A battlefield environment dynamic updating terrain matching high-precision positioning system

Abstract

The present application relates to the field of computer navigation positioning, and particularly relates to a battlefield environment dynamic updating terrain matching type high-precision positioning system, which comprises a multi-source battlefield environment perception module, a dynamic terrain modeling module, a candidate area generation module, a terrain matching evaluation module, a positioning fusion calculation module and a positioning credibility evaluation output module; the present application introduces a structured terrain particle representation and a constraint-driven voxel space discretization method, constructs a dynamic and updatable three-dimensional battlefield terrain model based on high-trust observation data, realizes continuous updating of the terrain model with changes in the battlefield environment, and improves the consistency of the terrain model with the actual environment; in the terrain matching evaluation process, the candidate area and the dynamic terrain model are jointly matched and scored in combination with a multi-modal observation constraint and a terrain feature consistency evaluation mechanism, the risk of false matching caused by single feature matching is reduced, and the accuracy of terrain matching, the stability of positioning results and the positioning credibility are significantly improved.

Description

Technical Field

[0001] This invention relates to the field of computer navigation and positioning, specifically to a terrain-matching high-precision positioning system that dynamically updates the battlefield environment. Background Technology

[0002] In complex battlefield environments, high-precision positioning is a crucial foundation for systems such as command and control, weapon guidance, and situational awareness. Existing positioning technologies typically rely on satellite navigation systems, inertial navigation systems, or combinations thereof. While they can achieve good positioning accuracy in open environments, under complex battlefield conditions, due to dramatic terrain undulations, severe obstruction, and frequent environmental changes, single or traditional fusion positioning methods struggle to operate stably over long periods. This is mainly reflected in the following two aspects:

[0003] On the one hand, existing terrain matching-based positioning technologies mostly use static or pre-built terrain models, which are difficult to reflect changes in terrain structure caused by natural changes or human factors in the battlefield environment in a timely manner. This can easily lead to inconsistencies between the terrain model and the actual environment, thereby reducing matching reliability and positioning accuracy.

[0004] On the other hand, existing terrain matching methods typically use a single geometric feature or a simple similarity index for comparison during the matching evaluation process. They lack a unified modeling and comprehensive evaluation of the constraint relationship between multi-source observation data, which can easily lead to mismatches in complex terrain or multiple solutions, resulting in unstable positioning results or insufficient confidence. Summary of the Invention

[0005] To address the above issues and overcome the shortcomings of existing technologies, this invention provides a terrain-matching high-precision positioning system with dynamic battlefield environment updates. This invention introduces structured terrain particle representation and a constraint-driven voxel space discretization method to construct a dynamically updatable three-dimensional battlefield terrain model based on high-reliability observation data. This enables continuous updates of the terrain model as the battlefield environment changes, effectively solving the problem of static terrain models in existing technologies that fail to reflect real-time battlefield environment changes, and improving the consistency between the terrain model and the actual environment. In the terrain matching evaluation process, this invention combines multimodal observation constraints and a terrain feature consistency evaluation mechanism to jointly match and score candidate regions with the dynamic terrain model, reducing the risk of mismatches caused by single-feature matching and significantly improving the accuracy of terrain matching, the stability of positioning results, and the reliability of location.

[0006] The technical solution adopted in this invention is as follows: This invention provides a terrain-matching high-precision positioning system with dynamic battlefield environment updates, comprising a multi-source battlefield environment perception module, a dynamic terrain modeling module, a candidate region generation module, a terrain matching evaluation module, a positioning fusion calculation module, and a positioning information evaluation output module, specifically including the following:

[0007] The multi-source battlefield environment perception module collects observation data of the battlefield environment, including satellite positioning signals, radar or laser point clouds, remote sensing images and sensor information, and performs preliminary fusion and filtering to obtain processed observation data.

[0008] The dynamic terrain modeling module constructs a three-dimensional battlefield terrain model based on the processed observation data. It uses a structured particle representation method to uniformly discretize complex terrain and annotate key attributes to generate a dynamic terrain model.

