A land and resource surveying and mapping data processing method based on space calculation

Abstract

The application discloses a kind of land and resources surveying and mapping data processing methods based on space calculation, it is related to geographic information technology field, including: obtaining land and resources surveying and mapping data, matching current surveying and mapping point with historical surveying and mapping point;Boundary chain is constructed and boundary continuous sampling point set is obtained by sampling;Write into RandLA-Net structure to generate surveying and mapping neighborhood coding feature;Calling local feature aggregation structure generates graph spot boundary enhancement feature;Difference decomposition is formed surveying and mapping difference residual;According to residual type, return to generation change identification result by branch;Update surveying and mapping data and output processing result.The application can improve boundary continuous keeping ability in land and resources surveying and mapping data processing, change identification accuracy and surveying and mapping data update integrity.

Description

Technical Field

[0001] This invention relates to the field of geographic information technology, and in particular to a method for processing land and resources surveying and mapping data based on spatial computing. Background Technology

[0002] In the field of geographic information technology, existing technologies mostly employ coordinate transformation, distance matching, spatial overlay analysis, manual rule verification, and point cloud feature extraction to update surveying data. For areas with a large number of surveying points, existing methods reduce the amount of data through sampling and then improve processing efficiency by utilizing neighborhood feature extraction, so as to complete change identification over a larger area.

[0003] Existing processing methods typically focus more on the locational differences of individual mapping points, neglecting the continuous connectivity relationships within patch boundaries. During sampling, the start and end points of boundaries, the locations of boundary direction changes, and the intersections of patch boundaries are easily weakened by ordinary sampling rules, resulting in subsequent neighborhood aggregation failing to accurately reflect the true boundary trends. Existing neighborhood feature extraction also largely relies on spatial distance, making it difficult to simultaneously reflect the correlation between boundary connectivity, patch affiliation, and historical point location changes.

[0004] Current change identification methods often rely on local differences as the basis for judgment, lacking a process of verification along boundary chains, map patch relationships, and spatial proximity relationships. This makes it easy to mistake surveying noise for real changes and to break down continuous boundary changes into multiple scattered change results, thus affecting the accuracy and completeness of land and resources surveying and mapping data updates.

[0005] Therefore, how to provide a spatial computing-based method for processing land and resources surveying and mapping data is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0006] One objective of this invention is to propose a spatial computing-based method for processing land and resources surveying and mapping data. This invention constructs a boundary chain and combines it with the RandLA-Net structure to perform continuous boundary sampling and mapping neighborhood coding on land and resources surveying and mapping data. It utilizes a residual split-path backhaul mechanism to achieve difference decomposition and change identification, which has the advantages of good boundary preservation, accurate change identification, and strong continuity of surveying and mapping data updates.

[0007] A land and resources mapping data processing method based on spatial computing according to an embodiment of the present invention includes the following steps: Acquire land and resources surveying and mapping data, perform spatial benchmark unification and point objectification processing, and match the current surveying and mapping points with historical surveying and mapping points; A boundary chain is constructed based on the matched current mapping points. Preservation sampling is performed on key boundary points in the boundary chain, and random sampling is performed on points outside the boundary chain to obtain a continuous set of boundary sampling points. Write the continuous set of boundary sampling points into the RandLA-Net structure, determine the neighborhood point set of the sampling points, extract the mapping neighborhood relationship between the sampling points and the neighborhood points, and generate mapping neighborhood coding features. The local feature aggregation structure in the RandLA-Net structure is called to calculate the neighborhood relationship score. The neighborhood relationship score is normalized to obtain the neighborhood point aggregation weight. Based on the neighborhood point aggregation weight, the mapping neighborhood coding features are fused to generate the patch boundary enhancement features. Based on the enhanced features of the map patch boundaries, the differences between the current survey point and the historical survey points are decomposed to form survey difference residuals; Extract the residual type of the mapping difference residual, select the backhaul path, perform split backhaul on the mapping difference residual, and generate change identification results; Based on the change identification results, update the mapping data associated with the current mapping point and output the land and resources mapping data processing results.

