Flight environment simulation multi-source data dynamic effect three-dimensional visualization generation method

Abstract

The present application relates to the field of flight environment simulation visualization, in particular to a flight environment simulation multi-source data dynamic effect three-dimensional visualization generation method, comprising: obtaining real-time state parameters of an aircraft, flow field disturbance, meteorological particles and dynamic environment parameters of terrain shielding through a sensor network and a meteorological server, synchronizing time stamps and aligning spatial coordinates of the data, and forming a unified space-time reference original flight scene data set; generating a dynamic three-dimensional effect field containing time evolution characteristics by using an improved adaptive fusion algorithm, including three-dimensional morphology of flow field disturbance, three-dimensional distribution of meteorological particles and three-dimensional gradient of terrain shielding; and outputting a three-dimensional flight environment simulation image with fused dynamic environment visual effects by real-time drawing in a graphics rendering pipeline according to the effect field and a basic three-dimensional geographic model. The present application realizes integrated rendering of multi-source data space-time fusion and environment dynamic effects, and is suitable for flight simulation, avionics simulation and airspace visualization scenes.

Description

Technical Field

[0001] This invention relates to the field of flight environment simulation and visualization technology, and in particular to a method for generating three-dimensional visualizations of dynamic effects from multi-source data in flight environment simulation. Background Technology

[0002] Current 3D visualization of flight environments often employs separate data sources and rendering modes. Aircraft status parameters are collected via sensors, meteorological data is provided by weather servers, and terrain data is provided by geographic information platforms. These data are processed independently and then overlaid for display. Flow field disturbances, meteorological particles, and terrain occlusion are typically rendered using static models or pre-generated texture maps, layered during the rendering phase, failing to create a unified 3D effect field under a consistent spatiotemporal reference.

[0003] In this approach, the timestamps of multi-source data are not synchronized, and there are deviations in spatial coordinates. Data fusion relies on simple weighting or linear stitching, which cannot adaptively match dynamically changing features. Flow field, meteorology, and terrain are only superimposed as independent visual elements, and their morphology and distribution cannot evolve continuously over time. Details of three-dimensional gradients and dynamic perturbations are missing.

[0004] Multi-source data cannot be accurately integrated within a unified spatiotemporal framework, dynamic environmental elements are difficult to form a continuously evolving three-dimensional effect field, and real-time integrated rendering of dynamic effect fields and basic geographic models cannot be achieved in the graphics rendering pipeline. The visual fusion effect and dynamic response accuracy cannot meet the simulation requirements of complex flight scenarios. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing technologies and propose a method for generating three-dimensional visualizations of dynamic effects from multi-source data in flight environment simulation.

[0006] To achieve the above objectives, the present invention employs the following technical solution: a method for generating three-dimensional visualizations of dynamic effects from multi-source data in flight environment simulation, comprising: The aircraft receives real-time status parameters and dynamic environmental parameters through a sensor network and a weather server. The dynamic environmental parameters include flow field disturbance information, meteorological particle information and terrain shading information. The real-time status parameters and dynamic environment parameters are time-stamped and aligned with spatial coordinates, and then merged into an original flight scenario dataset with a unified spatiotemporal reference. The original flight scene dataset is processed using an improved adaptive fusion algorithm to generate a dynamic three-dimensional effect field of the space where the aircraft is located. The dynamic three-dimensional effect field includes the three-dimensional morphology of flow field disturbance, the three-dimensional distribution of meteorological particles, and the three-dimensional gradient of terrain occlusion that can evolve over time. Based on the dynamic 3D effect field and combined with the preset basic 3D geographic model, the image is rendered in real time in the graphics rendering pipeline to generate and output a 3D flight environment simulation image that incorporates dynamic environmental visual effects.

[0007] As a further aspect of the present invention, the step of processing the original flight scene dataset using an improved adaptive fusion algorithm to generate a dynamic three-dimensional effect field of the space where the aircraft is located includes: The original flight scene dataset is subjected to noise filtering and threshold removal to obtain clean scene data; High dynamic components and quasi-static components are separated from the clean scene data. The high dynamic components include instantaneous vorticity information of the flow field and abrupt changes in cloud particle concentration. The quasi-static components include stable values ​​of terrain elevation. The improved adaptive fusion algorithm is invoked to process the high dynamic components. The improved adaptive fusion algorithm adaptively adjusts the size and shape of the spatial fusion kernel according to the data density and rate of change of the components, generating a continuous three-dimensional morphology of flow field disturbance and a three-dimensional distribution of meteorological particles. The quasi-static component is input into the terrain occlusion analysis model, which calculates the occlusion relationship between terrain elevation and preset ambient light based on the principle of ray tracing, and generates a three-dimensional terrain occlusion gradient. The three-dimensional morphology of the flow field disturbance, the three-dimensional distribution of meteorological particles, and the three-dimensional gradient of terrain shading are assigned and synthesized on a three-dimensional spatial grid to form the dynamic three-dimensional effect field that fully describes the dynamic effects of the environment.

[0008] As a further aspect of the present invention, the working principle of the improved adaptive fusion algorithm includes: Receive the input high dynamic range component and divide the high dynamic range component into multiple data blocks; For each data block, calculate the spatial density and numerical change rate of the data points within it, and map the spatial density and numerical change rate to a fusion kernel parameter adjustment factor. Based on the fusion kernel parameter adjustment factor, the size and anisotropic stretching coefficient of the optimal Gaussian fusion kernel suitable for the current data block are dynamically calculated. In regions with high data density and gentle changes, a small spherical kernel is used, while in regions with sparse data or drastic changes, an ellipsoidal kernel with increased size and stretched in the gradient direction is used. Using the calculated optimal Gaussian fusion kernel, perform convolution interpolation calculations on all data points in the current data block to fill the three-dimensional space of the entire block; The interpolation results between adjacent data blocks are smoothly transitioned in the block boundary region using a weighted average based on the overlapping region. After traversing and processing all data blocks, the processing results of each block are spliced ​​together to form a continuous three-dimensional scalar field covering the entire computational domain. The three-dimensional scalar field is used to express the morphology of flow field disturbances or particle distribution.

[0009] As a further aspect of the present invention, separating the high dynamic component and the quasi-static component from the cleaned scene data includes: A time series is established for the flow field information in the clean scene data, and the numerical difference of each spatial point at multiple consecutive sampling times is calculated; The flow field information where the numerical difference exceeds a preset dynamic threshold is marked as a high dynamic point, and the instantaneous vorticity information corresponding to the high dynamic point constitutes the instantaneous vorticity information of the flow field in the high dynamic component. For the cloud particle concentration information in the clean scene data, calculate its variance over time, and take the cloud particle concentration value of the spatial region where the variance exceeds the preset change threshold as the abrupt change value of cloud particle concentration in the high dynamic component. Extract the terrain elevation values ​​that do not change over time from the cleaned scene data and classify them as the quasi-static components; Data classified as high dynamic components are removed from the cleaned scene data, and the remaining environmental background steady-state values ​​are also classified as quasi-static components.

