Actual atmosphere rendering method based on virtual digital terrain

Abstract

The invention relates to the field of virtual digital terrains, in particular to an actual combat atmosphere rendering method based on the virtual digital terrains. According to the method, spatial distribution characteristics of a three-dimensional virtual digital terrain are automatically extracted and strengthened by using an intelligent algorithm based on deep learning, and cross-modal joint optimization is performed on the spatial distribution characteristics of the three-dimensional virtual digital terrain by taking actual atmosphere rendering style representation vectors as constraint and guide information; according to the method, the three-dimensional virtual digital topographic features guided by the rendering style are obtained, and the style renderer is constructed by using the diffusion model, so that the three-dimensional virtual digital topography is automatically generated and rendered based on the three-dimensional virtual digital topographic features guided by the rendering style, and the visual effect of the digital topography is enhanced.

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. In order to improve the realism and immersion of these applications, it is necessary to render the digital terrain in a combat atmosphere to make it more in line with the needs of actual scenarios. Specifically, combat atmosphere rendering refers to simulating visual effects and atmospheres close to the real battlefield environment through technical means to enhance training effects or provide a more realistic experience.

[0003] In traditional rendering methods, manual design and adjustment are usually relied on, but this method is highly dependent on the operator's technical level, knowledge level and experience, and has problems such as low efficiency and long time consumption. Therefore, an optimized combat atmosphere rendering method based on virtual digital terrain is expected. 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 real combat atmosphere based on a 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 real combat atmosphere based on a virtual digital terrain, which includes: Import three-dimensional virtual digital terrain; Import the 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; 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.

[0006] Further, according to the above-mentioned method for rendering combat atmosphere based on virtual digital terrain, wherein the extraction of terrain features of the three-dimensional virtual digital terrain to obtain a three-dimensional virtual digital 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.

[0007] Further, according to the above-mentioned actual combat atmosphere rendering method based on virtual digital terrain, wherein, the three-dimensional virtual digital terrain feature map is input into a feature attention selection enhancement module based on a compression-suppression structure to obtain an enhanced three-dimensional virtual digital terrain feature map, including: 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 feature vector representing the association between the three-dimensional virtual digitized terrain feature compression information; 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; 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; Normalizing the multi-scale associated feature vector of the three-dimensional virtual digitized terrain feature compression information to obtain a three-dimensional virtual digitized terrain weight feature vector; 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.

[0008] Further, according to the above-mentioned actual combat atmosphere rendering method based on virtual digital terrain, wherein, the multi-scale correlation feature vector of the three-dimensional virtual digital terrain feature compression information is normalized to obtain the three-dimensional virtual digital terrain weight feature vector, including: The Sigmoid function is used to perform a normalization operation on the multi-scale associated feature vector of the three-dimensional virtual digitized terrain feature compression information to obtain the three-dimensional virtual digitized terrain weight feature vector.

[0009] Further, according to the above-mentioned method for rendering a combat atmosphere based on virtual digital terrain, based on the three-dimensional virtual digital terrain weight feature vector, the three-dimensional virtual digital terrain feature map is subjected to feature amplification and suppression operations to obtain the enhanced three-dimensional virtual digital terrain feature map, including: The Kronecker product of the weighted feature vector of the three-dimensional virtual digitized terrain and each feature matrix of the three-dimensional virtual digitized terrain feature map along the channel dimension is calculated to obtain the enhanced three-dimensional virtual digitized terrain feature map.

[0010] Further, according to the above-mentioned actual combat atmosphere rendering method based on virtual digital terrain, the actual combat atmosphere rendering style representation vector and the enhanced three-dimensional virtual digital terrain feature map are cross-modally fused to obtain a three-dimensional virtual digital terrain feature map guided by the rendering style, including: 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.

[0011] Further, according to the above-mentioned actual combat atmosphere rendering method 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: 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.

[0012] Furthermore, according to the above-mentioned method for rendering combat atmosphere based on virtual digital terrain, the method further includes: 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.