[0008] Optionally, the spatial reference unification and point objectification processing includes the following steps: Transform land and resources surveying and mapping data into the same spatial coordinate system, and distinguish the current surveying point from the previous surveying point according to the surveying time phase; For the current survey point, select previous survey points that are spatially adjacent, have the same map patch affiliation, and have the same boundary affiliation from the previous survey points; Calculate the planar distance between the current survey point and the selected previous survey points, and take the previous survey point with the smallest planar distance as the historical survey point corresponding to the current survey point.

[0009] Optionally, the construction of the address chain and random sampling includes the following steps: Arrange the current survey points according to the endpoint connection relationship of the matched current survey points in the same patch boundary to obtain the boundary chain; The starting and ending points of the boundary chain, the locations where the boundary direction changes, the intersections of map patch boundaries, and the boundary points whose locations differ from historical survey points are identified as key boundary points. Preservation sampling is performed on critical boundary points, and random sampling is performed on points outside the boundary chain. The critical boundary points and the randomly sampled points outside the boundary chain are combined to form a continuous set of boundary sampling points.

[0010] Optionally, the generation of the mapping neighborhood coding features includes the following steps: Each sampling point in the continuous sampling point set at the boundary is written into the RandLA-Net structure. Based on the spatial proximity relationship of the sampling points in the continuous sampling point set at the boundary, neighborhood points are selected for each sampling point to determine the neighborhood point set. The sampling points are compared item by item with the neighboring points in the neighborhood point set. The comparison includes the patch number, the position of the boundary chain, the direction of the boundary extension, and the positional differences relative to the historical survey points, thus forming the survey neighborhood relationship. The mapping neighborhood relationships are encoded to form mapping neighborhood coding features.

[0011] Optionally, the calculation of the neighborhood relationship score includes the following steps: For each neighboring point under the same sampling point, calculate the planar distance between the neighboring point and the sampling point, and convert the planar distance into a distance score; Compare the connection status of neighboring points and sampling points in the boundary chain, and convert the connection status into boundary continuous scores; Based on the same affiliation, shared edge adjacency, non-shared edge adjacency, and no adjacency status between the neighboring points and the sampled points, a map patch affiliation score is generated. The positional difference direction between the neighboring points and the sampling points and the corresponding historical mapping points is matched with the current boundary chain extension direction to generate a temporal consistency score; The neighborhood relationship score is obtained by summing the distance score, boundary continuity score, patch affiliation score, and temporal consistency score.

[0012] Optionally, the generation of the patch boundary enhancement features includes the following steps: For the same sampling point, the neighborhood relationship scores corresponding to each neighboring point in the neighborhood point set are summarized to form a total score; Divide the neighborhood relationship score of each neighboring point by the sum of the scores to determine the neighborhood aggregation weight corresponding to each neighboring point; The mapping neighborhood coding features are weighted and fused according to the neighborhood point aggregation weight to form the patch boundary enhancement features.

[0013] Optionally, the formation of the mapping difference residual includes the following steps: Using the current survey point and the corresponding historical survey point corresponding to the enhanced feature of the map patch boundary as the difference object, extract the planar location, land type attribute, elevation value and map patch adjacency status of the difference object; The residual type is determined based on the plane position offset, land type attribute difference, elevation difference, and difference in the adjacency status of map patches; The residual type is associated with the corresponding difference content to form the surveying difference residual.

[0014] Optionally, the splitting and backhauling of the mapping difference residuals includes the following steps: Select the return path according to the residual type in the surveying difference residual; When the residual type is boundary position residual, the current mapping points with continuous planar position offsets are back-transmitted along the same boundary chain; When the residual type is land use attribute residual, the current mapping points with consistent land use attribute differences are back-transmitted along the adjacent state of the map patches; When the residual type is elevation change residual, the current mapping points with continuous elevation differences are back-transmitted along the spatial proximity relationship corresponding to the neighboring point set. When the residual type is the patch adjacency residual, the current mapping point with continuous differences in the adjacency state of the patch is back-transmitted along the shared boundary chain; The change identification result is generated based on the residual distribution in each return path.

