A method for monitoring surface changes and rapid updating of topographic data

Abstract

The present application relates to the field of 3D modeling technology, in particular to a kind of surface change monitoring and topographic data fast updating method, comprising the following steps: based on multi-period three-dimensional terrain data, the triangular mesh covering monitoring area is constructed, the elevation value of multi-period terrain data is compared by vertex by vertex, and the vertex time change rate set is generated.The present application can capture the spatial and temporal characteristics of topographic change by constructing the triangular mesh covering monitoring area in multi-period three-dimensional terrain data, comparing the elevation value by vertex by vertex to generate the time change rate set, and combining the single-period elevation field scanning connected branch to generate and merge event to obtain structural saliency score;On this basis, the basic triangular mesh contractible edge is traversed, the time change rate of two end vertices is extracted, the structural saliency score is calculated to calculate the comprehensive simplification generation value, and the ordered simplification operation sequence is formed in the order of generation value, which can realize the optimal contraction order under the premise of retaining key topographic features.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of 3D modeling, and particularly relates to a surface change monitoring and terrain data rapid updating method. BACKGROUND

[0002] The technical field of 3D modeling relates to a comprehensive technical system for digital three-dimensional reconstruction and expression of objects, scenes or terrains in the real world by using computer graphics and spatial data processing technology.

[0003] In the prior art, static data is often used for global reconstruction in the 3D modeling process, which leads to frequent full calculation of the entire model when dealing with dynamic changes in the terrain, thereby increasing the calculation amount and data processing pressure. In multi-period data comparison, the update inevitably affects the unchanged areas, resulting in low resource utilization and long update cycle. In the aspect of mesh simplification, the prior art more relies on a single index of geometric shape or the number of facets for edge contraction sorting, which is easy to lose important details in the key change area. Therefore, improvement is needed. SUMMARY

[0004] The purpose of the present application is to solve the shortcomings in the prior art and provide a surface change monitoring and terrain data rapid updating method.

[0005] In order to achieve the above purpose, the present application adopts the following technical scheme, a surface change monitoring and terrain data rapid updating method, comprising the following steps:

[0006] Based on multi-period three-dimensional terrain data, a triangular mesh covering the monitoring area is constructed, a vertex time change rate set is generated by comparing the elevation values of multi-period terrain data vertex by vertex, a single-period elevation field is taken as a scalar field, and a vertex structure significance score set is obtained by scanning the scalar field from low to high to track the generation and merging events of connected branches;

[0007] Based on the vertex time change rate set and the vertex structure significance score set, each contractible edge of the basic triangular mesh is traversed, and a comprehensive simplification generation value is calculated, the comprehensive simplification generation value of each edge is arranged, and an ordered simplification operation sequence is established;

[0008] Based on the ordered simplification operation sequence, vertex pair contraction is sequentially executed from the edge with the lowest comprehensive simplification generation value, a multi-level simplification record set is formed, different mesh simplification degrees are organized into a hierarchical data structure according to the information in the multi-level simplification record set, and an adaptive space-time hierarchical detail model is established;

[0009] Based on the new terrain data and the adaptive spatio-temporal hierarchical detail model, a monitoring area where the ground surface changes occurs is identified, the time change rate of the vertex in the monitoring area is recalculated and the integrated simplification cost is updated, the updated integrated simplification cost is compared with the original simplification cost, and the affected level in the adaptive spatio-temporal hierarchical detail model is locally reconstructed to generate a dynamically updated terrain level structure.

[0010] Preferably, the step of obtaining the vertex time change rate set comprises:

[0011] Based on multi-period three-dimensional terrain data, the coordinate fields X and Y and the elevation field Z are analyzed, the coordinate reference and the elevation zero base are unified, the effective points are clipped according to the monitoring area boundary, the adjacent points are connected into non-intersecting triangular patches according to the spatial proximity relationship, and the hanging points are removed to generate a triangular mesh;

[0012] Based on the triangular mesh, the elevation value sequence of the same spatial position in the multi-period three-dimensional terrain data is retrieved according to the vertex number, the missing marks and repeated time stamps are removed, the absolute value of the elevation value difference between adjacent time points is calculated, and the average elevation variation amplitude of each vertex is obtained by averaging the full sequence;

[0013] Based on the average elevation variation amplitude of each vertex, the adjacent time point recording interval is read and unified to a fixed time unit, the average elevation variation amplitude is converted into a rate according to the time unit and the vertex number index is attached, the vertex time change rate set is generated by sequentially collecting the identified entries according to the vertex number order.

[0014] Preferably, the step of obtaining the vertex structure significance score set comprises:

[0015] Based on a single-period elevation field, a threshold list arranged in ascending order of elevation value is generated, the threshold values are set in sequence, and the triangular mesh vertices with an elevation value not greater than the threshold value are extracted, the adjacent vertices are aggregated into connected branches according to the connection relationship of the triangular mesh edges, the first appearing connected branch is recorded as an appearance event and the appearance threshold value is registered, and when two connected branches share at least one edge at the boundary, a merging event is recorded and the merging threshold value is registered, and the triangular patch area covered by each connected branch at each threshold value is accumulated during the entire scanning process to form a connected branch event record containing the appearance threshold value, the merging threshold value and the area sequence.

[0016] According to the connected branch event record, the structure significance evaluation value is calculated;

[0017] Based on the structure significance evaluation value, the corresponding local minimum value vertex is located when each connected branch appears, the structure significance evaluation value of the connected branch is written into the record of the local minimum value vertex, and for the case where multiple structure significance evaluation value records exist for the same local minimum value vertex, the maximum value is retained as the only entry to generate the vertex structure significance score set.

[0018] Preferably, the step of obtaining the integrated simplification cost value is:

[0019] Based on the set of vertex time variation rates and the set of vertex structural significance scores, each contractible edge of the base triangular mesh is traversed, and the time variation rate and the structural significance score of the vertex at each end of the contractible edge are extracted in turn. The mean and the standard deviation of the time variation rates of all the vertices in the monitoring area and the mean and the standard deviation of the structural significance scores of all the vertices are counted and monitored. The contractible edge is stored in association with the attribute values of the two end vertices and the overall statistics, and a contractible edge attribute statistics list is formed.

[0020] According to the contractible edge attribute statistics list, the integrated simplification cost value is calculated.

[0021] Preferably, the step of obtaining the ordered simplification operation sequence is:

[0022] Based on the integrated simplification cost value, the corresponding integrated simplification cost value is registered for each contractible edge, and the contractible edges are arranged in ascending order of the integrated simplification cost value to form an ordered simplification operation sequence.

[0023] Preferably, the step of obtaining the multi-level simplification record set is:

[0024] Based on the ordered simplification operation sequence, vertex pair contraction is performed in order from the contractible edge with the smallest integrated simplification cost value, and the index of the removed vertex, the identification of the removed edge, and the identification of the removed triangular patch are recorded in turn. The three-dimensional coordinates of the newly generated vertex and the replacement relationship are recorded, and the adjacent connection and the normal are updated synchronously to form a multi-level simplification record set.

[0025] Preferably, the step of obtaining the adaptive spatio-temporal hierarchical detail model is:

[0026] Based on the multi-level simplification record set, the mesh snapshot after each contraction and the corresponding integrated simplification cost value are extracted. The mesh snapshots are binned and archived according to a set of pre-set integrated simplification cost threshold ranges. The vertex set, the edge set, and the triangular patch set in the snapshots are summarized for each threshold range, and an inter-layer parent-child reference is established. A hierarchical data structure is generated.