[0010] As a further aspect of the present invention, based on the dynamic three-dimensional effect field and combined with a preset basic three-dimensional geographic model, real-time rendering is performed in the graphics rendering pipeline, including: Load the preset base 3D geographic model from the graphics memory. The base 3D geographic model includes a 3D terrain mesh and a static feature model. The system reads the dynamic three-dimensional effect field from memory, converts the three-dimensional morphological data of the flow field disturbance in the dynamic three-dimensional effect field into a three-dimensional disturbance texture, converts the three-dimensional distribution data of meteorological particles into a particle density volume texture, and converts the three-dimensional gradient data of terrain occlusion into a shadow weight texture. During the vertex shading stage, the positions of the 3D terrain mesh vertices in the basic 3D geographic model are dynamically offset based on the 3D perturbation texture. In the geometric shading and subdivision shading stages, the generation density and distribution position of meteorological particles are controlled according to the particle density volume texture. During the fragment shading stage, the final color and transparency of each pixel are calculated by combining the model after vertex offset, the generated weather particles, the shadow weight texture, and the dynamic lighting parameters. The calculated pixel data is output to the frame buffer and post-processed to form the three-dimensional flight environment simulation image that incorporates dynamic environmental visual effects.

[0011] As a further aspect of the present invention, the step of dynamically offsetting the vertex positions of the three-dimensional terrain mesh in the basic three-dimensional geographic model based on the three-dimensional perturbation texture during the vertex shading stage includes: Obtain the original world coordinates of each vertex of the 3D terrain mesh in the basic 3D geographic model; Based on the original world coordinates of the vertex, sampling is performed in the three-dimensional perturbation texture to read the three-dimensional perturbation vector and perturbation intensity at the coordinate point; Multiplying the three-dimensional perturbation vector by the perturbation intensity yields the real-time displacement vector applied to the vertex; The original world coordinates of the vertex are added to the real-time displacement vector to calculate the new world coordinates of the vertex after the offset. The new world coordinates and other vertex properties are then passed to the subsequent shading stage.

[0012] As a further aspect of the present invention, the step of controlling the generation density and distribution position of meteorological particles according to the particle density volume texture in the geometric shading and subdivision shading stages includes: In the geometry shader, obtain the spatial bounding box information of the primitives to be processed; Based on the spatial bounding box information, the corresponding voxel region is queried in the particle density volume texture, and the particle density reference value stored in the voxel region is read. Based on the particle density reference value, the number of meteorological particles that should be generated in the current primitive space is calculated, and the number of meteorological particles is proportional to the particle density reference value. Within the spatial bounding box, the initial positions of a corresponding number of meteorological particles are generated according to a predefined random distribution algorithm; For each generated meteorological particle at its initial position, a subdivision shader or a computation shader is invoked to calculate the particle's size and basic shape parameters based on the gradient information of the particle density volume texture, and the particle is then instantiated and output.

[0013] As a further aspect of the present invention, in the fragment shading stage, the final color and transparency of each pixel are calculated by combining the vertex offset model, the generated weather particles, the shadow weight texture, and the dynamic lighting parameters, including: In the fragment shading stage, the model primitives processed by the vertex shading stage are received, and the model primitives have completed the dynamic offset of vertex positions according to the three-dimensional perturbation texture. Receive meteorological particle primitives generated during the geometric coloring and subdivision coloring stages; The dynamic lighting parameters are obtained, including the light source direction vector, the light source color intensity, and the ambient light color intensity. For the current pixel to be processed, sample the shadow weight texture to obtain the terrain occlusion weight value of the current pixel at the corresponding position in the world coordinate system; The color intensity of the light source is attenuated according to the terrain shading weight value, and the effective illumination intensity after shading correction is calculated. Based on the light source direction vector, the world coordinates of the pixel, and the normal information, the reflection color value of the model surface after vertex offset to the effective light intensity is calculated using the lighting model. For meteorological particle primitives, the scattering color and transmission color of the particles are calculated based on the particle density value sampled from the particle density volume texture and the effective illumination intensity, and the opacity of the particles is calculated in combination with the particle density value. The reflected color value of the model surface is superimposed and mixed with the scattered color and transmitted color of the particles. The superimposed and mixed ratio is determined by the opacity of the particles and the spatial depth relationship between the particles and the model surface. The superimposed and mixed color value is added to the ambient light color intensity to obtain the final color of the pixel, and the opacity of the model surface is mixed with the opacity of the particles to obtain the final transparency of the pixel.

[0014] As a further aspect of the present invention, the real-time state parameters and dynamic environment parameters are time-stamped and spatially aligned, and fused into a raw flight scene dataset with a unified spatiotemporal reference, including: The flow field disturbance information received from the sensor network, the meteorological particle information received from the meteorological server, and the terrain shading information obtained from the terrain database all have their own timestamps and spatial coordinate system identifiers. Establish a global world coordinate system and a simulation master clock; The timestamps of data from different sources are compared with the simulation master clock, and all data are aligned to the same simulation time using interpolation or forward hold methods. The spatial coordinates of data from different sources are uniformly transformed to the world coordinate system according to their coordinate system identifiers using a preset coordinate transformation matrix; The flow field disturbance information, meteorological particle information, terrain masking information, and real-time status parameters of the aircraft, after being aligned in time and space, are encapsulated and packaged according to a predefined data structure to generate the original flight scene dataset. Each data record in the original flight scene dataset contains world coordinates and simulation time.

[0015] As a further aspect of the present invention, the method further includes a step of optimizing the dynamic three-dimensional effect field before rendering: Receive the generated dynamic 3D effect field, and receive the current viewpoint position and viewing direction; Based on the viewpoint position and the direction of the line of sight, the spatial importance level of different regions in the dynamic three-dimensional effect field relative to the viewpoint is calculated. The spatial importance level is determined comprehensively based on distance, line of sight angle and the proportion of screen space occupied. Based on the spatial importance level, differentiated resolutions and update frequencies are assigned to data in different regions of the dynamic three-dimensional effect field. High-importance regions maintain or increase their original resolution and highest update frequency, while low-importance regions reduce their data resolution and update frequency. The dynamic 3D effect field data, after being adjusted for resolution and update frequency, is sent into the graphics rendering pipeline for subsequent rendering. When the viewpoint position or line of sight changes, the spatial importance level is recalculated and the data resolution and update frequency of each region are adjusted.

[0016] Compared with the prior art, the advantages and positive effects of the present invention are as follows: Synchronizing real-time state parameters with dynamic environmental parameters using timestamps and aligning them with spatial coordinates can eliminate the spatiotemporal discrepancies between multi-source data, providing a unified spatiotemporal reference for data from different sources and avoiding temporal misalignment and spatial offset. An improved adaptive fusion algorithm is used to process the original flight scene dataset under the unified spatiotemporal reference. This algorithm can dynamically match the changing characteristics of different types of parameters, such as flow field disturbances, meteorological particles, and terrain shading, generating a dynamic 3D effect field that includes the 3D morphology of flow field disturbances, the 3D distribution of meteorological particles, and the 3D gradient of terrain shading that evolves over time. This allows environmental parameters to form a continuous and spatiotemporally consistent expression in 3D space, fully representing the spatial distribution and change processes of various environmental elements, and realistically reflecting the correlations between parameters.