[0015] Optionally, updating the mapping data includes the following steps: When the change identification result is a change in the boundary position, the boundary of the patch associated with the current survey point is corrected according to the feedback segment corresponding to the change in the boundary position; When the change identification result is a change in land category attribute, the land category attribute of the current mapping point to which the map patch belongs will be updated to the land category attribute corresponding to the change identification result; When the change identification result is an elevation change, the elevation change content of the current survey point is written into the survey data associated with the current survey point; When the change identification result is a change in the adjacency of a patch, the adjacency status of the patch to which the current survey point belongs is updated to the adjacency status of the patch corresponding to the change identification result.

[0016] Optionally, the output of the land and resources surveying and mapping data processing results includes the following steps: The updated map patch boundaries, land use attributes, elevation changes, and map patch adjacency status are summarized to form updated surveying and mapping data. Write the current mapping point corresponding to the local anomaly result into the verification mark; The updated surveying and mapping data and the verification marks will be output together as the land and resources surveying and mapping data processing results.

[0017] The beneficial effects of this invention are: (1) This invention constructs a boundary chain for the current mapping points and performs retention sampling on the key boundary points in the boundary chain. While reducing the amount of data outside the boundary chain, it retains the start and end positions of the boundary, the position of boundary direction change and the intersection position of the map patch boundary, avoiding the destruction of the continuity of the map patch boundary by ordinary random sampling and improving the integrity of boundary data processing.

[0018] (2) The present invention determines the neighborhood point set through the RandLA-Net structure, and generates the mapping neighborhood coding feature by combining the map patch number, boundary chain arrangement position, boundary extension direction and historical mapping point position differences. The neighborhood aggregation process simultaneously reflects the spatial proximity relationship and the mapping business relationship, and improves the ability of the map patch boundary enhancement feature to express the real boundary trend.

[0019] (3) This invention selects the return path according to the residual type of the surveying difference residual and performs split return, which distinguishes continuous changes from local anomalies, reduces misjudgment caused by single-point error, and improves the accuracy of change identification results and land and resources surveying data update results. Attached Figure Description

[0020] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is an overall flowchart of a land and resources surveying and mapping data processing method based on spatial computing proposed in this invention; Figure 2 This is a schematic diagram illustrating the process of boundary chain construction and key boundary point retention sampling in this invention; Figure 3 This is a schematic diagram of the structure of RandLA-Net mapping neighborhood aggregation in this invention. Detailed Implementation

[0021] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0022] refer to Figures 1-3 A method for processing land and resources surveying and mapping data based on spatial computing includes the following steps: Acquire land and resources surveying and mapping data, perform spatial benchmark unification and point objectification processing, and match the current surveying and mapping points with historical surveying and mapping points; A boundary chain is constructed based on the matched current mapping points. Preservation sampling is performed on key boundary points in the boundary chain, and random sampling is performed on points outside the boundary chain to obtain a continuous set of boundary sampling points. Write the continuous set of boundary sampling points into the RandLA-Net structure, determine the neighborhood point set of the sampling points, extract the mapping neighborhood relationship between the sampling points and the neighborhood points, and generate mapping neighborhood coding features. The local feature aggregation structure in the RandLA-Net structure is called to calculate the neighborhood relationship score. The neighborhood relationship score is normalized to obtain the neighborhood point aggregation weight. Based on the neighborhood point aggregation weight, the mapping neighborhood coding features are fused to generate the patch boundary enhancement features. Based on the enhanced features of the map patch boundaries, the differences between the current survey point and the historical survey points are decomposed to form survey difference residuals; Extract the residual type of the mapping difference residual, select the backhaul path, perform split backhaul on the mapping difference residual, and generate change identification results; Based on the change identification results, update the mapping data associated with the current mapping point and output the land and resources mapping data processing results.

[0023] In this embodiment, the spatial reference unification and point objectification processing includes the following steps: Transform land and resources surveying and mapping data into the same spatial coordinate system, and distinguish the current surveying point from the previous surveying point according to the surveying time phase; The survey time phase is read in the order of acquisition time, result version time, and data entry time. If the acquisition time exists, it is used; if the acquisition time is missing, the result version time is used; if both the acquisition time and the result version time are missing, the data entry time is used. Points with the latest survey time phase are used as the current survey point, and points with a survey time phase earlier than the latest time phase are used as the preceding survey point.