[0009] The candidate region generation module uses sensors to collect the latest battlefield environment observation data, uses dynamic terrain models and the latest battlefield environment observation data to identify and locate candidate regions, filters out the most likely regions, and obtains candidate region information.

[0010] The terrain matching evaluation module compares candidate region information and dynamic terrain models using multimodal constraints and innovative terrain feature evaluation methods to generate a terrain matching score.

[0011] The positioning fusion and solving module integrates candidate region information and terrain matching scores, and performs high-precision positioning calculations based on the particle swarm optimization algorithm to obtain intermediate positioning results.

[0012] The location confidence assessment output module evaluates the confidence level of the intermediate location results and generates the final location information.

[0013] Furthermore, the dynamic terrain modeling module constructs a three-dimensional battlefield terrain model based on the processed observation data. It employs a structured particle representation method to uniformly discretize complex terrain and simultaneously annotates key attributes to generate a dynamic terrain model. Specifically, this includes the following steps:

[0014] Step S1: Data filtering. Based on the processed observation data, the spatial location description information contained in each processed observation data is extracted. The spatial location description information includes the coordinate information of the observation data in three-dimensional space. Using the spatial location description information as a constraint variable, an observation consistency constraint model with spatial location as the core variable is constructed. The multi-constraint fusion idea of ​​factor graph optimization is introduced to remove abnormal observation data and obtain a highly reliable observation set.

[0015] Step S2: Constraint-driven voxel spatial discretization. Based on the high-confidence observation set, spatial location description information is used as the statistical object to perform spatial statistics on the high-confidence observation set, obtain the spatial density of the observation data, and calculate the voxel resolution. The formula used is as follows:

[0016] ;

[0017] in, For voxel resolution, This represents the effective observation density per unit space. This is the scaling adjustment coefficient;

[0018] The three-dimensional spatial region covered by the high-confidence observation set is divided into a grid set composed of regular voxel units, represented as follows: ;

[0019] in, Represents a mesh set, each voxel Its location is determined by the center point of the three-dimensional spatial region it covers;

[0020] Step S3: Structured particle generation. Based on the mesh set, structured terrain particles are generated within each voxel as the basic computational unit of the terrain model. The position of each particle uses the center point of the 3D spatial region covered by its voxel as the initial spatial anchor point. The distribution of particles follows the voxel resolution. The particles inherit the statistical characteristics of the high-confidence observation set within their voxel. Using the mean elevation and local spatial variance of the observation data, a set of structured terrain particles is formed, as shown below:

[0021] ;

[0022] Each structured terrain particle Includes spatial location, mean elevation, and local spatial variance. Spatial location is represented as... The initial value is the voxel center position, and the average elevation is expressed as: The local spatial variance is expressed as , Represents a set of structured terrain particles. Indicates the first A structured terrain particle, Indicates the first Individual unit, express Located in voxels Inside, This indicates that the voxel unit belongs to the entire mesh set. Represents a set of highly reliable observations. Indicates the location of voxels A collection of highly reliable observational data within the database. The actual elevation value of the observed data. For particles The initial elevation is calculated using the average elevation from the voxel observation data. For particles The variance of the elevation of the observed data within the voxel;

[0023] Step S4: Particle attribute annotation based on a scoring mechanism. For the structured terrain particle set, a shadow matching scoring method is introduced. The scoring function is used to calculate the particle scoring results, which are used as the terrain attributes for particle annotation, including elevation, slope, and accessibility weights. The scoring function is defined as follows:

[0024] ;

[0025] in, This indicates the number of observed feature types that participated in the scoring. Represents particles With the A measure of the deviation between observed features of different classes. For the first Weighting coefficients of observed features;

[0026] Step S5: Particle consistency update. Based on the particle scoring results, the position and attributes of the structured terrain particles are adjusted to obtain the updated set of structured terrain particles.

[0027] Step S6: Terrain model generation. The updated set of structured terrain particles is spatially combined to obtain a continuous three-dimensional dynamic terrain model.