[0027] Based on the hierarchical data structure, the corresponding integrated simplification cost threshold range and the mesh snapshot of each layer are read. The inter-layer parent-child reference and the rollback pointer are established, and the contraction sequence bit to the hierarchical mapping and the time index are solidified. The metadata and the positioning index of each layer are integrated to form an adaptive spatio-temporal hierarchical detail model.

[0028] Preferably, the step of obtaining the dynamically updated terrain hierarchical structure is:

[0029] Based on the new period of terrain data and the adaptive spatio-temporal hierarchical detail model, the elevation values at the same position in the new period of terrain data and the model are compared one by one according to the monitoring area division unit, the unit with an elevation difference exceeding a preset change threshold is marked as a land surface change unit, the spatial range of the land surface change unit is extracted and adjacent units are merged to form a monitoring area, and a monitoring area boundary set is generated;

[0030] Based on the monitoring area boundary set, all vertices in the monitoring area are located, and a historical elevation sequence is re-read, missing values and repeated time stamps are removed, the absolute value mean of the time-adjacent elevation difference is calculated as a time change rate, the existing structure significance score is combined to recalculate the comprehensive simplification generation value, the original comprehensive simplification generation value is compared one by one, and the edges with value changes are marked to generate an affected edge list;

[0031] Based on the affected edge list, the arrangement position of the affected edge in the ordered simplification operation sequence is updated, the level where the affected edge is located in the adaptive spatio-temporal hierarchical detail model is searched, the associated local grid range is located, the invalid geometric elements are removed, and the vertices, edges and triangular facets at the corresponding positions are regenerated according to the new comprehensive simplification generation value to form a dynamically updated terrain hierarchical structure.

[0032] Compared with the prior art, the advantages and positive effects of the present application are that:

[0033] The present application can capture the spatial and temporal characteristics of terrain changes by constructing a triangular mesh covering the monitoring area in multi-period three-dimensional terrain data, generating a time change rate set by comparing elevation values point by point, and combining single-period elevation field scanning to generate and merge events to obtain a structure significance score. On this basis, the contractible edges of the basic triangular mesh are traversed, the time change rate and the structure significance score of the two end vertices are extracted, the comprehensive simplification generation value is calculated, and an ordered simplification operation sequence is formed by arranging according to the generation value. The optimal contraction order can be realized on the premise of retaining key terrain features. Starting from the edge with the lowest self-comprehensive simplification generation value, the vertex pair contraction is sequentially executed and the geometric changes are recorded to establish a multi-level simplification record set and organize a hierarchical data structure according to different simplification degrees, so that the model has consistency and traceability at multiple scales. In the update of the new period of terrain data, only the vertices in the monitoring area are recalculated and compared in terms of time change rate and comprehensive simplification cost, and the local grid reconstruction is combined to realize targeted update, effectively reducing the data recalculation and reconstruction workload of irrelevant areas, improving the update efficiency and maintaining the geometric accuracy and timeliness of the overall model. The fusion of the time change rate and the structure significance score improves the targeting of feature retention, the ordering and contraction based on the comprehensive simplification generation value ensure the global priority of structure optimization, and the localized dynamic update mechanism reduces the operation redundancy, realizing efficient maintenance of large-scale terrain data. BRIEF DESCRIPTION OF DRAWINGS

[0034] Figure 1 A schematic diagram of the steps of the present application. DETAILED DESCRIPTION

[0035] In order to make the purposes, technical solutions and advantages of the present application clearer, the present application is further described in detail below in combination with the drawings and examples. It should be understood that the specific examples described herein are only used to explain the present application and do not limit the present application.

[0036] Referring to Figure 1 The present application provides a technical solution, a ground surface change monitoring and terrain data rapid updating method, comprising the following steps:

[0037] Based on multi-period three-dimensional terrain data, a triangular mesh covering the monitoring area is constructed, a vertex time change rate set is generated by comparing the elevation values of the multi-period terrain data vertex by vertex, a single-period elevation field is taken as a scalar field, and a vertex structure significance score set is obtained by scanning the scalar field from low to high to track the generation and merging events of connected branches;

[0038] Based on the vertex time change rate set and the vertex structure significance score set, each contractible edge of the basic triangular mesh is traversed, and a comprehensive simplification generation value is calculated, all contractible edges are arranged according to the comprehensive simplification generation value of each edge, and an ordered simplification operation sequence is established;

[0039] Based on the ordered simplification operation sequence, vertex pair contraction is sequentially executed from the edge with the lowest comprehensive simplification generation value, a multi-level simplification record set is formed, different grid levels of different simplification degrees are organized into a hierarchical data structure according to the information in the multi-level simplification record set, and an adaptive spatio-temporal hierarchical detail model is established;

[0040] Based on the new-period terrain data and the adaptive spatio-temporal hierarchical detail model, the monitoring area where the ground surface changes occur is identified, the time change rate of the vertex in the monitoring area is recalculated and the comprehensive simplification generation value is updated, the updated comprehensive simplification generation value is compared with the original simplification generation value, and the affected levels in the adaptive spatio-temporal hierarchical detail model are locally reconstructed to generate a dynamically updated terrain level structure.

[0041] The acquisition step of the vertex time change rate set is:

[0042] Based on multi-period three-dimensional terrain data, the coordinate fields X and Y and the elevation field Z are analyzed, the coordinate reference and the elevation zero base are unified, the effective points are clipped according to the monitoring area boundary, the adjacent points are connected into non-intersecting triangular patches according to the spatial proximity relationship and the hanging points are removed, and a triangular mesh is generated;

[0043] Based on the triangular mesh, the elevation value sequence of the same spatial position in the multi-period three-dimensional terrain data is retrieved according to the vertex number, the missing mark and the repeated timestamp are removed, the absolute value of the elevation value difference of adjacent time points is calculated and the average of the whole sequence is taken, and the average elevation variation amplitude of each vertex is obtained;

[0044] Based on the average elevation variation amplitude of each vertex, the adjacent time point recording interval is read and unified to a fixed time unit, the average elevation variation amplitude is converted into a rate according to the time unit and the vertex number index is added, and the average elevation variation amplitude is summarized into an identified entry set according to the vertex number sequence to generate a vertex time change rate set.