[0024] For the current survey point, select previous survey points that are spatially adjacent, have the same map patch affiliation, and have the same boundary affiliation from the previous survey points; When the current survey point and the previous survey point are located in one of the following positional relationships: on the same boundary line segment, adjacent boundary line segment, or in the intersection area of ​​the same map patch, they are determined to be spatially adjacent; when the current survey point and the previous survey point have the same map patch number, they are determined to have the same map patch affiliation; when the current survey point and the previous survey point are both located on the boundary line segment corresponding to the boundary of the same map patch, they are determined to have the same boundary affiliation.

[0025] Calculate the planar distance between the current survey point and the selected previous survey points, and take the previous survey point with the smallest planar distance as the historical survey point corresponding to the current survey point.

[0026] The planar distance is calculated using horizontal coordinates under a unified spatial coordinate system, and elevation values ​​are not included in the calculation. The planar distance is calculated for previous surveying points that meet the requirements of spatial proximity, consistent map patch affiliation, and consistent boundary affiliation. The previous surveying point with the smallest planar distance is taken as the historical surveying point corresponding to the current surveying point.

[0027] In this embodiment, the construction of the address chain and random sampling includes the following steps: Arrange the current survey points according to the endpoint connection relationship of the matched current survey points in the same patch boundary to obtain the boundary chain; First, read the boundary of the current mapping point. Then, treat the current mapping points belonging to the same mapping point boundary as the same sorting object. When the endpoint of the previous current mapping point is connected to the endpoint of the next current mapping point, they are arranged according to the direction of boundary extension; when the endpoints cannot continue to be connected, the consecutively arranged current mapping points are treated as a boundary chain.

[0028] The starting and ending points of the boundary chain, the locations where the boundary direction changes, the intersections of map patch boundaries, and the boundary points whose locations differ from historical survey points are identified as key boundary points. The first and last current survey points in the boundary chain correspond to the start and end positions of the boundary chain. If two adjacent boundary segments extend in different directions, the current survey point at the point of direction change is designated as the boundary direction change position. If two or more map boundary segments connect to the same current survey point, the corresponding current survey point is designated as the map boundary intersection position. If the current survey point differs in position from its corresponding historical survey point, the corresponding current survey point is designated as the boundary point with the positional difference from the historical survey point.

[0029] Preservation sampling is performed on critical boundary points, and random sampling is performed on points outside the boundary chain. The critical boundary points and the randomly sampled points outside the boundary chain are combined to form a continuous set of boundary sampling points.

[0030] First, identify the current mapping points that are not involved in the boundary chain arrangement as boundary chain outside points. All critical boundary points are retained and not included in random sampling. Based on the proportion of critical boundary points to the total number of current mapping points, randomly select a similar proportion of point objects from the boundary chain outside points. The selected boundary chain outside points and critical boundary points together form the boundary continuous sampling point set; boundary chain outside points not selected are not included in the boundary continuous sampling point set.

[0031] In this embodiment, the generation of the mapping neighborhood coding features includes the following steps: Each sampling point in the continuous sampling point set at the boundary is written into the RandLA-Net structure. Based on the spatial proximity relationship of the sampling points in the continuous sampling point set at the boundary, neighborhood points are selected for each sampling point to determine the neighborhood point set. First, read the coordinates and boundary chain positions of each sampling point in the continuous boundary sampling point set. Sampling points adjacent to the sampling point on the same boundary chain are included in the spatial proximity relationship; points outside the boundary chain are included in the spatial proximity relationship according to their coordinate distance from the sampling point. A sampling point itself does not participate in the selection of its own neighboring points; neighboring points are selected according to the spatial proximity relationship to determine the corresponding sampling point's neighboring point set.