[0017] By rendering dynamic 3D effect fields and pre-built basic 3D geographic models in real time within the graphics rendering pipeline, dynamic environmental visual elements and geospatial models can be integrated and expressed in the same rendering process, with dynamic environmental changes and geospatial structures presented synchronously. Flow field disturbances, meteorological particle distribution, and terrain occlusion gradients in the dynamic 3D effect field directly participate in rendering calculations. Environmental visual effects are no longer presented as independent layers, but are consistent with the geographic model in terms of spatial location, lighting changes, and occlusion relationships. The rendered 3D flight environment simulation image possesses continuous dynamic evolution characteristics, realistically reflecting the spatial gradients and temporal changes of environmental elements. The integration degree between the dynamic environment and the basic geographic space is improved, ensuring the spatiotemporal consistency and visual integrity of the 3D visualization results. Attached Figure Description

[0018] Figure 1This is a state diagram of the three-dimensional visualization generation method for multi-source data dynamic effects in flight environment simulation as described in this invention. Figure 2 A flowchart for generating dynamic 3D effect fields for an improved adaptive fusion algorithm; Figure 3 A flowchart illustrating the improved adaptive fusion algorithm. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0020] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0021] See Figure 1 This invention provides a method for generating three-dimensional visualizations of dynamic effects from multi-source data in flight environment simulation. The overall method includes: Through a sensor network deployed in and around the aircraft, and a meteorological server connected to a remote data service, the system receives real-time status parameters of the aircraft and dynamic environmental parameters including flow field disturbance information, meteorological particle information, and terrain shading information. The received data, with different sources and timestamps, is fed into a data fusion preprocessing unit. In this unit, based on a global simulation master clock and a predefined world coordinate system, all data are time-stamped and spatially aligned, then fused and encapsulated into a raw flight scene dataset with a unified spatiotemporal reference. This dataset is then transmitted to a dynamic effect field generation module. The core of this module is an improved adaptive fusion algorithm that processes the raw flight scene dataset to calculate the dynamic three-dimensional effect field of the space in which the aircraft is located. This effect field is a comprehensive data field containing the three-dimensional morphology of flow field disturbances that evolves over time, the three-dimensional distribution of meteorological particles, and the three-dimensional gradient of terrain shading. In a graphics rendering engine, based on the generated dynamic 3D effect field and combined with the pre-set basic 3D geographic model, the vertex shader, geometry shader, subdivision shader and fragment shader are scheduled in the graphics rendering pipeline to perform real-time drawing calculations, generate and output a 3D flight environment simulation image that integrates dynamic environment visual effects to the display device.

[0022] In one embodiment of the present invention, see [reference] Figure 2 A digital filtering module applies noise filtering to the original flight scene dataset. This module uses a statistical outlier detection method to identify and remove outliers from the data, generating clean scene data. The clean scene data is then input into a component separation unit. In practice, the component separation unit performs time series analysis and feature extraction on the clean scene data. For the flow field information in the clean scene data, the component separation unit establishes a time series for each spatial point and calculates its numerical difference at consecutive sampling times. Flow field information with numerical differences exceeding a preset dynamic threshold is marked as high-dynamic points. The instantaneous vorticity information corresponding to these high-dynamic points constitutes the instantaneous vorticity information of the flow field in the high-dynamic component. For the cloud particle concentration information in the clean scene data, the component separation unit calculates its variance over time. Cloud particle concentration values ​​in spatial regions with variances exceeding a preset change threshold are extracted as abrupt changes in cloud particle concentration values ​​in the high-dynamic component. Stable values ​​of terrain elevation are directly extracted from the clean scene data and classified as quasi-static components. After removing all data marked as high dynamic components from the clean scene data, the remaining environmental background steady-state parameters are also classified as quasi-static components.

[0023] In practice, the separated high-dynamic components are fed into an improved adaptive fusion algorithm for processing. The improved adaptive fusion algorithm receives the high-dynamic components as input and dynamically adjusts the size and shape of the spatial fusion kernel based on two parameters: data density and numerical rate of change within the high-dynamic components. In regions with high data density and a gradual rate of change, the improved adaptive fusion algorithm uses a smaller spherical kernel for data fusion; in regions with sparse data or a rapid rate of change, it uses an ellipsoidal kernel with an increased size and stretched along the data gradient direction. Through this adaptive fusion process, the improved adaptive fusion algorithm interpolates discrete high-dynamic data points and generates a continuous, three-dimensional morphology of flow field perturbations and a three-dimensional distribution of meteorological particles covering three-dimensional space.

[0024] In practice, the quasi-static component is input into an independent terrain occlusion analysis model. The core calculation of this model is based on ray tracing principles. The model loads stable terrain elevation values ​​as the 3D terrain surface and receives preset ambient lighting direction parameters. It emits rays from each terrain sampling point towards the ambient lighting direction, detects ray collisions in 3D space, and quantifies the light accessibility of each sampling point by calculating the degree to which the rays are occluded by the terrain itself. This generates a 3D terrain occlusion gradient field that describes the continuous variation in the degree of terrain occlusion. The final synthesis step is performed in a 3D spatial meshing module. This module establishes a 3D spatial mesh covering the current airspace and assigns the 3D morphological data of flow field disturbances and the 3D distribution data of meteorological particles generated by the improved adaptive fusion algorithm, as well as the 3D terrain occlusion gradient data generated by the terrain occlusion analysis model, to the corresponding voxel vertices of the 3D spatial mesh. The assignment process is based on precise matching of spatial coordinates. For voxels with multiple data at the same spatial location, a weighted summation method is used to synthesize the data, ultimately forming a comprehensive data field that fully describes the dynamic effects of flow, particles, and occlusion, i.e., a dynamic three-dimensional effect field.

[0025] In one embodiment of the present invention, when the improved adaptive fusion algorithm is started, refer to... Figure 3The algorithm receives the separated high-dynamic components as input data and divides them into multiple uniformly sized or non-uniformly distributed data blocks in the spatial domain. For each data block, the algorithm calculates the average spatial density and average rate of change of all data points within the block. Spatial density is defined as the number of data points per unit volume of the block, and the rate of change of value is obtained by averaging the numerical gradient magnitudes of all data points within the block. The algorithm then applies a mapping function to the calculated spatial density and average rate of change of value to generate a fusion kernel parameter adjustment factor for the current data block. This adjustment factor dynamically controls the shape of the Gaussian fusion kernel, and its calculation formula is expressed as follows: in: The parameter adjustment factor for the fusion kernel is a multi-dimensional vector. This represents the calculated average spatial density; Represents a preset global reference space density; This represents the calculated average rate of change of the numerical value. Represents a preset global reference value rate of change; This represents a predefined mapping function. The adjustment factor is based on the calculated fusion kernel parameters. The improved adaptive fusion algorithm determines the optimal Gaussian fusion kernel size parameters and anisotropic stretching coefficient matrix suitable for the current data block in real time. In data block regions with high data density and a gradual rate of numerical change, the fusion kernel parameter adjustment factor... The improved adaptive fusion algorithm uses a smaller base size and identity matrix as the stretching coefficient, which is equivalent to using a small spherical kernel. In data blocks that are sparse or have drastic rate of change, the fusion kernel parameter adjustment factor... The improved adaptive fusion algorithm increases the base size and generates a non-identity matrix as an anisotropic stretching coefficient based on the principal gradient direction calculated within the data block. Its effect is equivalent to using an ellipsoidal kernel with increased size and stretched according to the gradient direction.