[0045] Specifically, based on the multi-period three-dimensional terrain data, first, through the preset data interface specification, the coordinate field X, Y and the elevation field Z in the original data file are automatically identified and parsed, and the data file formats include but are not limited to laser radar point cloud data format (LAS) or comma separated value text file (CSV), wherein the fields are located by column title or preset column index (for example, the 1st, 2nd and 3rd columns correspond to X, Y and Z respectively), then, in order to ensure that all data are compared under the same spatial reference, the system converts the coordinates of all points to a project pre-defined coordinate reference system, such as a certain specific zone of Universal Transverse Mercator projection (UTM) under WGS 84 coordinate system, at the same time, all elevation values are unified to the same elevation reference by adding or subtracting an offset, then, the system loads the monitoring area boundary in the pre-defined vector polygon format, and performs a ray casting algorithm for each data point to make a spatial judgment, only the points whose coordinates are located inside the polygon are retained as valid points, for the valid point set selected by screening, the system uses the constrained Delaunay triangulation algorithm to connect the spatially adjacent points according to the Euclidean distance between the points to construct a continuous surface composed of non-intersecting triangular patches, after the triangulation is completed, in order to improve the mesh quality, the system will perform a mesh cleaning operation, specifically, the quality factor of each triangular patch is calculated, for example, the ratio of its longest side to the corresponding height is calculated, and a quality threshold is set, which is dynamically set according to the average point spacing of the monitoring area point cloud, for example, if the average point spacing is 0.5 meters, the threshold can be set to 20, any long and narrow triangle with a ratio value exceeding 20 is regarded as a false image produced by calculation noise or boundary effect in the splitting process and is removed, after removing these poor quality triangles, the isolated vertices or vertices connected with only one or two edges, i.e. hanging points, which no longer constitute part of any triangular patch, are also removed, and the triangular mesh is generated.

[0046] Based on the triangular mesh, the system assigns a unique integer number to each vertex in the triangular mesh, and uses its X and Y coordinates in the unified coordinate system to search in the original multi-period 3D terrain data to collect all elevation records of the spatial position at different time points. In order to accurately match, a small spatial tolerance range is defined around the X and Y coordinates of each vertex when searching, and the tolerance radius is usually set to half of the nominal spatial resolution of the input data. For example, for terrain data with a resolution of 1 meter, the tolerance radius is set to 0.5 meters. All data points falling within this circular neighborhood are considered to be related to the vertex, thereby generating an original elevation value sequence containing multiple time stamps and corresponding elevation values for each vertex. Next, the system cleans up the elevation value sequence of each vertex. First, it checks and removes all missing data points marked with a preset special value (such as -9999) in the sequence. Then, the system checks the time stamps in the sequence. If it finds that there are multiple different elevation values corresponding to the same time stamp, it retains one according to a preset rule, such as retaining the first value appearing in the time sequence or taking the average value. After completing the data cleaning, for each vertex with at least two valid elevation records, the system calculates the absolute value of the elevation value change between each two adjacent time points along the time sequence, i.e. calculates all

[0047] |Z(t i:1 )-Z(t i )|, where Z(t i ) is the elevation value at the i-th time point. Finally, the system accumulates all the calculated absolute values of the elevation differences and divides the total number of differences to obtain the average value of the vertex. For example, the elevations of a vertex at three periods are 12.5 meters, 12.8 meters and 12.6 meters, respectively. The absolute value sequence of the elevation difference is |12.8-12.5| = 0.3 meters and |12.6-12.8| = 0.2 meters, and the average elevation variation amplitude is (0.3+0.2) / 2 = 0.25 meters. The average elevation variation amplitude of each vertex is obtained.

[0048] Based on the average elevation variation range of each vertex, the system first reads the timestamp sequence associated with the elevation value sequence of each vertex, calculates the time interval between each two adjacent record points, and then converts all the time intervals into a preset fixed time unit, such as converting all time intervals recorded in hours, minutes or seconds into "days" for standardized rate calculation, for example, a 48-hour interval will be converted to 2 days. After completing the unit conversion, the system calculates the average value of all adjacent time intervals of each vertex to obtain a value representing the average sampling period of the vertex data. Next, the system divides the average elevation variation range of each vertex obtained in the previous step by the average time interval corresponding to the vertex, thereby converting the elevation variation range into a standardized change rate, for example, if the average elevation variation range of a vertex is 0.25 meters and the average sampling interval is calculated to be 90 days, the time change rate is calculated as 0.25 meters divided by 90 days, which is approximately 0.00278 meters / day. This rate value is then associated with the unique index of the vertex to form a data pair, such as (vertex number: 54321, change rate: 0.00278). Finally, the system arranges and summarizes all vertex number and rate data pairs calculated by the system in ascending order of vertex number to form a structured identified entry set that records the dynamic change intensity of each grid vertex in the monitoring area, generating a vertex time change rate set.

[0049] The acquisition step of the vertex structure significance score set is:

[0050] Based on the single-phase elevation field, a threshold list is generated in ascending order of elevation value, and the threshold is set in sequence and the triangular mesh vertices with elevation values not greater than the threshold are extracted. Adjacent vertices are aggregated into connected branches according to the connection relationship of triangular mesh edges. The first occurrence of a connected branch is recorded as an occurrence event and the occurrence threshold is registered. At the same time, when two connected branches share at least one edge at the boundary, a merging event is recorded and the merging threshold is registered. During the entire scanning process, the area of triangular facets covered by each connected branch at each threshold is accumulated to form a connected branch event record containing occurrence threshold, merging threshold and area sequence.

[0051] According to the connected branch event record, the structure significance evaluation value is calculated, and the calculation formula is:

[0052]

[0053] Where a is the occurrence threshold in the occurrence event record of the connected branch, b is the merging threshold in the connected branch event record, Φ max is the maximum elevation value of the single-phase elevation field, Φ min is the minimum elevation value of the single-phase elevation field, Vc A is the volume of the connected component at threshold value a, and A is the volume of the connected component at threshold value b. total H is the total horizontal projection area of the monitoring region. unit P is the standard height unit. sig is the structural saliency evaluation value.

[0054] Based on the structural saliency evaluation value, the corresponding local minimum vertex is located when each connected component appears, and the structural saliency evaluation value of the connected component is written into the record of the local minimum vertex. In the case of multiple structural saliency evaluation value records for the same local minimum vertex, the maximum value is retained as the only entry, and a vertex structural saliency score set is generated.

[0055] Specifically, based on a single-phase elevation field, first, all unique Z coordinate elevation values are extracted from all the triangular mesh vertices that make up the elevation field, and these unique values are arranged in ascending order to generate a discrete threshold list. Each value in this list will be used as an independent scanning threshold. Then, the system starts from the smallest value in the threshold list and sets the values in the list as the current processing threshold one by one. For each current processing threshold, the system will traverse all the vertices in the triangular mesh and extract those vertices whose elevation values are less than or equal to the current processing threshold to form a temporary vertex subset. Then, based on the inherent edge connection relationship of the triangular mesh, the temporary vertex subset is analyzed for connectivity. Specifically, the system uses a breadth-first search algorithm to start from any vertex in the subset that has not been visited before, visit all adjacent vertices directly connected to it and also in the subset, and mark them as belonging to the same connected component. This process is repeated until all vertices in the subset are assigned to the corresponding connected component. During the scanning process, the system assigns a unique identifier to each newly discovered connected component and records the current processing threshold corresponding to the first appearance of the branch. This threshold is registered as the appearance threshold of the connected component. At the same time, the system continuously monitors the adjacency relationship between different connected components. When it is found that two previously independent connected components share at least one triangular mesh edge at the boundary due to the addition of new vertices or edges caused by the increase of the threshold, the system immediately records a merging event and registers the current processing threshold at this moment as the merging threshold of the two branches. In addition, during the entire scanning process from the lowest to the highest threshold, the system calculates and accumulates the total horizontal projection area of the triangular patches covered by each independent connected component at each threshold. These area data are associated with the event record of each connected component in sequence to form a connected component event record containing the unique identifier, appearance threshold, merging threshold, and area sequence of each connected component.