[0013] Further, according to the above-mentioned actual combat atmosphere rendering method based on virtual digital terrain, 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, including: Acquire training data, wherein the training data includes a training three-dimensional virtual digitized terrain, a training actual combat atmosphere rendering style representation vector, and a real value of rendering the 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-inhibition structure to obtain the 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 the cross-modal joint encoder based on the MetaNet model to obtain the 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 three-dimensional virtual digitized terrain feature map guided by the training rendering style to obtain a loss function value of the three-dimensional virtual digitized terrain guided by the rendering style; The weighted sum of the generated loss function value and the three-dimensional virtual digitized terrain loss function value guided by the rendering style is used as the loss function value, and 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.

[0014] According to the actual combat atmosphere rendering method based on virtual digital terrain in the embodiment of the present application, it 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 actual combat atmosphere rendering style representation vector as constraint and guidance information to perform cross-modal joint optimization on the spatial distribution characteristics of the three-dimensional virtual digital terrain, thereby obtaining the three-dimensional virtual digital terrain characteristics under the guidance of 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 under the guidance of the rendering style. This rendering method is highly efficient and can enhance the visual effect of the digital terrain. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings of the embodiments of the present application are briefly introduced below. Obviously, the drawings described below only relate to some embodiments of the present application, and are not intended to limit the present application.

[0016] 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; 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; Figure 3 A flowchart of sub-step S540 of the actual combat atmosphere rendering method based on virtual digital terrain in an embodiment of the present application is shown; Figure 4A schematic diagram of the structure of a combat atmosphere rendering system based on virtual digital terrain in an embodiment of the present application is shown; Figure 5 An application scenario diagram of a method for rendering a combat atmosphere based on virtual digital terrain in an embodiment of the present application is shown; Figure 6 A terrain white model according to an embodiment of the present application is shown. DETAILED DESCRIPTION

[0017] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments 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 creative work also fall within the scope of protection of the present application.

[0018] The terms used in this specification are those common terms currently widely used in the art 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 this case, 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 a general description based on the meaning of the terms and the present application.

[0019] 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.

[0020] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps may be processed in reverse order or simultaneously as required. At the same time, other operations may also be added to these processes, or a certain step or several steps of operations may be removed from these processes.

[0021] 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.

[0022] The terminal device 200 and the server 100 can be connected via the Internet to achieve mutual communication. Optionally, the above-mentioned Internet uses standard communication technology and / or protocol. The Internet is usually the Internet, but it can also be any network, including but not limited to any combination of 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 a virtual private network. In some embodiments, the data exchanged through the network can be represented by technologies and / or formats including Hyper Text Markup Language (HTML), Extensible Markup Language (XML), etc. In addition, conventional encryption technologies such as Secure Socket Layer (SSL), Transport Layer Security (TLS), Virtual Private Network (VPN), Internet Protocol Security (IPsec), etc. can be used to encrypt all or some links. In other embodiments, customized and / or dedicated data communication technologies can also be used to replace or supplement the above-mentioned data communication technologies.

[0023] The server 100 can provide various network services for the terminal device 200, wherein the server 100 can be a single server, a server cluster consisting of several servers, or a cloud computing center. Specifically, the server 100 may include a processor 110 (Center Processing Unit, CPU), a memory 120, an input device 130, and an output device 140, etc. The input device 130 may include a keyboard, a mouse, a touch screen, etc., and the output device 140 may include a display device, such as a liquid crystal display (Liquid Crystal Display, LCD), a cathode ray tube (Cathode Ray Tube, CRT), etc.

[0024] The memory 120 may include a read-only memory (ROM) and a random access memory (RAM), and provides the processor 110 with program instructions and data stored in the memory 120. In the 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 the embodiment of the present application.

[0025] 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 actual combat atmosphere based on virtual digitized terrain in the embodiments of the present application according to the obtained program instructions.

[0026] 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.

[0027] The following is a 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 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 the present application.

[0028] In response to the above technical problems, the technical concept of the present 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 to 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, thereby obtaining the three-dimensional virtual digitized terrain characteristics under the guidance of the rendering style, and to 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 under the guidance of the rendering style.