[0026] In some embodiments, after determining the optimal Gaussian fusion kernel, the improved adaptive fusion algorithm uses this Gaussian fusion kernel to perform three-dimensional convolution interpolation calculations on all discrete data points within the current data block. The convolution interpolation calculation traverses each interpolation target point in the three-dimensional space corresponding to the data block, and obtains the scalar value of the target point by weighted summation based on the spatial distance between the target point and each data point and the Gaussian fusion kernel function, thereby filling the three-dimensional space of the entire data block. To address the potential discontinuity in results between adjacent data blocks due to separate interpolation calculations, the improved adaptive fusion algorithm performs a distance-based weighted average calculation on the interpolation results from different data blocks within a pre-defined overlapping area at the data block boundaries, achieving a smooth transition. After traversing and processing all data blocks, the improved adaptive fusion algorithm stitches together the processing results of all data blocks according to their original spatial arrangement, ultimately forming a continuous and smooth three-dimensional scalar field covering the entire computational domain. This three-dimensional scalar field is used to express the three-dimensional morphology of flow field disturbances or the three-dimensional distribution of meteorological particles.

[0027] In practice, the process of separating the high-dynamic component and quasi-static component from clean scene data is executed independently of the improved adaptive fusion algorithm. The separation process operates on the flow field information in the clean scene data, establishing a buffer for each spatial point with a time series in the clean scene data, and continuously recording flow field parameter values ​​at multiple sampling times. For each spatial point, the numerical difference between its adjacent sampling times is calculated, resulting in a series of difference values. A preset dynamic threshold is set, and the flow field information corresponding to moments in the numerical difference sequence whose absolute value exceeds the preset dynamic threshold is marked as a high-dynamic point. The instantaneous vorticity information carried by all high-dynamic points is extracted and aggregated to constitute the instantaneous vorticity information of the flow field in the high-dynamic component. It can be understood that the separation process simultaneously processes the cloud particle concentration information in the clean scene data. Within a preset time window, the variance of the cloud particle concentration value at each spatial point as a function of time is calculated. A preset change threshold is set, and spatial regions whose variance calculation results exceed the preset change threshold are identified as regions of significant change. The cloud particle concentration values ​​of these spatial regions at the corresponding times are extracted as the abrupt change values ​​of cloud particle concentration in the high-dynamic component. In clean scene data, terrain elevation values ​​that do not change over time are directly classified as quasi-static components. After extracting instantaneous vorticity information of the flow field and abrupt changes in cloud particle concentration, these data that have been classified as high-dynamic components are removed from the original clean scene data. The remaining, gently changing steady-state values ​​of the environmental background in the clean scene data, such as large-scale background wind fields or basic humidity fields, are also classified as quasi-static components.

[0028] Optionally, the separation of high dynamic components and quasi-static components can be performed in parallel. The detection and extraction processes for flow field information, cloud particle concentration information, and terrain elevation are synchronously completed by different processing threads or dedicated computing units. The result of the separation process is the generation of two logically independent datasets: a high dynamic component dataset and a quasi-static component dataset. The high dynamic component dataset is fed into an improved adaptive fusion algorithm, while the quasi-static component dataset, particularly the stable values ​​of terrain elevation, is fed into a terrain occlusion analysis model for subsequent calculations. In some embodiments, the preset dynamic threshold and preset change threshold can be fixed parameters or adaptively adjusted based on historical data statistical analysis. It is understood that clean scene data is the input to the separation operation, and the thoroughness of the separation operation directly affects the processing effect of the subsequent improved adaptive fusion algorithm and terrain occlusion analysis model.

[0029] In one embodiment of the present invention, during the initialization of the graphics rendering engine, a pre-set basic 3D geographic model is loaded from the graphics memory. This basic 3D geographic model includes a 3D terrain mesh composed of triangular facets and static feature models representing buildings and vegetation. Simultaneously, data from a dynamic 3D effect field is read from system memory. This dynamic 3D effect field includes three data types: flow field perturbation 3D morphology, meteorological particle 3D distribution, and terrain occlusion 3D gradient. The graphics rendering engine converts the flow field perturbation 3D morphology data into 3D perturbation textures that can be sampled in the shader using the texture generation unit, converts the meteorological particle 3D distribution data into particle density volume textures, and converts the terrain occlusion 3D gradient data into shadow weight textures. These textures are sent to the graphics processor's video memory for subsequent stages of the rendering pipeline. The vertex shading stage of the rendering pipeline is executed, where the vertex shader processes each vertex in the basic 3D geographic model. During the vertex shading stage, the vertex positions of the 3D terrain mesh vertices in the basic 3D geographic model are dynamically offset based on the 3D perturbation textures. In the geometry shading and subdivision shading stages, the generation density and distribution position of meteorological particles are controlled according to the particle density volume textures. During the fragment shading stage, the final color and transparency of each pixel are calculated by combining the vertex offset model, generated weather particles, shadow weight textures, and dynamic lighting parameters. Finally, the calculated pixel data is output to the frame buffer, and after post-processing operations such as tone mapping and anti-aliasing, a 3D flight environment simulation image with dynamic environmental visual effects is formed.