[0028] Furthermore, the terrain matching evaluation module compares candidate region information and dynamic terrain models using multimodal constraints and innovative terrain feature evaluation methods, specifically including the following steps:

[0029] Step D1: Candidate region mapping. Based on the location information of the candidate regions, the corresponding spatial range is mapped to the dynamic terrain model, and the corresponding structured terrain particle subset is extracted.

[0030] Step D2: Multimodal feature alignment. The spatial location and elevation features of the observation data within the candidate region are matched one-to-one with the particle attributes within the structured terrain particle subset to obtain the aligned structured terrain particles.

[0031] Step D3: Terrain feature consistency calculation. Based on the aligned structured terrain particle attributes and observation data, calculate the terrain feature consistency index, as shown in the following expression:

[0032] ;

[0033] ;

[0034] in, Indicates consistency in terrain features, This represents the subset of structured terrain particles corresponding to the candidate region. For reference elevation of observation data, The local spatial variance of the particle;

[0035] Step D4: Multi-constraint joint matching score. Calculate the candidate region matching score based on terrain feature consistency. The formula used is as follows:

[0036] ;

[0037] in, Indicates the candidate region matching score. This is the weighting adjustment coefficient. Indicates the inherent constraint index of the candidate region;

[0038] Step D5: Output matching results. Sort the candidate regions by matching scores and output the region with the highest score as the terrain matching evaluation result.

[0039] The beneficial effects achieved by the present invention using the above solution are as follows:

[0040] (1) This invention introduces structured terrain particle representation and constraint-driven voxel space discretization method to construct a dynamically updatable three-dimensional battlefield terrain model based on high-reliability observation data, thereby realizing the continuous updating of the terrain model with changes in the battlefield environment. This effectively solves the problem that the terrain model is static and difficult to reflect real-time battlefield environment changes in the prior art, and improves the consistency between the terrain model and the actual environment.

[0041] (2) In the process of terrain matching evaluation, this invention combines multimodal observation constraints and terrain feature consistency evaluation mechanism to jointly match and score candidate areas with dynamic terrain models, which reduces the risk of mismatch caused by single feature matching and significantly improves the accuracy of terrain matching, the stability of positioning results and the reliability of positioning. Attached Figure Description

[0042] Figure 1 This is a schematic diagram of a terrain-matching high-precision positioning system that dynamically updates the battlefield environment, as proposed in this invention.

[0043] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. Detailed Implementation

[0044] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0045] Example 1, see Figure 1 In this embodiment, the application scenario is mobile target positioning in a complex battlefield environment. In mountainous, hilly, or battlefield areas with obstruction and interference, the position of the mobile platform is located with high precision. During the positioning process, multi-source observation data of the battlefield environment is first obtained through a multi-source battlefield environment perception module. The observation data includes satellite positioning signals, radar or laser point clouds, remote sensing images, and sensor information. The collected multi-source observation data is processed synchronously and abnormal data is removed to obtain the processed observation dataset.

[0046] The dynamic terrain modeling module performs three-dimensional terrain modeling of the battlefield area, maps the processed observation dataset to a unified spatial coordinate system, and uses structured particle representation to discretize the battlefield terrain.

[0047] When a target needs to be located, the latest collected battlefield environment observation data is used in conjunction with a dynamic terrain model to analyze and constrain the possible spatial range of the target, generate candidate areas for location, and exclude areas that are inconsistent with the current observations, thereby narrowing the location search space.

[0048] For each candidate region, a terrain matching evaluation is performed. The observed features of the candidate region are compared with the terrain information in the dynamic terrain model. The matching degree of the candidate region is calculated by the multimodal constrained terrain feature evaluation method, and the corresponding terrain matching score is obtained.

[0049] After obtaining the matching scores of candidate regions, the system integrates the candidate region information and terrain matching scores, and uses a particle swarm optimization algorithm to iteratively search and optimize the target location to obtain intermediate positioning results. Finally, the system performs confidence analysis on the intermediate positioning results to evaluate the reliability of the positioning results, and outputs the final positioning result and its corresponding confidence information.