[0029] Based on this, Figure 2 The flowchart of the actual combat atmosphere rendering method based on virtual digital terrain in the embodiment of the present application is shown. For example, the actual combat atmosphere rendering method 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 actual combat atmosphere rendering method based on virtual digital terrain according to an embodiment of the present application includes the steps of: S510, importing three-dimensional virtual digital terrain; S520, import actual combat atmosphere rendering style representation vector; S530, extracting terrain features from the three-dimensional virtual digitized terrain to obtain a three-dimensional virtual digitized terrain feature map; 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; S550, cross-modally fusing the actual combat atmosphere rendering style representation vector and the enhanced three-dimensional virtual digital terrain feature map to obtain a three-dimensional virtual digital terrain feature map guided by the rendering style; S560: Generate a rendered three-dimensional virtual digitized terrain based on the three-dimensional virtual digitized terrain feature map guided by the rendering style.

[0030] 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 express the terrain's elevation, slope, vegetation coverage 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 result meets the needs 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.

[0031] Then, the 3D virtual digitized terrain is input into the virtual terrain feature extractor based on the 3D convolutional neural network model to obtain the 3D virtual digitized terrain feature map. Among them, the 3D convolutional neural network (3D CNN) is particularly good at capturing implicit features in spatial data. In terrain data, these features may include terrain undulations, slopes, and terrain roughness.

[0032] 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 a three-dimensional virtual digitized terrain feature map.

[0033] 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 a higher weight to the important information in the feature distribution, so that the important terrain features are reused multiple times, thereby suppressing the flow of irrelevant background feature information. More specifically, the feature attention selection and enhancement module based on the compression-suppression structure uses the global average pooling (Avg-Pool) to compress the information of the three-dimensional virtual digitized terrain feature map in view of the idea of ​​the Squeeze and Excitation (SE) structure, and uses one-dimensional convolutional coding to capture the correlation information between the compressed information, and then uses a multi-layer perceptron (Multi-IayerPerception, MLP) and SiLU activation function to extract features, and then uses the Sigmoid function to obtain the weight values ​​of different channels after normalization, and finally combines the weight values ​​to perform feature amplification and suppression operations on the three-dimensional virtual digitized terrain feature map, so as to improve the network's ability to select three-dimensional virtual digitized terrain features.

[0034] Accordingly, in step S540, if Figure 3 As shown, 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: 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; S542, 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 correlation representation feature vector; S543, 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 the three-dimensional virtual digitized terrain feature compression information multi-scale representation vector; S544, 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; S545, performing a normalization operation on the multi-scale associated feature vector of the three-dimensional virtual digitized terrain feature compression information to obtain a three-dimensional virtual digitized terrain weight feature vector; 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.

[0035] Among them, in step S545, the multi-scale associated feature vector of the three-dimensional virtual digitized terrain feature compression information is normalized to obtain the three-dimensional virtual digitized terrain weight feature vector, including: using a Sigmoid function to normalize the multi-scale associated feature vector of the three-dimensional virtual digitized terrain feature compression information to obtain the three-dimensional virtual digitized terrain weight feature vector.

[0036] Among them, 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.

[0037] 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-inhibition 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-inhibition structure to process the three-dimensional virtual digitized terrain feature map with the following enhancement formula to obtain an enhanced three-dimensional virtual digitized terrain feature map; wherein the enhancement formula is:

[0038]

[0039]

[0040] in, It is a three-dimensional virtual digital terrain feature map. The coordinates in the channel are The characteristic value of and are the height and width of the three-dimensional virtual digital terrain feature map, It is the compressed information representation vector of the three-dimensional virtual digital terrain features. eigenvalues, It is a vector representing the compressed information of three-dimensional virtual digitized terrain features. represents one-dimensional convolutional coding, Indicates cascade, It is a multi-scale representation vector of compressed information of three-dimensional virtual digitized terrain features. represents a multi-layer 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.

[0041] Next, the actual combat atmosphere rendering style representation vector and the enhanced three-dimensional virtual digital terrain feature map are input into the cross-modal joint encoder based on the MetaNet model to obtain the three-dimensional virtual digital terrain feature map guided by the rendering style. That is, the style features of the actual combat atmosphere can be combined with the three-dimensional virtual terrain features through the cross-modal joint encoder based on the MetaNet model to form a terrain feature distribution with the actual combat atmosphere rendering style as a constraint, providing richer input for rendering and helping the system understand the rendering intention and expected effect.