[0056] formula: The advantage of the formula is that the first term The topological persistence of a feature is measured by the ratio of its "surviving" elevation interval (from emergence to merging) to the total elevation range. This method effectively identifies deep valleys or peaks that maintain their independent form over a large elevation range. (Second term) The geometric significance of a feature is measured by its relative volume. The introduction of a logarithmic function can smooth out extreme differences in volume size, so that the score is not overly sensitive to changes in volume. At the same time, it can distinguish between large-scale and small landform units. The multiplication of the two results in only those landform features that are both persistent and have considerable volume being able to obtain high scores, thereby identifying key landforms that have an important impact on the overall structure of the terrain, such as major valleys and basins, rather than small depressions that appear instantaneously.

[0057] The steps to obtain 'a' (the occurrence threshold in the connected component occurrence event record) are as follows: This parameter comes directly from the connected component event record generated in the previous step. It marks the lowest elevation point where an independent terrain feature (such as a depression) begins to form. In topography, this corresponds to a local minimum point. When the elevation scan threshold reaches this value, a new connected component is created. For example, in a monitoring area, through elevation scanning, it is found that when the scan threshold reaches 85.2 meters, a new connected component numbered C-101 appears for the first time. Then, the occurrence threshold 'a' of the connected component C-101 is 85.2 meters.

[0058] The steps for obtaining b (the merging threshold in the connected component event record) are as follows: This parameter also comes from the connected component event record. It marks the critical elevation at which an independent terrain feature is connected to another larger or more important terrain feature (usually its "parent" feature) due to the continued "rise" of the water level. This elevation point corresponds to a saddle point on the terrain, which is the lowest watershed between two depressions. When the scanning threshold reaches this value, the original connected component is merged into another connected component, and its independent life cycle ends. For example, for the connected component C-101 that appeared at 85.2 meters, as the scanning threshold continues to rise, when it reaches 102.5 meters, it is connected to another existing connected component C-58 at the boundary. The system records the merging event, so the merging threshold b of the connected component C-101 is 102.5 meters.

[0059] Φ maxThe acquisition step of the maximum elevation value of the single period elevation field is that the maximum value of the Z coordinate value of all the triangular mesh vertices in the single period three-dimensional terrain data used in this analysis is found through a global traversal and comparison, which represents the highest point of the terrain in the entire monitoring area. For example, after scanning all 5 million data points in a certain mining area, the system determines that the highest point is 547.3 meters, and then Φ max is 547.3 meters.

[0060] Φ min The acquisition step of the minimum elevation value of the single period elevation field is that the minimum value of the Z coordinate value of all the triangular mesh vertices in the single period three-dimensional terrain data used in this analysis is found through a global traversal and comparison, which represents the lowest point of the terrain in the entire monitoring area. For example, after scanning all the data points in the same mining area, the system determines that the lowest point is 65.1 meters, and then Φ min is 65.1 meters.

[0061] V c The acquisition step of the volume of the connected branch in the range of threshold values from a to b is that the volume is approximately calculated by integrating the horizontal projection area of the connected branch at each discrete elevation threshold value. First, the area sequence associated with the branch is extracted from the connected branch event record, which records the area of the branch at each elevation scanning step from the appearance threshold value a to the merging threshold value b. The volume is calculated using the numerical integration method, and the formula is: where h i is the discrete elevation threshold sequence from a to b, A(h i ) is the area of the connected branch at the elevation h i , and N is the number of threshold values in this range. For example, for the connected branch C-101, the threshold range is 85.2 meters to 102.5 meters, and during this period, it experiences 30 elevation steps and records 31 area values. Through the above trapezoidal rule integration, the calculated cumulative volume value is 125,000 cubic meters, and then V c is 125,000.

[0062] A total The acquisition step of the total horizontal projection area of the monitoring area is that the parameter is calculated according to the monitoring area boundary polygon initially set by the project. By applying the Gauss area formula or other polygon area calculation methods to the polygon, the projection area in the two-dimensional plane can be obtained. This value is fixed in a monitoring task, for example, the monitoring range of a mining area is defined as an irregular polygon, and the area calculated by the geographic information system software is 5 square kilometers, i.e. 5,000,000 square meters, and then A total is 5,000,000.

[0063] H unit The acquisition step of (standard height unit) is that this parameter is a constant for dimensionless normalization, and its value is set to 1, which is consistent with the length unit of height and area. In this example, the height unit is meters, and the area unit is square meters. Therefore, H unit is set to 1 meter, which makes the denominator A total ·H unit The dimension of the denominator becomes cubic meters, which matches the dimension of the numerator V c , so that the ratio becomes a dimensionless pure number, which is convenient for logarithmic operation and cross-regional comparison.

[0064] Calculation process:

[0065] According to the above parameter acquisition step, the structural significance evaluation value P sig of the connected branch C-101 is calculated,

[0066] The known parameters are as follows:

[0067] a = 85.2 meters;

[0068] b = 102.5 meters;

[0069] Φ max = 547.3 meters;

[0070] Φ min = 65.1 meters;

[0071] V c = 125000 cubic meters;

[0072] A total = 5000000 square meters;

[0073] H unit = 1 meter;

[0074] The calculation process is as follows:

[0075]

[0076] P sig = 0.035877·ln(1.025);

[0077] P sig = 0.035877·0.024692;

[0078] P sig ≈0.000886;

[0079] The result shows that the structural significance value of the connected branch C-101 is about 0.000886, which is a relative measure, and its absolute size needs to be compared with the evaluation values of all features in the region. Generally, the evaluation values are distributed between 0 and 1, and a value close to 0, such as 0.000886 in this example, indicates that the topographic feature (depression) has very low significance in terms of both elevation depth (persistence) and spatial volume relative to the entire monitoring area, and may only represent a small, local surface depression. In subsequent grid simplification, the vertices and edges associated with this feature will be assigned a lower retention priority. Conversely, if the calculation result is close to 1, it means that the feature is an extremely important geomorphic unit in the region, such as a deep main valley.

[0080] Based on the structural significance evaluation value, the system will backtrack the occurrence event of each connected branch, and locate the unique TIN vertex triggering the occurrence event when each connected branch first occurs. This vertex is the local minimum point of the topographic feature (usually a depression or basin) represented by the connected branch, and its elevation value is equal to the occurrence threshold of the connected branch. The locating process is completed by querying the occurrence threshold and initial vertex identifier in the connected branch event record. Once the local minimum point is located, the system writes the complete structural significance evaluation value previously calculated for the connected branch into the record of the vertex as an attribute value. However, in complex mountainous or hilly terrain, due to the nested structure of the topography, a vertex may be the local minimum of multiple depressions of different scales at the same time. For example, a small pit is located at the bottom of a large pit. As the lowest point of the small pit, the vertex is associated with a structural significance evaluation value of the small pit. When the water level (scanning threshold) submerges the small pit and continues to rise, the vertex is still part of the large pit and may be considered as a secondary center of the large pit under certain topological analysis, and thus may be assigned another structural significance evaluation value calculated based on the large pit. In order to handle the case where one vertex corresponds to multiple structural significance evaluation value records, the system sets a clear rule: compare all candidate evaluation values for the vertex, and only keep the maximum value as the final and unique structural significance score. For example, if vertex V-789 is assigned an evaluation value of 0.15 because it is the lowest point of small depression C-200, and an evaluation value of 0.62 because it belongs to a larger basin C-80, the system will compare 0.15 and 0.62, and finally assign 0.62 as the evaluation value of vertex V-789. After traversing all connected branches and completing the above assignment and selection operations, the system finally assigns a determined evaluation value to each vertex that is a local minimum, representing the importance of the most significant topographic structure it is in. The structural significance score of other vertices that are not local minima is set to 0 by default, and a set of vertices and their corresponding highest structural significance scores is finally formed, which is the vertex structural significance score set.