[0050] Example 2, based on the above examples, describes a dynamic terrain modeling module that constructs a 3D battlefield terrain model based on processed observation data. It employs a structured particle representation method to uniformly discretize complex terrain and simultaneously annotates key attributes to generate a dynamic terrain model. The specific steps include:

[0051] Step S1: Data filtering. Based on the processed observation data, the spatial location description information contained in each processed observation data is extracted. The spatial location description information includes the coordinate information of the observation data in three-dimensional space. Using the spatial location description information as a constraint variable, an observation consistency constraint model with spatial location as the core variable is constructed. The multi-constraint fusion idea of ​​factor graph optimization is introduced to remove abnormal observation data and obtain a highly reliable observation set.

[0052] Step S2: Constraint-driven voxel spatial discretization. Based on the high-confidence observation set, spatial location description information is used as the statistical object to perform spatial statistics on the high-confidence observation set, obtain the spatial density of the observation data, and calculate the voxel resolution. The formula used is as follows:

[0053] ;

[0054] in, For voxel resolution, This represents the effective observation density per unit space. This is the scaling adjustment coefficient;

[0055] The three-dimensional spatial region covered by the high-confidence observation set is divided into a grid set composed of regular voxel units, represented as follows: ;

[0056] in, Represents a mesh set, each voxel Its location is determined by the center point of the three-dimensional spatial region it covers;

[0057] Step S3: Structured particle generation. Based on the mesh set, structured terrain particles are generated within each voxel as the basic computational unit of the terrain model. The position of each particle uses the center point of the 3D spatial region covered by its voxel as the initial spatial anchor point. The distribution of particles follows the voxel resolution. The particles inherit the statistical characteristics of the high-confidence observation set within their voxel. Using the mean elevation and local spatial variance of the observation data, a set of structured terrain particles is formed, as shown below:

[0058] ;

[0059] Each structured terrain particle Includes spatial location, mean elevation, and local spatial variance. Spatial location is represented as... The initial value is the voxel center position, and the average elevation is expressed as: The local spatial variance is expressed as , Represents a set of structured terrain particles. Indicates the first A structured terrain particle, Indicates the first Individual unit, express Located in voxels Inside, This indicates that the voxel unit belongs to the entire mesh set. Represents a set of highly reliable observations. Indicates the location of voxels A collection of highly reliable observational data within the database. The actual elevation value of the observed data. For particles The initial elevation is calculated using the average elevation from the voxel observation data. For particles The variance of the elevation of the observed data within the voxel;

[0060] Step S4: Particle attribute annotation based on a scoring mechanism. For the structured terrain particle set, a shadow matching scoring method is introduced. The scoring function is used to calculate the particle scoring results, which are used as the terrain attributes for particle annotation, including elevation, slope, and accessibility weights. The scoring function is defined as follows:

[0061] ;

[0062] in, This indicates the number of observed feature types that participated in the scoring. Represents particles With the A measure of the deviation between observed features of different classes. For the first Weighting coefficients of observed features;

[0063] Step S5: Particle consistency update. Based on the particle scoring results, the position and attributes of the structured terrain particles are adjusted to obtain the updated set of structured terrain particles.

[0064] Step S6: Terrain model generation. The updated set of structured terrain particles is spatially combined to obtain a continuous three-dimensional dynamic terrain model.

[0065] In this embodiment, the code used is as follows:

[0066] import numpy as np

[0067] from collections import defaultdict

[0068] # =========================

[0069] # Basic Data Structures

[0070] # =========================

[0071] class Observation:

[0072] def __init__(self, x, y, z, features=None):

[0073] self.x = x

[0074] self.y = y

[0075] self.z = z

[0076] self.features = features or {}

[0077] class TerrainParticle:

[0078] def __init__(self, x, y, z_mean, z_var):

[0079] self.x = x

[0080] self.y = y

[0081] self.z_mean = z_mean

[0082] self.z_var = z_var

[0083] self.score = 0.0

[0084] self.slope = 0.0

[0085] self.traversability = 0.0

[0086] # =========================

[0087] # Dynamic terrain modeling module

[0088] # =========================

[0089] class DynamicTerrainModel:

[0090] def __init__(self, alpha=1.0):