[0042] 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.

[0043] Then, the rendering style-guided 3D virtual digitized terrain feature map is input into the style renderer based on the diffusion model to obtain the rendered 3D virtual digitized terrain. In this way, the diffusion model can apply a specific visual style to the 3D virtual digitized terrain according to the given input, and realize automatic rendering.

[0044] Accordingly, in step S560, based on the three-dimensional virtual digitized terrain feature map guided by the rendering style, a rendered three-dimensional virtual digitized terrain is generated, including: 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 a rendered three-dimensional virtual digitized terrain.

[0045] In a preferred example, the enhanced three-dimensional virtual digitized terrain feature map expresses 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 obtained three-dimensional virtual digitized terrain feature map 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 under the guidance of the rendering style.

[0046] In the preferred embodiment, 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 a rendered three-dimensional virtual digitized terrain, including: Calculating the sum of the absolute values ​​of each characteristic value 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 characteristic value 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; After performing dot-wise subtraction on the three-dimensional virtual digitized terrain feature map guided by the rendering style and the three-dimensional virtual digitized terrain and modulation value guided by the second rendering style, dot-wise multiplication is performed with the number of eigenvalues ​​of the three-dimensional virtual digitized terrain feature map guided by the rendering style and the reciprocal of the three-dimensional virtual digitized terrain and modulation value guided by the first rendering style, and the reciprocal of each eigenvalue is taken to obtain the phase conversion feature map of the three-dimensional virtual digitized terrain guided by the first rendering style; After performing dot-wise subtraction on the three-dimensional virtual digitized terrain feature map guided by the rendering style and the three-dimensional virtual digitized terrain and the modulation value guided by the first rendering style, dot-wise multiplication is performed with the square root of the number of eigenvalues ​​of the three-dimensional virtual digitized terrain feature map guided by the rendering style and the reciprocal of the three-dimensional virtual digitized terrain and the modulation value guided by the second rendering style, and the reciprocal of each eigenvalue is taken to obtain the phase conversion feature map of the three-dimensional virtual digitized terrain guided by the second rendering style; The first rendering style guided three-dimensional virtual digitized terrain phase conversion feature map is point-subtracted from the second rendering style guided three-dimensional virtual digitized terrain phase conversion feature map and the dot-multiplied feature map of the weighted hyperparameter to obtain an optimized rendering style guided three-dimensional virtual digitized terrain feature map; 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.

[0047] Here, the 3D virtual digital terrain feature map guided by the rendering style The optimization expression is:

[0048]

[0049]

[0050]

[0051]

[0052]

[0053]

[0054] 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 3D virtual digital terrain feature map guided by the rendering style. The first part represents the 3D virtual digitized terrain feature map guided by rendering style. The eigenvalues ​​at the positions, It represents the number of feature values ​​of the 3D virtual digitized terrain feature map guided by the rendering style. Indicates the three-dimensional virtual digitized 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, It means point multiplication by position. is matrix multiplication, It means subtracting 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.

[0055] 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 representation forms, 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 semantic changes of details, 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.

[0056] 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.

[0057] The training step includes: obtaining training data, the training data including training three-dimensional virtual digitized terrain, training actual combat atmosphere rendering style representation vector and real value of rendering 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-inhibition 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 invention relates to a method for extracting a three-dimensional virtual digitized terrain feature map guided by a training rendering style; inputting the three-dimensional virtual digitized terrain feature map guided by a 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 guided by a training rendering style to obtain a loss function value of the three-dimensional virtual digitized terrain guided by the rendering style; taking the weighted sum of the generation loss function value and the loss function value of the three-dimensional virtual digitized terrain guided by the 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 reinforcement 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.