[0030] In practice, the vertex shader program specifically implements the process of dynamically offsetting the vertex positions of the 3D terrain mesh in the basic 3D geographic model based on the 3D perturbation texture during the vertex shading stage. The vertex shader obtains the original world coordinates of each vertex of the 3D terrain mesh in the basic 3D geographic model. The original world coordinates are the 3D position vectors of the vertex in the global world coordinate system. Based on the original world coordinates of the vertices, the vertex shader samples the 3D perturbation texture. The 3D perturbation texture is a 3D texture in which each texel stores a 3D perturbation vector and a scalar perturbation intensity. The sampling operation reads the 3D perturbation vector and perturbation intensity corresponding to the current vertex coordinate position. The vertex shader multiplies the sampled 3D perturbation vector by the perturbation intensity value to obtain the real-time displacement vector applied to the current vertex. Subsequently, the vertex shader adds the original world coordinates of the vertex to the real-time displacement vector to calculate the new world coordinates after the vertex offset. This addition calculation can be expressed by the formula: in: This represents the new world coordinates after the vertex offset. The original world coordinates representing the vertex. This represents the real-time displacement vector obtained by multiplying the 3D perturbation vector by the perturbation intensity. The vertex shader passes the calculated new world coordinates and other vertex attributes, such as normal vectors and texture coordinates, to subsequent geometry shading and subdivision shading stages. In some embodiments, the correspondence and uses of the above data and textures are shown in Table 1: Table 1: Correspondence between rendering stages and textures Optionally, the sampling process for 3D perturbation textures may involve coordinate transformation and interpolation. Since the original world coordinates of vertices are continuous values, while the 3D perturbation texture is discrete volume data, the vertex shader uses a trilinear interpolation algorithm to obtain smooth perturbation values ​​during sampling. After the vertex shader completes all vertex processing and outputs the results, the 3D terrain mesh, after vertex offsetting, is assembled into primitives and passed to subsequent shading stages. In the geometry shading and subdivision shading stages, the system instantiates and generates meteorological particles representing effects such as clouds, fog, and precipitation in 3D space based on information provided by the particle density volume texture. The particle density volume texture defines the expected meteorological particle density reference value at each voxel location in 3D space. The geometry shader determines the number of particles to generate by querying the voxel region corresponding to the primitive space bounding box, reading the particle density reference value, and randomly distributing the initial particle positions within the bounding box. The subdivision shader or computation shader further adjusts the size and shape parameters of the generated particles based on the local gradient of the particle density volume texture.

[0031] The fragment shading stage is the final pixel shading step. The fragment shader receives primitives processed by all previous stages, including a 3D terrain mesh with vertex offsets and a static terrain model surface, as well as generated weather particle sprites or patches. The fragment shader simultaneously accesses the shadow weight texture and dynamic lighting parameters passed in by the application. For each pixel fragment requiring shading, the fragment shader samples the shadow weight texture based on its world coordinates to obtain a terrain occlusion weight value, which is used to modulate the direct lighting intensity in the dynamic lighting parameters. The fragment shader uses the lighting model to calculate the occlusion-corrected model surface color. For weather particles covering the same pixel location, the fragment shader calculates the scattered and transmitted light colors based on their density and lighting. The fragment shader performs an alpha blending operation, superimposing the model surface color and particle color based on the particle's opacity, and finally adding it to the ambient light to obtain the final output color value for that pixel. All pixels output by the fragment shader constitute the complete image. In some embodiments, the dynamic lighting parameters may include information from multiple light sources; the fragment shader needs to calculate occlusion and shading for each light source and accumulate the contributions. Optionally, the resolution of the particle density volume texture can differ from that of the 3D perturbation texture to accommodate different levels of detail. It can be understood that the entire graphics rendering pipeline, from vertex shading to fragment shading, is a pipeline process where each stage relies on the output of the previous stage and specific texture or parameter data, working together to generate the final 3D flight environment simulation image that includes dynamic flow fields, weather particles, and terrain occlusion visual effects.

[0032] The sampling process for 3D perturbation textures mentions the use of a "trilinear interpolation algorithm" to obtain smooth perturbation values ​​from discrete volume data. This method is the core of achieving high-quality data sampling in continuous 3D space. The implementation of the trilinear interpolation algorithm begins with coordinate normalization and voxel localization. The vertex shader maps the original world coordinates of vertices to the texture coordinate system of the 3D perturbation texture through an affine transformation matrix. This texture coordinate system is a normalized 3D space where the coordinates in all three dimensions range from 0 to 1. The mapped coordinate values ​​indicate the relative position of the vertex in the texture volume data. Since the texture volume data consists of discrete voxels, the mapped continuous coordinate points must lie within a minimum cube unit consisting of 8 adjacent voxel vertices. The trilinear interpolation algorithm needs to obtain the data values ​​stored by each of these 8 voxel vertices. The data stored by each voxel vertex is a data structure containing a 3D perturbation vector and a scalar perturbation intensity. The algorithm calculates the local relative coordinates of the current sampling point within this cube unit. This coordinate is a 3D decimal used to indicate the offset ratio of the sampling point from the lower left corner of the cube unit's preceding voxel vertex. The specific interpolation calculation is completed in three linear steps. Four linear interpolations are performed along the X-axis, weighting and mixing four pairs of voxel vertices located on the same X-plane based on their local X-coordinates to generate two intermediate interpolation points on the YZ-plane. Linear interpolation is then performed along the Y-axis on these two intermediate points, generating one intermediate result along the Z-axis based on its local Y-coordinates. A final linear interpolation is performed along the Z-axis, mixing the two final candidate values ​​along the Z-axis based on their local Z-coordinates to calculate the final 3D perturbation vector and perturbation intensity scalar value at the sampling point. Through this sequential combination of linear interpolations along the three axes in 3D space, the trilinear interpolation algorithm can simulate a continuous and smooth data field. This prevents the texture sampling results based on vertex coordinates from exhibiting abrupt, step-like changes due to the discreteness of the volume data. This is crucial for ensuring a natural transition in terrain vertex offsets caused by flow field perturbations. The entire process is completed entirely within the vertex shader using hardware-accelerated texture sampling and arithmetic instructions, ensuring the efficiency of real-time rendering.

[0033] In the fragment shading stage description, the color synthesis of weather particles and model surfaces mentions the execution of an "alpha blending operation," a fundamental graphics method for handling the overlay of semi-transparent and transparent visual effects. The specific implementation of the alpha blending operation relies on pre-calculated color, opacity, and depth values ​​within a pixel fragment. After the fragment shader calculates the color and opacity of a weather particle, this pixel fragment does not immediately overwrite the existing pixel in the framebuffer, which may contain colors from previously drawn terrain models or other opaque objects. The alpha blending operation determines how the old and new colors are combined based on a predefined blending equation, typically configured as an "over" operation: the new fragment color's opacity + the original target color (1 - new fragment opacity). This calculation is performed by fixed-function hardware after the graphics processor's rasterization stage and before writing to the framebuffer. To achieve the correct visual overlay order, especially when multiple semi-transparent particles intersect in depth, the rendering pipeline usually requires depth sorting or the use of order-independent transparency rendering techniques. In the implementation described in the example, for multiple meteorological particle fragments that may exist at the same pixel location, the system first sorts these fragments from back to front according to their depth values. Following the order from farthest to nearest, each particle fragment is sequentially alpha-blended with the accumulated color in the current frame buffer, with each blend using the result of the previous blend as the new target color. For blending terrain model surfaces with particles, since terrain is typically considered an opaque surface with an opacity of 1, the terrain surface color is written to the frame buffer. Subsequently, the particle colors entering the blending stage are superimposed with the terrain color based on the particle's opacity, thereby simulating the visual effect of particles attaching to the terrain surface or floating in front of the terrain. The configuration state of the alpha blending operation is set by the graphics application interface before rendering, including enabling blending, specifying the source blending factor, and the target blending factor. In the fragment shader, the calculated color output value contains three color channels (red, green, and blue) and one alpha channel representing opacity. The value of this alpha channel is directly used in the calculation of the blending equation. By precisely controlling the opacity of each particle and the edges of the model, the Alpha Blend operation allows dense clouds to obscure the background scenery, while sparse fog can be softly superimposed on them, ultimately generating a composite image with realistic depth and texture.