[0091] self.alpha = alpha

[0092] # -------------------------

[0093] # S1: Data screening

[0094] # -------------------------

[0095] def filter_observations(self, observations, distance_thresh=2.0):

[0096] """

[0097] Outlier removal based on the spatial consistency of factor graphs

[0098] """

[0099] high_confidence = []

[0100] for obs in observations:

[0101] residuals = []

[0102] For other observations:

[0103] If obs is other:

[0104] continue

[0105] d = np.linalg.norm(

[0106] np.array([obs.x, obs.y, obs.z]) -

[0107] np.array([other.x, other.y, other.z]) )

[0109] residuals.append(d)

[0110] # Factor Residual Statistics

[0111] mean_residual = np.mean(residuals)

[0112] if mean_residual < distance_thresh:

[0113] high_confidence.append(obs)

[0114] return high_confidence

[0115] # -------------------------

[0116] # S2: Voxel space discretization

[0117] # -------------------------

[0118] def voxelize_space(self, observations):

[0119] """

[0120] Adaptive Voxel Resolution Calculation and Spatial Discretization

[0121] """

[0122] coords = np.array([[ox, oy, oz] for o in observations])

[0123] volume = np.ptp(coords, axis=0).prod()

[0124] rho = len(observations) / (volume + 1e-6)

[0125] h = self.alpha / rho # Voxel resolution

[0126] voxels = defaultdict(list)

[0127] for o in observations:

[0128] key = (

[0129] int(ox / / h),

[0130] int(oy / / h),

[0131] int(oz / / h) )

[0133] voxels[key].append(o)

[0134] return voxels, h

[0135] # -------------------------

[0136] # S3: Structured Terrain Particle Generation

[0137] # -------------------------

[0138] def generate_structured_particles(self, voxels, h):

[0139] particles = []

[0140] for voxel_key, obs_list in voxels.items():

[0141] xs = np.array([o.x for o in obs_list])

[0142] ys = np.array([o.y for o in obs_list])

[0143] zs = np.array([o.z for o in obs_list])

[0144] x_c = (voxel_key[0] + 0.5) * h

[0145] y_c = (voxel_key[1] + 0.5) * h

[0146] z_mean = np.mean(zs)

[0147] z_var = np.var(zs)

[0148] particle = TerrainParticle(x_c, y_c, z_mean, z_var)

[0149] particles.append(particle)

[0150] return particles

[0151] # -------------------------

[0152] # S4: Scoring and Attribute Annotation

[0153] # -------------------------

[0154] def score_particles(self, particles, observations, weights):

[0155] """

[0156] Generalized implementation of the shadow matching scoring idea

[0157] """

[0158] for p in particles:

[0159] score = 0.0

[0160] for obs in observations:

[0161] dz = abs(p.z_mean - obs.z)

[0162] slope_penalty = dz / (np.sqrt(p.z_var) + 1e-6)

[0163] score += (

[0164] weights["height"] * dz +

[0165] weights["slope"] * slope_penalty )

[0167] p.score = np.exp(-score / max(len(observations), 1))

[0168] p.slope = np.sqrt(p.z_var)

[0169] p.traversability = 1.0 / (1.0 + p.slope)

[0170] # -------------------------

[0171] # S5: Particle Consistency Update

[0172] # -------------------------

[0173] def update_particles(self, particles, lr=0.2):

[0174] """

[0175] Spatial consistency update based on score

[0176] """

[0177] for p in particles:

[0178] adjustment = lr * (p.score - 0.5)

[0179] p.z_mean += adjustment

[0180] p.traversability = np.clip(p.traversability + adjustment, 0, 1)

[0181] # -------------------------

[0182] # S6: Terrain Model Generation

[0183] # -------------------------

[0184] def build_terrain_model(self, particles):

[0185] """

[0186] Output continuous terrain model

[0187] """

[0188] terrain = []

[0189] for p in particles:

[0190] terrain.append({

[0191] "x": px,

[0192] "y": py,

[0193] "z": p.z_mean,

[0194] "slope": p.slope,

[0195] "traversability": p.traversability

[0196] })

[0197] return terrain.