[0058] 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 based on the loss function through gradient back propagation 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 three-dimensional virtual digitized terrain feature vector guided by the training rendering style, wherein the first and second training rendering style guided three-dimensional virtual digitized terrain weight matrices are The feature values ​​of the positions are the first and second feature vectors of the 3D virtual digitized terrain guided by the training rendering style. Eigenvalues ​​and The mean and half of the absolute value of the difference of the eigenvalues; perform query-type matrix multiplication of the 3D virtual digitized terrain feature vector guided by the training rendering style with the 3D virtual digitized terrain weight matrix guided by the first training rendering style and the 3D virtual digitized terrain weight matrix guided by the second training rendering style to obtain the intermediate vector of the 3D virtual digitized terrain guided by the first training rendering style and the intermediate vector of the 3D virtual digitized terrain guided by the second training rendering style; calculate the vector inner product of the intermediate vector of the 3D virtual digitized terrain guided by the first training rendering style and the intermediate vector of the 3D virtual digitized terrain guided by the second training rendering style to obtain the loss term of the 3D virtual digitized terrain guided by the first training rendering style; perform matrix multiplication of the 3D virtual digitized terrain weight matrix guided by the first training rendering style and the 3D virtual digitized terrain weight matrix 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 under the guidance of the second training rendering style to obtain the three-dimensional virtual digitized terrain loss function value under the guidance of the rendering style; based on the weighted sum of the three-dimensional virtual digitized terrain loss function value under the guidance of the rendering style and the generation loss function value, the model parameters are optimized through gradient back propagation.

[0059] Among them, 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:

[0060]

[0061]

[0062] in, is the loss function value of the 3D virtual digitized terrain guided by the rendering style, It is the feature vector of 3D virtual digitized terrain guided by the training rendering style. and They are the first and second feature vectors of 3D virtual digitized terrain guided by the training rendering style. and eigenvalues, It is the first training rendering style guided 3D virtual digitized terrain weight matrix. is the weight matrix of the three-dimensional virtual digitized terrain guided by the second training rendering style, and They are 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. The characteristic value of is a predetermined weight hyperparameter, express norm, represents matrix multiplication, It means taking the transpose, Indicates taking the absolute value.

[0063] That is, in the above preferred example, the loss function value of the three-dimensional virtual digitized terrain guided by the rendering style is trained by training the structured feature representation of the short-range and long-range cross-scale detail links of the three-dimensional virtual digitized terrain feature map guided by the rendering style to train the query composition of the detail inner product space within the three-dimensional virtual digitized terrain feature map guided by the rendering style, so as to approximate the low-rank independent observable composition of the linked detail composition provided by the structured detail interaction of the three-dimensional virtual digitized terrain feature map guided by the training style. In this way, by training with the loss function value of the three-dimensional virtual digitized terrain guided by the rendering style, the distributed detail group of the three-dimensional virtual digitized terrain feature map guided by the rendering style can be trained to perform detail group decomposition on the basis of detail complexity, so as to promote the diffusion generation regression decomposition recognition of the complex feature structure of the three-dimensional virtual digitized terrain feature map guided by the training rendering style, and improve the training efficiency of the style renderer based on the diffusion model.

[0064] Based on the above embodiments, see Figure 4 As shown, it is a structural schematic diagram of a real combat atmosphere rendering system 800 based on virtual digital terrain in an embodiment of the present application. The real combat atmosphere rendering system 800 based on virtual digital terrain includes: a terrain import module 810, which is used to import a three-dimensional virtual digital terrain; a vector import module 820, which is used to import a real combat atmosphere rendering style representation vector; a terrain feature extraction module 830, which is used to extract terrain features from the three-dimensional virtual digital terrain to obtain a three-dimensional virtual digital terrain feature map; an enhancement module 840, which is used to input 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, which is used to cross-modally fuse the real combat atmosphere rendering style representation vector and the enhanced three-dimensional virtual digital terrain feature map to obtain a three-dimensional virtual digital terrain feature map under the guidance of a rendering style; a generation module 860, which is used to generate a rendered three-dimensional virtual digital terrain based on the three-dimensional virtual digital terrain feature map under the guidance of a rendering style.