[0081] The comprehensive simplification generation value acquisition step is as follows:

[0082] Based on the vertex time change rate set and the vertex structural significance score set, each contractible edge of the basic TIN is traversed, and the time change rate and structural significance score of the vertices at both ends of the contractible edge are extracted in turn. The mean and standard deviation of the time change rates of all vertices in the monitoring area and the mean and standard deviation of the structural significance scores of all vertices are calculated. The contractible edge and the attribute values of the vertices at both ends are associated and stored, forming a contractible edge attribute statistical list.

[0083] According to the contractible edge attribute statistical list, the comprehensive simplification generation value is calculated, and the calculation formula is:

[0084]

[0085] wherein C e is the simplified value of the contractible edge e i,j is the normalized attribute difference vector, is the normalized time variation rate of vertex i, is the normalized time variation rate of vertex j, is the normalized structural saliency score of vertex i, is the normalized structural saliency score of vertex j, μ r is the mean of all vertex time variation rates, σ r is the standard deviation of all vertex time variation rates, μ P is the mean of all vertex structural saliency scores, σ P is the standard deviation of all vertex structural saliency scores, Σ' is the two-dimensional correlation coefficient matrix calculated from all vertex normalized attribute pairs (r' k , P' sig,k ), and k is the vertex index in the monitoring area.

[0086] Specifically, based on the vertex time variation rate set and the vertex structural saliency score set, the system first traverses each edge in the basic triangular mesh and applies a contractibility judgment criterion, which is based on the topology of the two end points of the edge and their neighborhoods. For example, an edge is contractible if and only if the contraction operation does not change the mesh topological manifold, i.e., it does not produce non-manifold edges or non-manifold vertices after contraction, nor does it cause the mesh to flip. For each edge that passes the contractibility judgment, the system accurately extracts the corresponding attribute values of its two end vertices, i.e., the time variation rate and the structural saliency score, from the vertex time variation rate set and the vertex structural saliency score set through vertex number indexing. Next, the system performs global statistical analysis on all vertices in the monitoring area. It reads the time variation rate data of all vertices from the vertex time variation rate set and calculates the mean and standard deviation thereof. Similarly, it reads the structural saliency scores of all vertices from the vertex structural saliency score set and calculates the mean and standard deviation thereof. These global statistics, including the mean and standard deviation of the time variation rate, and the mean and standard deviation of the structural saliency score, will serve as the basis for subsequent normalization. After completing the statistics, the system constructs a detailed record list that associates and stores the unique identifier of each contractible edge with its two end vertices' original time variation rates, original structural saliency scores, and the four global statistics (two mean values and two standard deviation values) just calculated, forming a contractible edge attribute statistics list.

[0087] Formula:​ The advantage of this formula lies in its use of Mahalanobis distance to calculate the comprehensive simplification cost, which offers a significant advantage over the traditional Euclidean distance. Mahalanobis distance considers the correlation between the time rate of change and the structural significance score, and calculates this through the inverse Σ′ of the covariance matrix. ；1 To eliminate the influence of this correlation, it also utilizes the variance of each variable (reflected in the standardization process) for scale normalization. This makes the cost calculation unaffected by the original dimensions of the variables, enabling a more objective and fair measurement of the comprehensive differences in spatiotemporal characteristics between the two vertices, ultimately yielding a simplified cost C. e It is a dimensionless "distance" that accurately reflects the amount of information loss caused by shrinking an edge: a small cost means that the two vertices are very similar in terms of dynamic changes and static structure, and shrinking them has little impact on the overall expression of the model; conversely, a large cost means that the two vertices have significantly different features, and shrinking them will cause a large loss of information, so they need to be retained first. This method can ensure that during the mesh simplification process, the vertex pairs with the most similar spatiotemporal features are merged first, thereby preserving the key dynamic change areas and important static structural features of the terrain to the greatest extent.

[0088] The steps to obtain Δ′ (the standardized attribute difference vector) are as follows: this vector is obtained by analyzing the contractible edge e. i,j The result is obtained by subtracting the standardized attribute values ​​of the two endpoints i and j. Specifically, it comprises two components: the standardized difference of the vertex time change rate and the standardized difference of the vertex structure saliency score. Its calculation relies on the standardization process of the original attribute values, and the formula is as follows: Where r′ i and r′ j P′ is the standardized value of the time rate of change of vertices i and j. sig,i and P′ sig,j These are the standardized structural saliency scores for vertices i and j. These standardized scores are obtained by subtracting the global mean from the original values ​​and then dividing by the global standard deviation. For example, for an edge connecting vertices 201 and 205, if r′ is calculated... 201 =0.5, r′ 205 =0.6, P′ sig,201 =-1.2, P′ sig,205 = -1.0, then the attribute difference vector of this edge is

[0089] μ r The steps to obtain the mean of the time change rate of all vertices are as follows: This parameter is obtained by reading the list of shrinkable edge attribute statistics generated in the previous step and extracting the time change rate r of all N vertices in the monitoring area. k And calculate their average value, the calculation formula is as follows: For example, in a monitoring area containing 1 million vertices, the system collects the time change rates of all vertices from the time change rate set of vertices, calculates the mean of the time change rates of all vertices by accumulating summation and dividing by the total number of vertices, and obtains the mean μ r = 0.0015.

[0090] σ r (the standard deviation of the time change rates of all vertices) is obtained by reading the collapsible edge attribute statistics list, after the mean μ r is calculated, the system continues to calculate the square sum of the difference between each vertex time change rate and the mean, then divides by the total number of vertices N, and finally takes the square root, and the calculation formula is For example, based on the data of the aforementioned 1 million vertices and the calculated mean 0.0015 m / day, the system calculates the standard deviation as 0.005 m / day, and σ r = 0.005.

[0091] μ P (the mean of the structural significance scores of all vertices) is obtained by reading the collapsible edge attribute statistics list, extracting the structural significance scores P sig,k of all N vertices in the monitoring area recorded therein, and calculating the average value, and the calculation formula is For example, in the same monitoring area, the system collects the scores of all vertices from the vertex structural significance score set (most of the scores of non-local minimum points are 0), calculates the mean as 0.0002, and μ P = 0.0002.

[0092] σ P (the standard deviation of the structural significance scores of all vertices) is obtained after the mean μ P is calculated, the system continues to calculate the square sum of the difference between each vertex structural significance score and the mean, then divides by the total number of vertices N, and finally takes the square root, and the calculation formula is For example, based on the aforementioned data and the calculated mean 0.0002, the system calculates the standard deviation as 0.001, and σ P = 0.001.