[0034] In one embodiment of the present invention, after the vertex shading stage, the graphics rendering pipeline passes primitives to the geometry shader. In the geometry shader, the bounding box information of the primitive to be processed is obtained. The bounding box is a minimal axis-aligned hexahedron that can completely contain the primitive, and its information is defined by the maximum and minimum coordinates of all vertices of the primitive. Based on the bounding box information, the corresponding voxel region is queried in the particle density volume texture. The particle density volume texture is a three-dimensional texture, where each voxel stores a particle density reference value representing the number of reference particles in that unit space. The query operation reads the particle density reference values ​​stored in all voxels covered by the bounding box. Based on the read particle density reference values, the number of meteorological particles to be generated in the current primitive space is calculated. The number of meteorological particles is proportional to the particle density reference value, and the relationship can be expressed by the formula: in: This represents the calculated number of meteorological particles. It is a preset proportional coefficient. This represents the spatial average of the particle density reference values ​​obtained from the particle density volume texture. This represents the volume of the current primitive bounding box. Within the bounding box, a corresponding number of initial positions for meteorological particles are generated according to a predefined random distribution algorithm. This algorithm ensures that the particles are uniformly distributed or distributed according to a specific probability density function within the bounding box. For each generated initial position of a meteorological particle, the system calls the subdivision shader or computation shader to calculate the particle's size and basic shape parameters based on the gradient information of the particle density volume texture at the particle's position. The calculation of the particle's size and basic shape parameters takes into account the direction and magnitude of the gradient vector. The meteorological particle with its final determined position, size, and basic shape parameters is then instantiated and output as a new point sprite or patch primitive, which is then sent to the rasterization stage.

[0035] In practice, the fragment shading stage receives primitives processed in previous shading stages and performs final shading calculations. It also receives model primitives processed by the vertex shading stage, whose vertex positions have been dynamically offset based on the 3D perturbation texture. Simultaneously, the fragment shading stage receives meteorological particle primitives generated in the geometry and subdivision shading stages. The fragment shading stage obtains dynamic lighting parameters from a unified shader parameter buffer, including the light source direction vector, light source color intensity, and ambient light color intensity. For the current pixel to be processed, the fragment shader samples the shadow weight texture, which stores previously calculated terrain occlusion 3D gradient information. The sampling operation obtains the terrain occlusion weight value of the current pixel at its corresponding position in the world coordinate system. The light source color intensity is attenuated based on the terrain occlusion weight value to calculate the effective lighting intensity after occlusion correction. Based on the light source direction vector, the pixel's world coordinates, and normal information, the diffuse and specular reflection color values ​​of the model surface after vertex offset are calculated using the Von Leigh lighting model or a similar lighting model. For meteorological particle primitives, the fragment shader calculates the scattered and transmitted colors of the particles based on the particle density values ​​sampled from the particle density volume texture at the particle positions and the calculated effective illumination intensity, combined with Mie scattering theory. It then calculates the particle opacity based on the particle density values. The reflected color value of the model surface is superimposed and mixed with the scattered and transmitted colors of the particles. The superposition and mixing ratio is determined by the particle opacity and the spatial depth relationship between the particles and the model surface. The superimposed and mixed color value is added to the ambient light color intensity obtained in the fragment shading stage to obtain the final color of the pixel. The opacity of the model surface is then mixed with the opacity of the particles to obtain the final transparency of the pixel. The final color and transparency data are written to the frame buffer. In some embodiments, the key computational data flow of the above fragment shading stage is shown in Table 2. Table 2: Data Calculation Relationships in the Fragment Coloring Stage Optionally, particle opacity can be calculated directly and linearly related to the particle density value sampled from the particle density volume texture, or the density can be mapped to opacity using a preset transfer function. When calculating particle scattering color, the path length of light passing through the particle medium needs to be considered. This path length can be estimated using the particle diameter or the distance the current pixel's line of sight travels through the particle volume. It's understood that the opacity of the model surface is considered completely opaque in most cases, i.e., a value of 1.0. The fragment shading stage is performed pixel-by-pixel. For each pixel on the screen, the fragment shader may be called multiple times to process fragments from different primitives, and the final displayed color is determined through depth testing and blending operations. In some embodiments, to optimize performance, particle fragments completely occluded by the opaque model can be discarded in an early depth test. Optionally, the ambient light color intensity can be a constant or derived from an environment cubemap to provide more realistic scene lighting.

[0036] In one embodiment of the present invention, the flow field disturbance information received from the sensor network, the meteorological particle information received from the meteorological server, and the terrain shading information obtained from the terrain database, when transmitted to the system, all carry a timestamp of their respective acquisition time and a spatial coordinate system identifier describing their origin and axis direction. The timestamp accuracy reaches the millisecond level, and the spatial coordinate system identifier explicitly points to, for example, the WGS-84 geographic coordinate system, the body coordinate system, or the local meteorological grid coordinate system. During system initialization, a global world coordinate system covering the entire simulation domain is established as a unified reference framework for all spatial data, and a high-precision simulation master clock is started as a unified time reference for all time data. The timestamp synchronization process compares the timestamps carried by data from different sources with the current time of the simulation master clock. For data whose timestamps are earlier or later than the simulation master clock, linear interpolation or forward hold methods are used to adjust their data values ​​to the same simulation time that is completely aligned with the simulation master clock. The spatial coordinate alignment process uses the coordinate system identifier attached to each data point to find a pre-defined coordinate transformation matrix. This matrix defines the rotation, translation, and scaling relationships from the source coordinate system to the global world coordinate system. Matrix multiplication is then used to uniformly transform the spatial coordinates of all data to the global world coordinate system. The time- and space-aligned flow field disturbance information, meteorological particle information, terrain occlusion information, and the aircraft's real-time status parameters are then packaged and encapsulated in a predefined structured format to generate the original flight scenario dataset. Each data record in the original flight scenario dataset contains unified world coordinate values, the aligned simulation time, and other measurement or computational attributes.