[0032] The sampling points are compared item by item with the neighboring points in the neighborhood point set. The comparison includes the patch number, the position of the boundary chain, the direction of the boundary extension, and the positional differences relative to the historical survey points, thus forming the survey neighborhood relationship. Each sampling point is compared with every neighboring point in the neighborhood point set. If the patch numbers are the same, it is recorded as patch consistency; if they are located in adjacent positions within the same boundary chain, it is recorded as continuous boundary chain; the boundary extension direction is determined by the vector pointing from the coordinates of the preceding sampling point to the coordinates of the following sampling point in the boundary chain arrangement, and if the boundary extension directions corresponding to the sampling point and the neighboring point are the same, it is recorded as consistent direction; if both the sampling point and the neighboring point have positional differences relative to their respective historical mapping points, the consistency of their positional differences is recorded. All of the above records constitute the mapping neighborhood relationship.

[0033] The mapping neighborhood relationships are encoded to form mapping neighborhood coding features.

[0034] Map feature consistency, boundary chain continuity, direction consistency, and positional difference with direction consistency are each recorded as 1, while inconsistencies are recorded as 0. Each coded value is arranged according to the order of the sampling point and its neighboring points in the neighboring point set, forming the mapping neighborhood coding feature.

[0035] In this embodiment, the calculation of the neighborhood relationship score includes the following steps: For each neighboring point under the same sampling point, calculate the planar distance between the neighboring point and the sampling point, and convert the planar distance into a distance score; For the same sampling point, the planar distance is calculated for each neighboring point in the neighborhood point set. The planar distance is calculated using the horizontal coordinates of the sampling point and its neighboring points under a unified spatial coordinate system; elevation values ​​are not included in the distance calculation. The planar tolerance is determined by the mapping accuracy level corresponding to the land and resources surveying data. When the planar distance does not exceed the planar tolerance, the distance score is 3; when the planar distance exceeds the planar tolerance but does not exceed twice the planar tolerance, the distance score is 2; when the planar distance exceeds twice the planar tolerance, the distance score is 1.

[0036] Compare the connection status of neighboring points and sampling points in the boundary chain, and convert the connection status into boundary continuous scores; The boundary chain connection state is determined according to the positions of the sampling point and neighboring points in the boundary chain. When the sampling point and neighboring points are located in the same boundary chain and are continuous, the boundary continuity score is 3; when the sampling point and neighboring points are located in adjacent boundary chains with a common endpoint, the boundary continuity score is 2; when both the sampling point and neighboring points are boundary points but do not satisfy the first two connection states, the boundary continuity score is 1; when the neighboring point is not a boundary point, the boundary continuity score is 0.

[0037] Based on the same affiliation, shared edge adjacency, non-shared edge adjacency, and no adjacency status between the neighboring points and the sampled points, a map patch affiliation score is generated. The attribution status of a land parcel is determined based on the spatial relationship between the land use parcels to which the sampling point and its neighboring points belong. When the sampling point and its neighboring point belong to the same land use parcel, the attribution score is 3; when the sampling point and its neighboring point belong to two adjacent parcels sharing the same boundary chain, the attribution score is 2; when the sampling point and its neighboring point belong to two adjacent parcels but do not share the same boundary chain, the attribution score is 1; when the sampling point and its neighboring point do not have an adjacency relationship, the attribution score is 0.

[0038] The positional difference direction between the neighboring points and the sampling points and the corresponding historical mapping points is matched with the current boundary chain extension direction to generate a temporal consistency score; The direction of positional difference is determined by the vector pointing from the coordinates of the corresponding historical mapping point to the coordinates of the current sampling point. The direction of the current boundary chain extension is determined by the vector pointing from the coordinates of the previous sampling point to the coordinates of the next sampling point in the boundary chain. When both the sampling point and its neighboring points have corresponding historical mapping points, and the direction of their positional difference is consistent with the direction of the current boundary chain extension, the temporal consistency score is 3. When both the sampling point and its neighboring points have corresponding historical mapping points, but only one point has a boundary offset, the temporal consistency score is 2. When both the sampling point and its neighboring points have corresponding historical mapping points, but the direction of their positional difference is inconsistent, the temporal consistency score is 1. When at least one of the sampling point or its neighboring points does not have a corresponding historical mapping point, the temporal consistency score is 0.

[0039] The neighborhood relationship score is obtained by summing the distance score, boundary continuity score, patch affiliation score, and temporal consistency score.