[0093] Σ'(the two-dimensional correlation coefficient matrix calculated by the normalized attribute pairs (r' k , P' sig,k ) of all vertices) is obtained as follows: first, the system needs to normalize the two attributes of all N vertices to obtain (r' k , P' sig,k) sequences, then the Pearson correlation coefficient p between these two normalized sequences is calculated, which is Since the data has been normalized, its mean is 0 and its standard deviation is 1, so the formula can be simplified as The correlation coefficient matrix is a 2x2 symmetric matrix, whose diagonal elements are 1 (the correlation of a variable with itself), and the non-diagonal elements are p, i.e. For example, through calculation, it is found that there is a slight negative correlation between the time change rate and the structural significance score, p = -0.2, so the correlation coefficient matrix is

[0094] The calculation process is as follows:

[0095] Now, for the edge e 201,205 connecting vertex 201 and vertex 205, its integrated simplification cost C e is calculated as

[0096] The known parameters are as follows:

[0097]

[0098] First, the inverse matrix of Σ' is calculated: ；1

[0099]

[0100] Next, the square of C e , i.e. the square of Mahalanobis distance, is calculated:

[0101]

[0102] Finally, the square root of C e is calculated:

[0103]

[0104] This result shows that the integrated simplification cost value of the contractible edge e 201,205 is about 0.2458, which is a comprehensive measure after correction of multi-dimensional features (time change rate and structural significance) and their internal correlation, representing the amount of spatio-temporal information loss caused by merging vertex 201 and 205 into one point. In the entire simplification process, such a cost value will be calculated for all contractible edges, which will be used to sort all simplification operations. The smaller the cost value, the smaller the error introduced by the contraction operation, and the higher the priority of its execution. For example, if the cost value of another edge is 0.1, it will be contracted before this edge.

[0105] The steps for obtaining the ordered simplification operation sequence are as follows: ​

[0106] Based on the comprehensive simplification generation value, the corresponding comprehensive simplification generation value of each contractible edge is recorded and arranged in ascending order of the comprehensive simplification generation value, forming an ordered simplification operation sequence.

[0107] Specifically, based on the comprehensive simplification generation value, the system first records the corresponding comprehensive simplification generation value calculated in the previous step for each contractible edge recorded in the contractible edge attribute statistics list, that is, this value is attached to the record of each edge as a new attribute field. After completing the generation value recording of all contractible edges, the system collects all edge entries with generation values and applies an efficient sorting algorithm, such as quicksort or heapsort, to perform global sorting. The sorting basis is the comprehensive simplification generation value, and the sorting is arranged in ascending order from small to large. The result of the sorting is a list or queue, in which the edge with the lowest comprehensive simplification generation value is located at the front of the list, and the edge with the highest generation value is located at the end. This sorted list is a sequence of vertex pair contraction operations that are scheduled to be executed, which precisely defines the order of mesh simplification and ensures that each simplification operation is the choice with the smallest information loss among all possible operations at the moment, forming an ordered simplification operation sequence.

[0108] The acquisition step of the multi-level simplification record set is:

[0109] Based on the ordered simplification operation sequence, vertex pair contraction is performed in order from the contractible edge with the smallest comprehensive simplification generation value, and the removed vertex index, removed edge identifier, and removed triangle patch identifier are recorded. The three-dimensional coordinates of the newly generated vertex and the replacement relationship are recorded and the adjacent connection and normal are updated synchronously, forming a multi-level simplification record set.

[0110] Specifically, based on the ordered simplification operation sequence, the system starts an iterative simplification process from the starting position of the ordered simplification operation sequence, that is, the contractible edge with the smallest comprehensive simplification generation value, and performs vertex pair contraction operations in order. In each iteration, the system takes out the edge with the lowest cost in the current sequence, for example, edge e i,j , connecting vertices v i and v j , and performs contraction. According to the preset rules, the system removes one of the vertices, for example, the vertex with higher structural saliency score or lower time change rate is retained as the target vertex, and v j is retained and v i is removed. Before performing the removal, the system creates a detailed record entry for this operation, which first records the unique index of the removed vertex (i.e., i), the unique identifier of the removed edge (i.e., the ID of e i,j ), and the identifier list of all triangles that are invalidated due to this contraction, which includes all triangles that share edge e i,jDirectly adjacent to two triangular facets, then the record entry stores the information of the newly generated vertex, which is the preserved vertex v j here, and explicitly records the replacement relationship of vertex v i to v j , which is a key information for subsequent construction of parent-child links between levels. After the record is completed, the system updates the topology of the mesh, reconnects all edges originally connected to the removed vertex v i to the target vertex v j , and deletes the redundant edges and degenerate facets (i.e. facets with zero or near-zero area) thus generated. At the same time, since the set of triangular facets around the target vertex v j has changed, the system recalculates the surface normal of v j and all vertices in its one-ring neighborhood, usually by weighted average of all adjacent facet normals, where the weights can be the areas of the facets. After one contraction and record, the system removes the processed edge from the ordered sequence of simplification operations, and continues to process the next edge with the lowest cost. Repeat this process until the sequence is empty or the preset simplification target is reached, forming a multi-level simplification record set.

[0111] The steps for obtaining the adaptive spatio-temporal hierarchical detail model are as follows:

[0112] Based on the multi-level simplification record set, extract the mesh snapshot after each contraction and the corresponding comprehensive simplification cost value. According to the preset comprehensive simplification cost threshold range set, archive the mesh snapshot, summarize the vertex set, edge set and triangular facet set in the snapshot for each threshold range, and establish the parent-child reference between layers to generate the hierarchical data structure.

[0113] Based on the hierarchical data structure, read the corresponding comprehensive simplification cost threshold range and mesh snapshot of each layer, establish the parent-child reference and rollback pointer between layers, and solidify the contraction sequence position to layer mapping and time index. Integrate the metadata and positioning index of each layer to form the adaptive spatio-temporal hierarchical detail model.

[0114] Specifically, based on the multi-level simplification record set, the system first extracts a series of mesh snapshots representing different levels of simplification from the record set, each time the vertex pair contraction operation corresponds to a unique comprehensive simplification generation value, which can be regarded as the "cost" of this simplification operation, the system associates these cost values with the corresponding mesh state, then the system needs to discretize and stratify these continuous mesh snapshots according to a set of pre-set comprehensive simplification cost threshold ranges, the setting of this threshold range set is based on the statistical distribution of the comprehensive simplification generation values of all contraction operations, for example, the system can calculate the deciles of all generation values, that is, find the values at the 10%, 20%,..., 90% positions after arranging all generation values from small to large, and use these percentile values as the boundary of hierarchical division, thereby generating, for example, 10 cost range intervals such as [0, 0.05], (0.05, 0.12], (0.12, 0.3] and the like, which can ensure that each level contains approximately equivalent changes in simplification details, after determining the threshold range, the system archives all mesh snapshots in buckets, for each threshold range, for example, (0.12, 0.3], the system selects the mesh state after the last contraction operation in this range as the representative snapshot of this level (or called Level of Detail), then the system summarizes all geometric information contained in the representative snapshot of each level defined by the threshold range, that is, the current vertex set, edge set and triangular patch set, and uses the replacement relationship in the multi-level simplification record set (for example, records that vertex v i is contracted to vertex v j ), to establish an explicit parent-child reference relationship between adjacent two levels, specifically, the vertex v j existing in the coarser level is the "parent" node of the predecessor vertices v i and v j in the next finer level, and a hierarchical data structure is generated.