[0037] In practice, the pre-rendering optimization step for the dynamic 3D effect field aims to improve rendering efficiency. It receives the dynamic 3D effect field generated by the dynamic effect field generation module and simultaneously receives the current viewpoint position and viewing direction vector from the view management module. Based on the viewpoint position and viewing direction, a spatial importance assessment module calculates the spatial importance level of different data regions in the dynamic 3D effect field relative to the viewpoint. The calculation of the spatial importance level comprehensively considers the distance factor, the viewing angle factor, and the screen space occupancy factor. The calculation formula is as follows: in: It represents the calculated spatial importance level and is a dimensionless scalar value; Represents the Euclidean distance from the boundary of a data region in a dynamic 3D effect field to the viewpoint position; This represents the angle between the line-of-sight vector and the line-of-sight vector from the center of the data region to the viewpoint; This represents the ratio of the pixel area occupied by the data region when projected onto the current view frustum near clipping plane to the total pixel area of ​​the screen; This represents a weighted calculation function that integrates three factors. The spatial importance level is calculated based on this. The dynamic detail level management module assigns differentiated resolutions and update frequencies to data in different regions of the dynamic 3D effect field. For regions with high spatial importance, such as those close to the viewer, located at the center of the field of view, or occupying a large portion of the screen, the original resolution and maximum update frequency of the data are maintained or increased. For regions with low spatial importance, such as those far away, located at the edge of the field of view, or occupying a small portion of the screen, the data resolution and update frequency are reduced. Methods for reducing resolution include downsampling the 3D volume data, and methods for reducing update frequency include skipping data updates for some frames. The dynamic 3D effect field data, after resolution and update frequency adjustments, is then fed into the graphics rendering pipeline for subsequent real-time rendering. When the viewpoint position or line of sight changes due to user interaction or aircraft movement, the spatial importance assessment module re-triggers calculations, recalculates the new spatial importance level of each region in the dynamic 3D effect field, and dynamically adjusts the data resolution and update frequency of each region based on the new level.

[0038] In some embodiments, the establishment of the coordinate transformation matrix relies on precise coordinate system calibration parameters, which are loaded from the configuration file during system initialization. The linear interpolation method in timestamp synchronization is suitable for situations where data is available at both the preceding and following timestamps, while the forward hold method is used for the blank period before the arrival of the latest data, filling it with the last valid data value. Spatial importance ranking calculation function. The specific form can be a linear weighted sum or a discrete mapping based on a decision table. When reducing data resolution, the dynamic level of detail (LMD) management module can employ downsampling methods such as mean pooling or median filtering to reduce information loss. Optionally, the update frequency can be adjusted in sync with the system frame rate; for example, high-importance regions are updated every frame, medium-importance regions every two frames, and low-importance regions every four frames. It can be understood that pre-rendering optimization is a continuously running dynamic process. The spatial importance assessment module and the dynamic LMD management module execute once per frame or every few frames in the rendering loop to ensure that computational resources are always concentrated on visually sensitive areas. In some embodiments, the screen space occupancy factor... The spatial importance can be approximated by projecting the bounding box of the data region onto screen space and calculating its two-dimensional convex hull area. Optionally, for data regions completely outside the view frustum, their spatial importance level can be set to the lowest, thus completely skipping the processing and rendering of data in that region during the frame.

[0039] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A method for generating three-dimensional visualizations of dynamic effects from multi-source data in flight environment simulation, characterized in that: The method includes: The aircraft receives real-time status parameters and dynamic environmental parameters through a sensor network and a weather server. The dynamic environmental parameters include flow field disturbance information, meteorological particle information, and terrain shading information. The real-time status parameters and dynamic environment parameters are time-stamped and aligned with spatial coordinates, and then merged into an original flight scenario dataset with a unified spatiotemporal reference. The original flight scene dataset is processed using an improved adaptive fusion algorithm to generate a dynamic three-dimensional effect field of the space where the aircraft is located. The dynamic three-dimensional effect field includes the three-dimensional morphology of flow field disturbance, the three-dimensional distribution of meteorological particles, and the three-dimensional gradient of terrain occlusion that can evolve over time. Based on the dynamic 3D effect field and combined with the preset basic 3D geographic model, the image is rendered in real time in the graphics rendering pipeline to generate and output a 3D flight environment simulation image that incorporates dynamic environmental visual effects.

2. The method for generating three-dimensional visualization of dynamic effects from multi-source data in flight environment simulation according to claim 1, characterized in that, The process of using an improved adaptive fusion algorithm to process the original flight scene dataset and generate a dynamic three-dimensional effect field of the space where the aircraft is located includes: The original flight scene dataset is subjected to noise filtering and threshold removal to obtain clean scene data; High dynamic components and quasi-static components are separated from the clean scene data. The high dynamic components include instantaneous vorticity information of the flow field and abrupt changes in cloud particle concentration. The quasi-static components include stable values ​​of terrain elevation. The improved adaptive fusion algorithm is invoked to process the high dynamic components. The improved adaptive fusion algorithm adaptively adjusts the size and shape of the spatial fusion kernel according to the data density and rate of change of the components, generating a continuous three-dimensional morphology of flow field disturbance and a three-dimensional distribution of meteorological particles. The quasi-static component is input into the terrain occlusion analysis model, which calculates the occlusion relationship between terrain elevation and preset ambient light based on the principle of ray tracing, and generates a three-dimensional terrain occlusion gradient. The three-dimensional morphology of the flow field disturbance, the three-dimensional distribution of meteorological particles, and the three-dimensional gradient of terrain shading are assigned and synthesized on a three-dimensional spatial grid to form the dynamic three-dimensional effect field that fully describes the dynamic effects of the environment.

3. The method for generating three-dimensional visualization of dynamic effects from multi-source data in flight environment simulation according to claim 2, characterized in that, The working principle of the improved adaptive fusion algorithm includes: Receive the input high dynamic range component and divide the high dynamic range component into multiple data blocks; For each data block, calculate the spatial density and numerical change rate of the data points within it, and map the spatial density and numerical change rate to a fusion kernel parameter adjustment factor. Based on the fusion kernel parameter adjustment factor, the size and anisotropic stretching coefficient of the optimal Gaussian fusion kernel suitable for the current data block are dynamically calculated. In regions with high data density and gentle changes, a small spherical kernel is used, while in regions with sparse data or drastic changes, an ellipsoidal kernel with increased size and stretched in the gradient direction is used. Using the calculated optimal Gaussian fusion kernel, convolution interpolation is performed on all data points in the current data block to fill the three-dimensional space of the entire block; The interpolation results between adjacent data blocks are smoothly transitioned in the block boundary region using a weighted average based on the overlapping region. After traversing and processing all data blocks, the processing results of each block are spliced ​​together to form a continuous three-dimensional scalar field covering the entire computational domain. The three-dimensional scalar field is used to express the morphology of flow field disturbances or particle distribution.

4. The method for generating three-dimensional visualization of dynamic effects from multi-source data in flight environment simulation according to claim 2, characterized in that, Separating high-dynamic components and quasi-static components from the cleaned scene data includes: A time series is established for the flow field information in the clean scene data, and the numerical difference of each spatial point at multiple consecutive sampling times is calculated; The flow field information where the numerical difference exceeds a preset dynamic threshold is marked as a high dynamic point, and the instantaneous vorticity information corresponding to the high dynamic point constitutes the instantaneous vorticity information of the flow field in the high dynamic component. For the cloud particle concentration information in the clean scene data, calculate its variance over time, and take the cloud particle concentration value of the spatial region where the variance exceeds the preset change threshold as the abrupt change value of cloud particle concentration in the high dynamic component. Extract the terrain elevation values ​​that do not change over time from the cleaned scene data and classify them as the quasi-static components; Data classified as high dynamic components are removed from the cleaned scene data, and the remaining environmental background steady-state values ​​are also classified as quasi-static components.