[0198] Example 3, based on the above examples, describes a terrain matching and evaluation module that compares candidate region information and dynamic terrain models using multimodal constraints and innovative terrain feature evaluation methods. Specifically, it includes the following steps:

[0199] Step D1: Candidate region mapping. Based on the location information of the candidate regions, the corresponding spatial range is mapped to the dynamic terrain model, and the corresponding structured terrain particle subset is extracted.

[0200] Step D2: Multimodal feature alignment. The spatial location and elevation features of the observation data within the candidate region are matched one-to-one with the particle attributes within the structured terrain particle subset to obtain the aligned structured terrain particles.

[0201] Step D3: Terrain feature consistency calculation. Based on the aligned structured terrain particle attributes and observation data, calculate the terrain feature consistency index, as shown in the following expression:

[0202] ;

[0203] ;

[0204] in, Indicates consistency in terrain features, This represents the subset of structured terrain particles corresponding to the candidate region. For reference elevation of observation data, The local spatial variance of the particle;

[0205] Step D4: Multi-constraint joint matching score. Calculate the candidate region matching score based on terrain feature consistency. The formula used is as follows:

[0206] ;

[0207] in, Indicates the candidate region matching score. This is the weighting adjustment coefficient. Indicates the inherent constraint index of the candidate region;

[0208] Step D5: Output matching results. Sort the candidate regions by matching scores and output the region with the highest score as the terrain matching evaluation result.

[0209] In this embodiment, candidate region The corresponding structured terrain particle subset representation is as follows:

[0210] ;

[0211] The mean elevation and local spatial variance of the particles are as follows:

[0212] ;

[0213] ;

[0214] ;

[0215] The observation reference elevation is expressed as follows:

[0216] ;

[0217] The formula for calculating the local standard deviation of a particle is:

[0218] ;

[0219] The formula for calculating the consistency of terrain features is:

[0220] ;

[0221] Substituting the values ​​for each particle, the results are as follows:

[0222] ; ; ;

[0223] Candidate region The matching score formula is:

[0224] ;

[0225] Pick , Substituting into the calculation, we get:

[0226] ;

[0227] Other candidate region matching scores are:

[0228] ;

[0229] Finally, based on the matching scores, the candidate regions with the highest matching scores are output. As a result of terrain matching evaluation.

[0230] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0231] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

[0232] The present invention and its embodiments have been described above. This description is not restrictive, and the accompanying drawings are only one embodiment of the present invention; the actual structure is not limited thereto. In conclusion, if those skilled in the art are inspired by this description and design similar structures and embodiments without departing from the spirit of the invention, such designs should fall within the protection scope of the present invention.

Claims

1. A terrain-matching high-precision positioning system with dynamic battlefield environment updates, characterized in that: It includes a multi-source battlefield environment perception module, a dynamic terrain modeling module, a candidate region generation module, a terrain matching evaluation module, a positioning fusion calculation module, and a positioning information evaluation output module, specifically including the following: The multi-source battlefield environment perception module collects observation data of the battlefield environment, performs preliminary fusion and filtering, and obtains processed observation data. The dynamic terrain modeling module constructs a three-dimensional battlefield terrain model based on the processed observation data. It uses a structured particle representation method to uniformly discretize complex terrain and annotate key attributes to generate a dynamic terrain model. The candidate region generation module uses sensors to collect the latest battlefield environment observation data, uses dynamic terrain models and the latest battlefield environment observation data to identify and locate candidate regions, filters out the most likely regions, and obtains candidate region information. The terrain matching evaluation module compares candidate region information and dynamic terrain models using multimodal constraints and innovative terrain feature evaluation methods to generate a terrain matching score. The positioning fusion solution module integrates candidate region information and terrain matching score, and performs high-precision positioning calculation based on particle swarm optimization algorithm to obtain intermediate positioning results. The location confidence assessment output module evaluates the confidence level of the intermediate location results and generates the final location information.