[0040] In this embodiment, the generation of the patch boundary enhancement feature includes the following steps: For the same sampling point, the neighborhood relationship scores corresponding to each neighboring point in the neighborhood point set are summarized to form a total score; The set of neighboring points corresponding to the same sampling point is used as the statistical range, and the neighborhood relationship scores of each neighboring point are added together to obtain the total score.

[0041] Divide the neighborhood relationship score of each neighboring point by the sum of the scores to determine the neighborhood aggregation weight corresponding to each neighboring point; The mapping neighborhood coding features are weighted and fused according to the neighborhood point aggregation weight to form the patch boundary enhancement features.

[0042] The aggregated weight of each neighborhood point only applies to the mapping neighborhood coding feature corresponding to the same neighborhood point. First, the aggregated weight of the neighborhood point is multiplied by the corresponding mapping neighborhood coding feature, and then all weighted results under the same sampling point are summed to form the patch boundary enhancement feature of the corresponding sampling point.

[0043] In this embodiment, the formation of the mapping difference residual includes the following steps: Using the current survey point and the corresponding historical survey point corresponding to the enhanced feature of the map patch boundary as the difference object, extract the planar location, land type attribute, elevation value and map patch adjacency status of the difference object; The planar location is represented by horizontal coordinates under a unified spatial coordinate system, and the elevation value is extracted separately. The land use attribute is the land use type of the map patch to which the current survey point belongs. The adjacency status of map patches is determined according to the shared boundary chain between the map patch to which the current survey point belongs and the adjacent map patches. If a shared boundary chain exists, the adjacent map patch number is recorded; if no shared boundary chain exists, the record is empty.

[0044] The residual type is determined based on the plane position offset, land type attribute difference, elevation difference, and difference in the adjacency status of map patches; When the horizontal coordinates of the current survey point are inconsistent with those of the corresponding historical survey point, the residual type is determined to be boundary position residual; when the land use type is inconsistent, the residual type is determined to be land category attribute residual; when the elevation values ​​are inconsistent, the residual type is determined to be elevation change residual; when the adjacent patch numbers are inconsistent, the residual type is determined to be patch adjacency residual.

[0045] The residual type is associated with the corresponding difference content to form the surveying difference residual.

[0046] Boundary position residuals are associated with the planar position offset between the current survey point and its corresponding historical survey point; land use attribute residuals are associated with differences in land use types; elevation change residuals are associated with elevation differences; and parcel adjacency residuals are associated with differences in adjacent parcel numbers. The residual types and the associated differences together form the surveying difference residuals.

[0047] In this embodiment, the splitting and backhauling of the mapping difference residuals includes the following steps: Select the return path according to the residual type in the surveying difference residual; When the residual type is boundary position residual, the current mapping points with continuous planar position offsets are back-transmitted along the same boundary chain; The direction of planar position offset is determined by the vector pointing from the coordinates of the corresponding historical survey point to the coordinates of the current survey point. When the planar position offset directions of adjacent current survey points in the same boundary chain are consistent, they are included in the same boundary position residual backhaul path; when the planar position offset directions are inconsistent, the current backhaul path stops.

[0048] When the residual type is land use attribute residual, the current mapping points with consistent land use attribute differences are back-transmitted along the adjacent state of the map patches; Differences in land use attributes are determined based on the land use attributes of the current mapping point and the corresponding historical mapping points. When the land use attribute types of adjacent current mapping points are consistent before and after the change, they are included in the same land use attribute residual backhaul path; when the land use attribute types are inconsistent before and after the change, the current backhaul path is stopped.

[0049] When the residual type is elevation change residual, the current mapping points with continuous elevation differences are back-transmitted along the spatial proximity relationship corresponding to the neighboring point set. The elevation difference is determined by subtracting the corresponding historical elevation value from the current elevation value of the surveyed point. In the spatial proximity relationship corresponding to the neighboring point set, when the elevation difference directions of adjacent current surveyed points are consistent, they are included in the same elevation change residual backhaul path; when the elevation difference directions are inconsistent, the current backhaul path is stopped.