[0065] Here, those skilled in the art can understand that the specific functions and operations of each module in the above-mentioned actual combat atmosphere rendering system 800 based on virtual digital terrain have been described in the above reference. Figure 2 to Figure 3 The description of the actual combat atmosphere rendering method based on virtual digitized terrain has been introduced in detail, and therefore, its repeated description will be omitted.

[0066] Figure 5 FIG. 1 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 (e.g., 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 the 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.

[0067] It should be understood that according to the needs, coastal wetlands, plateau mountains, plains and hills, snowy and cold terrains are mainly constructed to form virtual digital terrain. The virtual maintenance training system has high requirements for virtual scenes. It is necessary to simulate a certain area at a specific longitude and latitude in the real world for training, and build a terrain scene with a size of about 5km*5km with the longitude and latitude as the center point. Considering the complex environment of terrain mountains, waters, etc., the production progress is slow through conventional methods. Therefore, it can be combined with existing map data and constructed by generating elevation map terrain through terrain generation tools.

[0068] First, you need to make a scene terrain elevation map to restore terrain features, such as mountains, slopes, cliffs, 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 a complete terrain white model for the area. After completing the production of the terrain white model, use the model map to make a 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 locations. Among them, the terrain white model is such as Figure 6 shown.

[0069] The complex and special environment is composed of three parts: three-dimensional virtual digital terrain, virtual weather conditions, and combat atmosphere rendering. These three parts form a complete and vital virtual combat environment and situation, so that virtual combat has terrain conditions, weather conditions, different combat targets and target conditions, firepower deployment and strike conditions in different regional environments (coastal wetlands, plateau mountains, plains and hills, snowy and cold areas). The three can be freely set and combined to form various types of variable combat situations, making the variability and simulation of virtual combat vital.

[0070] The construction of complex and special environments includes the construction of complex terrain and the rendering of environmental effects. The complex terrain is produced according to the requirements of the designated area's landform characteristics, such as mountains, rivers, vegetation and other scenes. The environmental effects are mainly the rendering of the scene effects, including the atmosphere rendering of the actual combat atmosphere, the rendering of weather effects such as sunshine, snow, rain, fog, and the rendering of lights, etc.

[0071] Regarding virtual environment rendering, virtual scenes require a combination of terrain and environment to present realistic effects and achieve the most realistic sense of immersion. To complete the construction of complex terrain, it is necessary to render the overall environmental effect. Environmental rendering includes scene lighting rendering and weather effect rendering.

[0072] Lights are used to illuminate the scene and give it an atmosphere. The setting of lights is crucial to the atmosphere of the scene. Without lights, there will be no light map, no shadows, no special regions, weather, and combat atmosphere, and the entire scene will lack a sense of three-dimensionality. In a three-dimensional scene, the role of lights is not only to illuminate objects, but also to convey more information to trainees through lighting effects. To achieve the final real effect of the scene, many different lights need to be established, because in the real world, light sources are multifaceted, such as sunlight, candlelight, fluorescent lights, etc., and the effects of things observed under different light sources will also be different. In the Unity3D engine, there are four commonly used light sources: Directional Light, Point Light, Spotlight, and Area Light.

[0073] Among them, parallel light usually exists as the main light source of the scene to simulate the effect of sunlight. Parallel light can set common properties such as color, rendering mode, and light intensity. It can also set the shadow type of objects affected by light. Point light source is similar to a spherical light source. It only affects the lighting effect of objects within the scope of the scene 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 realized by point light sources. For example, the indicator lights of the console are realized by point light sources to simulate the indicator light effect. The spotlight effect is similar to the spotlight effect. The illumination range is cone-shaped and can be used to simulate the light effect of a flashlight. The spotlight property panel can be set to adjust the illumination range, illumination color, and illumination intensity. One of the disassembly and assembly scenes of a certain type of vehicle components in the virtual maintenance training system is simulated in an indoor environment and rendered using spotlights. Range light needs to be baked in the scene to take effect. The baking effect is to render the 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 rendering picture. In addition to adding common types of lights to the scene lighting effects, you can also use environmental reflection balls, light source searchlights, etc. to set special lighting effects. In addition to the light source, the light and shadow effects produced by the object itself are also greatly affected by the object's own shader, material, etc. For example, in the shader (Shader), you can set whether to receive light source illumination, whether to reflect the light source, etc. Therefore, the rendering of complex scene lighting effects requires the coordinated use of the light source itself and the object shader, etc.

