as the loss function value to train the virtual terrain feature extractor based on the three-dimensional convolutional neural network model, the feature attention selection and enhancement module based on the compression-suppression structure, the cross-modal joint encoder based on the MetaNet model, and the style renderer based on the diffusion model.

Images