[0050] When the residual type is the patch adjacency residual, the current mapping point with continuous differences in the adjacency state of the patch is back-transmitted along the shared boundary chain; Differences in the adjacency status of map features are determined based on the changes in the adjacency status of map features on both sides of the shared boundary chain. When the direction of the adjacency status change of map features corresponding to adjacent current mapping points is consistent, they are included in the same map feature adjacency residual backhaul path; when the direction of the adjacency status change of map features is inconsistent, the current backhaul path is stopped.

[0051] The change identification result is generated based on the residual distribution in each return path.

[0052] When the same residual type and direction appear consecutively within the same backhaul path, the corresponding backhaul path is identified as a change identification result. When the residual type and direction of adjacent current mapping points are inconsistent, the current backhaul path stops, and the consecutive consistent backhaul paths before stopping are identified as change identification results separately; the inconsistent locations are marked as local anomalies.

[0053] In this embodiment, updating the surveying data includes the following steps: When the change identification result is a change in the boundary position, the boundary of the patch associated with the current survey point is corrected according to the feedback segment corresponding to the change in the boundary position; The planar position offset of each current survey point within the transmission segment is used as the correction amount for the map patch boundary. The corresponding map patch boundary is adjusted according to the arrangement order of the transmission segment in the boundary chain. When the endpoints of adjacent corrected map patch boundaries are connected, the boundary continuity is maintained according to the endpoint connection relationship.

[0054] When the change identification result is a change in land category attribute, the land category attribute of the current mapping point to which the map patch belongs will be updated to the land category attribute corresponding to the change identification result; When the change identification result is an elevation change, the elevation change content of the current survey point is written into the survey data associated with the current survey point; The elevation change information includes the elevation difference and the location of the elevation change. The elevation difference is entered into the elevation field corresponding to the current survey point, and the location of the elevation change uses the coordinates of the current survey point.

[0055] When the change identification result is a change in the adjacency of a patch, the adjacency status of the patch to which the current survey point belongs is updated to the adjacency status of the patch corresponding to the change identification result.

[0056] When a change in the adjacency of two patches corresponds to a change in the shared boundary chain, the adjacency status of the two patches is adjusted according to the change identification result. If the two patches are still connected after the change in the shared boundary chain, the adjacency status is retained. If the shared boundary chain is broken, the adjacency status is canceled.

[0057] In this embodiment, the output of the land and resources surveying and mapping data processing results includes the following steps: The updated map patch boundaries, land use attributes, elevation changes, and map patch adjacency status are summarized to form updated surveying and mapping data. Using the current survey point number as the summary index, the corrected map patch boundaries, updated land use attributes, elevation changes, and map patch adjacency status are grouped into the same update record to form updated survey data.

[0058] Write the current mapping point corresponding to the local anomaly result into the verification mark; The current survey point corresponding to a local anomaly will not participate in the automatic update of survey data. The verification mark records the current survey point number, anomaly type, and anomaly location.

[0059] The updated surveying and mapping data and the verification marks will be output together as the land and resources surveying and mapping data processing results.

[0060] Example 1: To verify the feasibility of this invention in practice, it was applied to a land and resources surveying and mapping results update scenario. In this scenario, the area to be processed contains a large number of irregular map patch boundaries, some of which exhibit turning, intersection, and local offset. There are inconsistencies in point density, boundary acquisition order, local elevation changes, and changes in map patch adjacency relationships between historical and current surveying and mapping results. Existing processing methods typically involve direct matching based on point distance, followed by identification of changes through ordinary neighborhood aggregation. This approach is prone to mistakenly deleting key boundary points in densely populated boundary areas and misinterpreting errors in individual surveying points as actual boundary changes.

[0061] In this scenario, the current and historical surveying results are first acquired and converted to the same spatial coordinate system. Then, the current surveying point is distinguished from the previous surveying point according to the surveying time phase. For each current surveying point, points that are spatially adjacent, have the same map patch affiliation, and have the same boundary affiliation are selected from the previous surveying points. The corresponding historical surveying point is then determined using the minimum planar distance rule. After matching, a boundary chain is constructed based on the endpoint connection relationship of the current surveying point within the same map patch boundary. The starting and ending positions of the boundary chain, the positions where the boundary direction changes, the intersection positions of map patch boundaries, and the boundary points with positional differences from historical surveying points are identified as key boundary points. All key boundary points are retained, and points outside the boundary chain are randomly sampled to obtain a continuous sampling point set for the boundary. This reduces the number of non-key points while retaining the key positions that affect the expression of map patch boundaries.