2. The terrain-matching high-precision positioning system for dynamic battlefield environment updates according to claim 1, characterized in that: The dynamic terrain modeling module constructs a three-dimensional battlefield terrain model based on processed observation data. It uses a structured particle representation method to uniformly discretize complex terrain and annotate key attributes to generate a dynamic terrain model. The specific steps include: Step S1: Data filtering. Based on the processed observation data, the spatial location description information contained in each processed observation data is extracted. The spatial location description information includes the coordinate information of the observation data in three-dimensional space. Using the spatial location description information as a constraint variable, an observation consistency constraint model with spatial location as the core variable is constructed. The multi-constraint fusion idea of ​​factor graph optimization is introduced to remove abnormal observation data and obtain a highly reliable observation set. Step S2: Constraint-driven voxel spatial discretization. Based on the high-confidence observation set, spatial location description information is used as the statistical object to perform spatial statistics on the high-confidence observation set, obtain the spatial density of the observation data, and calculate the voxel resolution. The formula used is as follows: ； in, For voxel resolution, This represents the effective observation density per unit space. This is the scaling adjustment coefficient; The three-dimensional spatial region covered by the high-confidence observation set is divided into a grid set composed of regular voxel units, represented as follows: ; in, Represents a mesh set, each voxel Its location is determined by the center point of the three-dimensional spatial region it covers; Step S3: Structured particle generation. Based on the mesh set, structured terrain particles are generated within each voxel as the basic computational unit of the terrain model. The position of each particle uses the center point of the 3D spatial region covered by its voxel as the initial spatial anchor point. The distribution of particles follows the voxel resolution. The particles inherit the statistical characteristics of the high-confidence observation set within their voxel. Using the mean elevation and local spatial variance of the observation data, a set of structured terrain particles is formed, as shown below: ； Each structured terrain particle Includes spatial location, mean elevation, and local spatial variance. Spatial location is represented as... The initial value is the voxel center position, and the average elevation is expressed as: The local spatial variance is expressed as , Represents a collection of structured terrain particles. Indicates the first A structured terrain particle, Indicates the first Individual unit, express Located in voxels Inside, This indicates that the voxel unit belongs to the entire mesh set. Represents a set of highly reliable observations. Indicates the location of voxels A collection of highly reliable observational data within the database. The actual elevation value of the observed data. For particles The initial elevation is calculated using the average elevation from the voxel observation data. For particles The variance of the elevation of the observed data within the voxel; Step S4: Particle attribute annotation based on a scoring mechanism. For the structured terrain particle set, a shadow matching scoring method is introduced. The scoring function is used to calculate the particle scoring results, which are used as the terrain attributes for particle annotation, including elevation, slope, and accessibility weights. The scoring function is defined as follows: ； in, This indicates the number of observed feature types that participated in the scoring. Represents particles With the A measure of the deviation between observed features of different classes. For the first Weighting coefficients of observed features; Step S5: Particle consistency update. Based on the particle scoring results, the position and attributes of the structured terrain particles are adjusted to obtain the updated set of structured terrain particles. Step S6: Terrain model generation. The updated set of structured terrain particles is spatially combined to obtain a continuous three-dimensional dynamic terrain model.

3. The terrain-matching high-precision positioning system for dynamic battlefield environment updates according to claim 2, characterized in that: The terrain matching and evaluation module compares candidate region information and dynamic terrain models using multimodal constraints and innovative terrain feature evaluation methods, specifically including the following steps: Step D1: Candidate region mapping. Based on the location information of the candidate regions, the corresponding spatial range is mapped to the dynamic terrain model, and the corresponding structured terrain particle subset is extracted. Step D2: Multimodal feature alignment. The spatial location and elevation features of the observation data within the candidate region are matched one-to-one with the particle attributes within the structured terrain particle subset to obtain the aligned structured terrain particles. Step D3: Calculate terrain feature consistency. Based on the aligned structured terrain particle attributes and observation data, calculate the terrain feature consistency index. Step D4: Multi-constraint joint matching score, calculate the candidate region matching score based on the consistency of terrain features; Step D5: Output matching results. Sort the candidate regions by matching scores and output the region with the highest score as the terrain matching evaluation result.

Images