[0115] Based on the hierarchical data structure, the system performs the final data integration and structure solidification. First, the system traverses each layer in the generated hierarchical data structure, reads the corresponding comprehensive simplification cost threshold range and the representative mesh snapshot. During the traversal process, the system further strengthens the connection relationship between layers. In addition to the established parent-child reference, a rollback pointer is established for each vertex in the coarse level, which points to a specific entry in the multi-level simplification record set. This entry describes in detail how the vertex is generated from two vertices in the next fine level through the contraction operation, including all connection information of the removed vertex, thereby providing a direct basis for "decompression" or detail recovery from coarse to mesh. Then, the system solidifies a direct mapping relationship from the sequence number of the contraction operation to the layer number it belongs to, creating a lookup table that allows the detail level of the mesh after a specific simplification step (e.g., the 5000th contraction) to be immediately queried based on the sequence number of the step. At the same time, the system indexes the key time information related to model construction, including the specific acquisition timestamp sequence of the three-dimensional terrain data used to calculate the vertex time change rate, and associates it with the entire model as part of the metadata. Finally, the system integrates all metadata for each layer, including the total number of vertices and faces, coordinate reference system definition, data precision, and constructs an R-tree or quadtree spatial index structure for each layer's vertex dataset. This index structure can greatly speed up subsequent spatial location-based queries and data retrieval speeds, forming an adaptive spatiotemporal hierarchical detail model.

[0116] The acquisition steps of the dynamically updated terrain hierarchical structure are as follows:

[0117] Based on the new terrain data and the adaptive spatiotemporal hierarchical detail model, the elevation values at the same location in the new terrain data and the model are compared one by one for each division unit in the monitoring area. The units with an elevation difference exceeding the preset change threshold are marked as surface change units. The spatial range of the surface change units is extracted and adjacent units are merged to form a monitoring area, generating a monitoring area boundary set.

[0118] Based on the monitoring area boundary set, all vertices within the monitoring area are located and the historical elevation sequence is re-read. Missing values and duplicate timestamps are removed, and the absolute value mean of the time-adjacent elevation difference is calculated as the time change rate. The comprehensive simplification cost value is recalculated in combination with the existing structure significance score, and compared with the original comprehensive simplification cost value one by one to mark the value changes, generating an affected edge list.

[0119] Based on the affected edge list, the arrangement position of the affected edge in the ordered simplification operation sequence is updated, the level where the affected edge is located in the adaptive spatio-temporal hierarchical detail model is retrieved, the associated local grid range is located, the invalid geometry elements are removed, and the vertices, edges and triangular facets in the corresponding positions are regenerated according to the new integrated simplification generation value, so as to form a dynamically updated terrain level structure.

[0120] Specifically, based on the new period of terrain data and the adaptive spatio-temporal hierarchical detail model, the system first divides the entire monitoring area on the horizontal projection plane into a regular grid cell system, for example, for a 5 square kilometer monitoring area, a 10 meter by 10 meter cell size can be set, thereby generating 50,000 independent monitoring cells, then the system iterates through these cells one by one, for each cell, all points falling within the cell are extracted from the new period of terrain data, and the average value of the elevation values is calculated, at the same time, the spatial index of the highest resolution level (i.e. the basic triangular grid) in the adaptive spatio-temporal hierarchical detail model is used to quickly retrieve all vertices whose centers are located within the cell, and the average value of the elevations of these vertices is calculated, then the absolute difference between the two average elevation values is calculated, and the difference is compared with a preset change threshold, which is determined according to historical data and measurement device accuracy, the setting method is to calculate the average value and standard deviation of the measurement error of all past period terrain data relative to the true ground surface, for example, if the historical data shows that the average elevation error of unmanned aerial vehicle oblique photography measurement technology is 0.1 meters, and the standard deviation is 0.03 meters, then the change threshold can be set to the average error plus three times the standard deviation, i.e. 0.1+3*0.03=0.19 meters, any cell with an elevation difference greater than 0.19 meters is marked as a land surface change cell, after all cells are checked and marked, the system uses a connected component analysis algorithm to merge all spatially adjacent (i.e. sharing edges or corners) land surface change cells into one or more continuous regions, and calculates the outer boundary contour of each continuous region to generate a set of monitoring area boundaries.

[0121] Based on the set of monitoring region boundary polygons, the system performs a spatial query on the base triangular mesh of the adaptive spatio-temporal hierarchical detail model, pinpointing and extracting all the triangular mesh vertices that are completely or partially inside the boundaries. For each of the pinpointed vertices, the system accesses the back-end database, re-reads its complete historical elevation sequence, and appends the elevation value at the corresponding spatial location in the new period of terrain data to the end of the sequence. Then, the system performs a data cleaning on the updated elevation sequence, removing all the missing records marked with special values such as -9999, and checking for duplicate time-stamped entries due to data integration errors, merging them if any. After obtaining a clean and complete historical elevation sequence, the system re-computes the average of the absolute values of the time-adjacent elevation differences of the vertex, obtaining a new time change rate. This process does not re-compute the structural significance score of the vertex, because the score reflects the static structure of the terrain, which usually remains unchanged before a major terrain reconstruction. Next, the system iterates through all the collapsible edges with the updated vertex as one of the endpoints, combines the new time change rate of the vertex with the unchanged time change rate of the other endpoint, as well as the original structural significance scores of both endpoints, and re-computes the integrated simplification cost values of the edges using the global statistics computed during the model construction (the mean and standard deviation of the time change rates of all vertices, the mean and standard deviation of the structural significance scores of all vertices, and the two-dimensional correlation coefficient matrix). Finally, the system compares the newly computed integrated simplification cost values with the original integrated simplification cost values stored in the model, and records the unique identifiers of any edges whose values have changed (exceeding a small floating-point comparison tolerance, such as le-9), generating a list of affected edges.

[0122] Based on the affected edge list, the system first performs a local update on the original global ordered simplification operation sequence. For each edge in the affected edge list, the system removes its old entry in the ordered simplification operation sequence, and then re-inserts the edge and its newly calculated integrated simplification cost into the correct position in the sequence, maintaining the ascending order of the sequence. Subsequently, the system queries the original simplification order of each affected edge in the adaptive spatio-temporal hierarchical detail model to retrieve the levels at which the edge was contracted, and marks these levels as to-be-updated levels. For each to-be-updated level, the system locates an associated local grid range by expanding two-ring neighborhoods outward from the vertices involved in the edges whose simplification cost changed after the update. This range defines the area that needs to be reconstructed. Then, the system performs reverse simplification operations, i.e., decompression, on all geometric elements in the local grid range using the rollback pointers stored in the model, until the highest resolution state of the base triangular mesh is restored. Next, the system uses the restored local high-resolution mesh as a basis, and executes vertex pair contraction operations in the new order according to the part of the updated ordered simplification operation sequence that only involves edges within the local range, to regenerate the simplified representation of the area at each affected level. Finally, the newly generated local grid blocks are seamlessly embedded back into the corresponding levels in the adaptive spatio-temporal hierarchical detail model, replacing the original invalid geometric elements, to form a dynamically updated terrain level structure.

[0123] The above merely describes the preferred embodiments of the present application, but the present application is not limited to the above, and any modification or change made by those skilled in the art based on the above disclosure without departing from the technical solution of the present application shall fall within the protection scope of the present application.