[0074] 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.

[0075] The simulation of sunny, snowy, rainy, foggy and other weather conditions is mainly produced using shaders and particle effects in the Unity3D engine. First, select different shaders according to the weather to be simulated, set the light position, light reception, cloud thickness, cloud color, density and other parameters. These attribute parameters form a visual component panel in the Unity3D engine. The weather effect can be previewed by adjusting the range value of the corresponding parameters in the component panel, and the parameter attribute value can also be controlled by code to dynamically simulate various different weather conditions.

[0076] Special weather such as rain and snow needs 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, and sets various snowflake parameters, including the duration of snowflakes, whether they appear in a loop, the range of snowflakes, the speed of snowflakes, and other properties. The style of 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.

[0077] In rainy weather, the sky will be overcast, and you can see raindrops falling. At the same time, the ground will gradually show the effect of stagnant water or moisture. In foggy weather, there will be bursts of fog in the scene, and the distant visual effect will appear gray, and the field of vision will be reduced. In addition to the four weather types of sunny, snowy, rainy, and foggy, we also created weather effects such as sandstorms, thunder and lightning, cloudy, overcast, sunny and rainy, and each weather has distinct characteristics.

[0078] Furthermore, the weather management panel can be used for selective control switching. When switching from one weather to another, there will be a transition buffer effect and preview effect, and the volume of the ambient sound effect can be adjusted. The weather buffer effect is processed by the weather management module. For example, when switching from cloudy weather to clear weather, the clouds in the sky will gradually dissipate. When switching from rainy and snowy weather to clear weather, the rain and snow will gradually disappear, and the accumulated water and snow will gradually disappear and turn into a clear ground effect.

[0079] Regarding the day and night simulation, the day and night function is simulated by the passage of time. The time can be set to real time or virtual time. The real time is synchronized with the real environment time. The time passes slowly, and the changes in the scene are not obvious. The virtual time passes quickly, and the scene effect changes can be clearly perceived. The changes include the position of the sun and the changes in brightness during the day and night changes, the position of the moon at night, the degree of roundness, the changes of stars, etc. You can quickly control the passage of virtual time by dragging the time slider of the weather switching panel, and feel the effects of various changes more intuitively. This effect is to present the switching of the sun and the moon by controlling the rotation of the sky box (Sky Box) of the entire scene. The sky box operates through a set of time systems, and the brightness of the lights depends on the time axis. Therefore, controlling the changes in time can control the changes in day and night. The roundness of the moon also changes with the flow of time, but its changes are handled by a separate set of rules.

[0080] Virtual combat atmosphere rendering mainly develops vehicles and objects in three-dimensional virtual combat, such as various weapons, targets, military facilities and fortifications, personnel roles, etc. Virtual combat atmosphere mainly develops various special effects related to real combat, such as attack explosion effects, damage status, combat sound effects, bombing, artillery attacks, etc. in virtual combat, and sets and loads them in the virtual combat environment independently to create a real and cruel war environment, so that trainees can obtain real and immersive combat experience in the virtual combat environment, so that trainees can feel the pressure of combat in the virtual combat, overcome the fear of combat, and complete various tasks at the same time, conduct targeted psychological training for personnel, cultivate trainees' courageous spirit in the face of danger, exercise their ability to adapt to tense, cruel and dangerous situations, establish a high degree of confidence in defeating the enemy, and improve trainees' combat capabilities.

[0081] Regarding the actual combat atmosphere, the creation of the actual combat atmosphere is based on sound, light, and electricity. Various models and special effects that meet the characteristics of actual combat are added for rendering. The models include equipment, vehicle wreckage, shells, etc.; special effects are made using the particle system of the Unity3D engine, including explosion and combustion effects, shell tail effects, and smoke effects. In terms of sound, 3D sound effects are used to simulate the explosion sound, shell flight sound, bullet shooting sound, etc. from far to near. The effects of sound, light, and electricity are usually used as static resources. To make the actual combat atmosphere real and vivid, it is necessary to combine sound, light, and electricity, and control the timing of appearance and rendering. By using the Timeline component of the Unity3D engine, these resources can be well combined and controlled. Timeline is like a timeline with many tracks on the timeline. Different resources are controlled on different tracks. When different tracks run to the same time segment, the resources on each track will be controlled to show their corresponding effects. The integration and utilization of actual combat resources using Timeline can ultimately simulate a realistic actual combat environment effect.