5. The method for generating three-dimensional visualization of dynamic effects from multi-source data in flight environment simulation according to claim 1, characterized in that, Based on the dynamic 3D effect field and combined with the preset basic 3D geographic model, real-time rendering is performed in the graphics rendering pipeline, including: Load the preset base 3D geographic model from the graphics memory. The base 3D geographic model includes a 3D terrain mesh and a static feature model. The system reads the dynamic three-dimensional effect field from memory, converts the three-dimensional morphological data of the flow field disturbance in the dynamic three-dimensional effect field into a three-dimensional disturbance texture, converts the three-dimensional distribution data of meteorological particles into a particle density volume texture, and converts the three-dimensional gradient data of terrain occlusion into a shadow weight texture. During the vertex shading stage, the positions of the 3D terrain mesh vertices in the basic 3D geographic model are dynamically offset based on the 3D perturbation texture. In the geometric shading and subdivision shading stages, the generation density and distribution position of meteorological particles are controlled according to the particle density volume texture. During the fragment shading stage, the final color and transparency of each pixel are calculated by combining the model after vertex offset, the generated weather particles, the shadow weight texture, and the dynamic lighting parameters. The calculated pixel data is output to the frame buffer and post-processed to form the three-dimensional flight environment simulation image that incorporates dynamic environmental visual effects.

6. The method for generating three-dimensional visualization of dynamic effects from multi-source data in flight environment simulation according to claim 5, characterized in that, In the vertex shading stage, the dynamic offset of the vertex positions of the 3D terrain mesh in the basic 3D geographic model based on the 3D perturbation texture includes: Obtain the original world coordinates of each vertex of the 3D terrain mesh in the basic 3D geographic model; Based on the original world coordinates of the vertex, sampling is performed in the three-dimensional perturbation texture to read the three-dimensional perturbation vector and perturbation intensity at the coordinate point; Multiplying the three-dimensional perturbation vector by the perturbation intensity yields the real-time displacement vector applied to the vertex; The original world coordinates of the vertex are added to the real-time displacement vector to calculate the new world coordinates of the vertex after the offset. The new world coordinates and other vertex properties are then passed to the subsequent shading stage.

7. The method for generating three-dimensional visualization of dynamic effects from multi-source data in flight environment simulation according to claim 5, characterized in that, In the geometric shading and subdivision shading stages, controlling the generation density and distribution position of meteorological particles based on the particle density volume texture includes: In the geometry shader, obtain the spatial bounding box information of the primitives to be processed; Based on the spatial bounding box information, the corresponding voxel region is queried in the particle density volume texture, and the particle density reference value stored in the voxel region is read. Based on the particle density reference value, the number of meteorological particles that should be generated in the current primitive space is calculated, and the number of meteorological particles is proportional to the particle density reference value. Within the spatial bounding box, the initial positions of a corresponding number of meteorological particles are generated according to a predefined random distribution algorithm; For each generated meteorological particle at its initial position, a subdivision shader or a computation shader is invoked to calculate the particle's size and basic shape parameters based on the gradient information of the particle density volume texture, and the particle is then instantiated and output.

8. The method for generating three-dimensional visualization of dynamic effects from multi-source data in flight environment simulation according to claim 5, characterized in that, In the fragment shading stage, the final color and transparency of each pixel are calculated by combining the vertex-offset model, the generated weather particles, the shadow weight texture, and dynamic lighting parameters, including: In the fragment shading stage, the model primitives processed by the vertex shading stage are received, and the model primitives have completed the dynamic offset of vertex positions according to the three-dimensional perturbation texture. Receive meteorological particle primitives generated during the geometric coloring and subdivision coloring stages; The dynamic lighting parameters are obtained, including the light source direction vector, the light source color intensity, and the ambient light color intensity. For the current pixel to be processed, sample the shadow weight texture to obtain the terrain occlusion weight value of the current pixel at the corresponding position in the world coordinate system; The color intensity of the light source is attenuated according to the terrain shading weight value, and the effective illumination intensity after shading correction is calculated. Based on the light source direction vector, the world coordinates of the pixel, and the normal information, the reflection color value of the model surface after vertex offset to the effective light intensity is calculated using the lighting model. For meteorological particle primitives, the scattering color and transmission color of the particles are calculated based on the particle density value sampled from the particle density volume texture and the effective illumination intensity, and the opacity of the particles is calculated in combination with the particle density value. The reflected color value of the model surface is superimposed and mixed with the scattered color and transmitted color of the particles. The superimposed and mixed ratio is determined by the opacity of the particles and the spatial depth relationship between the particles and the model surface. The superimposed and mixed color value is added to the ambient light color intensity to obtain the final color of the pixel, and the opacity of the model surface is mixed with the opacity of the particles to obtain the final transparency of the pixel.

9. The method for generating three-dimensional visualization of dynamic effects from multi-source data in flight environment simulation according to claim 1, characterized in that, The real-time state parameters and dynamic environment parameters are timestamped and aligned with spatial coordinates, and then merged into a raw flight scene dataset with a unified spatiotemporal reference, including: The flow field disturbance information received from the sensor network, the meteorological particle information received from the meteorological server, and the terrain shading information obtained from the terrain database all have their own timestamps and spatial coordinate system identifiers. Establish a global world coordinate system and a simulation master clock; The timestamps of data from different sources are compared with the simulation master clock, and all data are aligned to the same simulation time using interpolation or forward hold methods. The spatial coordinates of data from different sources are uniformly transformed to the world coordinate system according to their coordinate system identifiers using a preset coordinate transformation matrix; The flow field disturbance information, meteorological particle information, terrain masking information, and real-time status parameters of the aircraft, after being aligned in time and space, are encapsulated and packaged according to a predefined data structure to generate the original flight scene dataset. Each data record in the original flight scene dataset contains world coordinates and simulation time.

10. The method for generating three-dimensional visualization of dynamic effects from multi-source data in flight environment simulation according to claim 2, characterized in that, The method further includes a step of pre-rendering optimization of the dynamic three-dimensional effect field: Receive the generated dynamic 3D effect field, and receive the current viewpoint position and viewing direction; Based on the viewpoint position and the direction of the line of sight, the spatial importance level of different regions in the dynamic three-dimensional effect field relative to the viewpoint is calculated. The spatial importance level is determined comprehensively based on distance, line of sight angle and the proportion of screen space occupied. Based on the spatial importance level, differentiated resolutions and update frequencies are assigned to data in different regions of the dynamic three-dimensional effect field. High-importance regions maintain or increase their original resolution and highest update frequency, while low-importance regions reduce their data resolution and update frequency. The dynamic 3D effect field data, after being adjusted for resolution and update frequency, is sent into the graphics rendering pipeline for subsequent rendering. When the viewpoint position or line of sight changes, the spatial importance level is recalculated and the data resolution and update frequency of each region are adjusted.

Images