Claims

1. A method of ground change monitoring and rapid update of topographic data, characterized by, The method comprises the following steps: Based on multi-period three-dimensional terrain data, a triangular mesh covering the monitoring area is constructed, the elevation values of multi-period terrain data are compared by vertex by vertex, a vertex time change rate set is generated, a single-period elevation field is taken as a scalar field, the generation and merging events of connected branches are tracked by scanning the scalar field from low to high, and a vertex structure significance score set is obtained; Based on the vertex time change rate set and the vertex structure significance score set, each contractible edge of the basic triangular mesh is traversed, and a comprehensive simplification generation value is calculated, the comprehensive simplification generation values of each edge are arranged according to the comprehensive simplification generation values, and an ordered simplification operation sequence is established; Based on the ordered simplification operation sequence, vertex pair contraction is sequentially executed from the edge with the lowest comprehensive simplification generation value, a multi-level simplification record set is formed, different grid levels of different simplification degrees are organized into a hierarchical data structure according to the information in the multi-level simplification record set, and an adaptive space-time hierarchical detail model is established; Based on new-period terrain data and the adaptive space-time hierarchical detail model, a monitoring area where surface changes occur is identified, the time change rate of the vertex in the monitoring area is recalculated and the comprehensive simplification cost is updated, the original simplification cost is compared, and the affected levels in the adaptive space-time hierarchical detail model are locally reconstructed to generate a dynamically updated terrain level structure.

2. The method of ground change monitoring and rapid terrain data updating according to claim 1, characterized in that, The vertex time change rate set is obtained by: Based on multi-period three-dimensional terrain data, the coordinate fields X and Y and the elevation field Z are analyzed, the coordinate reference and the elevation zero base are unified, the effective points are clipped according to the monitoring area boundary, the adjacent points are connected into non-intersecting triangular patches according to the spatial proximity relationship, and the hanging points are removed to generate a triangular mesh; Based on the triangular mesh, the elevation value sequence of the same spatial position in the multi-period three-dimensional terrain data is retrieved according to the vertex number, the missing marks and repeated time stamps are removed, the absolute value of the elevation value difference between adjacent time points is calculated, and the average is taken for the whole sequence to obtain the average elevation variation amplitude of each vertex; Based on the average elevation variation amplitude of each vertex, the adjacent time point record interval is read and unified to a fixed time unit, the average elevation variation amplitude is converted into a rate according to the time unit and is indexed with the vertex number, and the identified item set is summarized in order of the vertex number to generate a vertex time change rate set.

3. The method of ground change monitoring and rapid terrain data updating according to claim 1, characterized in that, The vertex structure significance score set is obtained by: Based on the single-period elevation field, a threshold list arranged in ascending order of elevation value is generated, the threshold values are set in sequence, and the triangular mesh vertices with an elevation value not greater than the threshold value are extracted, the adjacent vertices are aggregated into connected branches according to the connection relationship of the triangular mesh edges, the first appearing connected branch is recorded as an appearance event and the appearance threshold value is registered, and when two connected branches share at least one edge at the boundary, a merging event is recorded and the merging threshold value is registered, and the triangular patch area covered by each connected branch at each threshold value is accumulated during the whole scanning process to form a connected branch event record containing the appearance threshold value, the merging threshold value and the area sequence; According to the connected branch event record, the structure significance evaluation value is calculated. Based on the structural saliency evaluation value, a corresponding local minimum vertex is located when each connected branch appears, the structural saliency evaluation value of the connected branch is written into the record of the local minimum vertex, and in the case that there are multiple structural saliency evaluation value records for the same local minimum vertex, only one entry is reserved by taking the maximum value, and a vertex structural saliency score set is generated.

4. The method of ground change monitoring and rapid terrain data updating according to claim 1, characterized in that, The comprehensive simplification generation value obtaining step is: Based on the vertex time variation rate set and the vertex structural saliency score set, each contractible edge of the basic triangular mesh is traversed, the time variation rate and the structural saliency score of the vertices at both ends of the contractible edge are sequentially extracted, the mean value and the standard deviation of the time variation rates of all the vertices in the monitoring area and the mean value and the standard deviation of the structural saliency scores of all the vertices are counted and monitored, the contractible edge and the attribute values of the vertices at both ends and the overall statistics are associated and stored, and a contractible edge attribute statistical list is formed. The comprehensive simplification generation value is calculated according to the contractible edge attribute statistical list.

5. The method of ground change monitoring and rapid terrain data updating according to claim 1, characterized in that, The ordered simplification operation sequence obtaining step is: Based on the comprehensive simplification generation value, the corresponding comprehensive simplification generation value is registered for each contractible edge, and the comprehensive simplification generation values are arranged in ascending order to form an ordered simplification operation sequence.

6. The method of ground change monitoring and rapid terrain data updating according to claim 1, characterized in that, The multi-level simplification record set obtaining step is: Based on the ordered simplification operation sequence, vertex pair contraction is sequentially performed from the contractible edge with the smallest comprehensive simplification generation value, and the removed vertex index, the removed edge identifier and the removed triangular patch identifier are recorded in sequence, the three-dimensional coordinates of the newly generated vertex and the replacement relationship are recorded, and the adjacent connection and the normal are synchronously updated to form a multi-level simplification record set.

7. The method of ground change monitoring and rapid terrain data updating according to claim 1, characterized in that, The adaptive spatio-temporal hierarchical detail model obtaining step is: Based on the multi-level simplification record set, the mesh snapshot after each contraction and the corresponding comprehensive simplification generation value are extracted, the mesh snapshot is binned and archived according to a preset comprehensive simplification cost threshold range set, the vertex set, the edge set and the triangular patch set in the snapshot are summarized for each threshold range, and the inter-layer parent-child reference is established to generate a hierarchical data structure; Based on the hierarchical data structure, the corresponding comprehensive simplification cost threshold range and the mesh snapshot of each layer are read, the inter-layer parent-child reference and the rollback pointer are established, the contraction sequence bit is fixed to the layer mapping and the time index, the metadata and the positioning index of each layer are integrated, and an adaptive spatio-temporal hierarchical detail model is formed.

8. The method of ground change monitoring and rapid terrain data updating according to claim 1, characterized in that, The dynamic update terrain hierarchical structure obtaining step is: Based on the new period terrain data and the adaptive spatio-temporal hierarchical detail model, the elevation values at the same position in the new period terrain data and the model are compared one by one according to the monitoring area division unit, the unit with an elevation difference exceeding a preset change threshold is marked as a ground surface change unit, the spatial range of the ground surface change unit is extracted and adjacent units are merged to form a monitoring area, and a monitoring area boundary set is generated; Based on the monitoring area boundary set, all the vertices in the monitoring area are located and the historical elevation sequence is re-read, the missing values and the repeated time stamps are removed, the absolute value mean of the time-adjacent elevation difference is calculated as the time variation rate, the comprehensive simplification generation value is recalculated in combination with the existing structural saliency score, the original comprehensive simplification generation value is compared one by one, and the edges with value changes are marked to generate an affected edge list; Based on the affected edge list, the arrangement position of the affected edge in the ordered simplification operation sequence is updated, the level where the affected edge is located is searched in the adaptive space-time hierarchical detail model, the associated local grid range is located, the invalid geometry elements are removed, and the vertices, edges and triangular facets in the corresponding positions are regenerated according to the new integrated simplification generation value, so as to form a dynamically updated terrain level structure.

Images