[0082] Based on the above embodiment, another exemplary embodiment of an electronic device is also provided in the embodiment of the present application. In some possible implementations, the electronic device in the embodiment of 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 can implement the steps of the actual combat atmosphere rendering method based on virtual digitized terrain in the above embodiment when executing the program.

[0083] For example, in the case of electronic equipment Figure 1 Taking the server 100 in the example as an example, 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.

[0084] The embodiment of the present application also provides a computer-readable storage medium, on which computer-executable instructions are stored. When the computer-executable instructions are executed by the processor, the actual combat atmosphere rendering method based on virtual digitized terrain according to the embodiment 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, for example, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and / or cache memory (cache), etc. Non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc.

[0085] The embodiment of the present application also provides a computer program product or a computer program, which includes computer executable instructions, and the computer executable instructions are stored in a computer readable storage medium. The processor of the computer device reads the computer executable instructions from the computer readable storage medium, and the processor executes the computer executable instructions, so that the computer device executes the actual combat atmosphere rendering method based on virtual digitized terrain according to the embodiment of the present application.

[0086] Those skilled in the art will appreciate that the contents disclosed in this application may be subject to various variations and improvements. For example, the various devices or components described above may be implemented by hardware, or by software, firmware, or a combination of some or all of the three.

[0087] In addition, 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 can be used and run on the client and / or server. The units are only illustrative, and different aspects of the system and method can use different units.

[0088] Those skilled in the art will appreciate that all or part of the steps in the above method can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium, such as a read-only memory, a disk or an optical disk. Optionally, 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 function modules. The present application is not limited to any particular form of combination of hardware and software.

[0089] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by those 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 extremely formal sense, unless explicitly defined as such herein.

[0090] 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 the 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; 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.

2. The method for rendering combat atmosphere based on virtual digital terrain according to claim 1, characterized in that: The extracting of terrain features from 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 2, characterized in that: The step of 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 comprises: 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 feature vector representing the association between the three-dimensional virtual digitized terrain feature compression information; 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; 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; Normalizing the multi-scale associated feature vector of the three-dimensional virtual digitized terrain feature compression information to obtain a three-dimensional virtual digitized terrain weight feature vector; 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.

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

5. The method for rendering combat atmosphere based on virtual digital terrain according to claim 4, 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 weighted feature vector of the three-dimensional virtual digitized terrain and each feature matrix of the three-dimensional virtual digitized terrain feature map along the channel dimension is calculated to obtain the enhanced three-dimensional virtual digitized terrain feature map.

6. The method for rendering combat atmosphere based on virtual digital terrain according to claim 5, 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.

7. The method for rendering combat atmosphere based on virtual digital terrain according to claim 6, characterized in that: The generating and rendering a three-dimensional virtual digitized terrain based on the three-dimensional virtual digitized terrain feature map guided by the rendering style includes: 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.

8. The method for rendering combat atmosphere based on virtual digital terrain according to claim 7, 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 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.

9. The method for rendering combat atmosphere based on virtual digital terrain according to claim 8, characterized in that: 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: Acquire training data, wherein the training data includes a training three-dimensional virtual digitized terrain, a training actual combat atmosphere rendering style representation vector, and a real value of rendering the 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-inhibition structure to obtain the 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 the cross-modal joint encoder based on the MetaNet model to obtain the 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 three-dimensional virtual digitized terrain feature map guided by the training rendering style to obtain a loss function value of the three-dimensional virtual digitized terrain guided by the rendering style; The weighted sum of the generated loss function value and the three-dimensional virtual digitized terrain loss function value guided by the rendering style is used as the loss function value, and 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.

Images