[0062] A continuous set of boundary sampling points is written into the RandLA-Net structure. For each sampling point, neighboring points are selected according to spatial proximity to form a neighboring point set. The differences in patch number, boundary chain arrangement, boundary extension direction, and position relative to historical mapping points between the sampling point and its neighboring points are compared to form mapping neighborhood relationships. These mapping neighborhood relationships are then encoded to obtain mapping neighborhood coding features. The local feature aggregation structure calculates distance score, boundary continuity score, patch affiliation score, and temporal consistency score for each neighboring point under the same sampling point. The sum of these four scores forms the neighborhood relationship score. The neighborhood relationship score is then normalized to obtain the neighborhood point aggregation weight. The mapping neighborhood coding features are then fused according to the neighborhood point aggregation weight to form the patch boundary enhancement features. Neighborhood aggregation no longer relies solely on spatial distance but simultaneously reflects boundary connectivity, patch relationships, and historical change relationships.

[0063] After obtaining the enhanced features of the map patch boundaries, the differences between the current survey point and its corresponding historical survey points are decomposed to form surveying difference residuals. Residual types include boundary location residuals, land use attribute residuals, elevation change residuals, and map patch adjacency residuals. Different backhaul paths are selected based on the residual type: boundary location residuals are backhauled along the same boundary chain; land use attribute residuals are backhauled along the map patch adjacency status; elevation change residuals are backhauled along spatial proximity relationships; and map patch adjacency residuals are backhauled along shared boundary chains. When the same residual type and direction appear consecutively within the same backhaul path, a change identification result is formed; when the residual type and direction are inconsistent, the current backhaul path stops, and the inconsistent location is recorded as a local anomaly result. The surveying data associated with the current survey point is updated based on the change identification result, local anomalies are marked for verification, and the land and resources surveying data processing result is output.

[0064] To verify the processing effect of this invention, its method was compared with conventional distance matching. Both methods used the same input data, including the current survey point, historical survey points, map boundary, elevation information, and land use attribute information. During the comparison, manually verified survey results were used as the verification basis, and the matching of survey points, the correspondence of boundary points, the location of boundary position changes, the misjudgment of local anomalies, and the updating of survey data were statistically analyzed. Specific comparison results are shown in Table 1.

[0065] Table 1 Comparison of the Update Effects of Land and Resources Surveying and Mapping Results

[0066] As shown in Table 1, the accuracy of mapping point matching and the completeness of boundary point correspondence of the method of the present invention are both higher than those of conventional distance matching. The accuracy of mapping point matching increased from 84.3% to 96.1%, and the completeness of boundary point correspondence increased from 82.7% to 95.4%. This indicates that the present invention, through spatial benchmark unification, point objectification processing, and screening of patch and boundary assignments, can reduce erroneous correspondences between different patch boundaries; through boundary chain construction and key boundary point retention sampling, it can preserve the start and end positions of boundaries, the positions of boundary direction changes, and the intersection positions of patch boundaries, and the sampled patch boundaries still have good continuity and completeness.

[0067] In terms of boundary change localization and local anomaly control, the boundary position change localization deviation of the method of this invention is 0.17 meters, lower than the 0.48 meters of conventional distance matching; the local anomaly misjudgment rate is 4.3%, lower than the 12.9% of conventional distance matching. This indicates that the present invention, through mapping neighborhood relationship encoding, neighborhood relationship score calculation, and patch boundary enhancement feature generation, can more accurately express the boundary connection relationship and historical point position difference relationship between the current mapping point and neighboring points; furthermore, through mapping difference residual backhaul, it can distinguish continuous and consistent real boundary changes from local anomaly points, reducing the misjudgment of changes caused by single-point errors.

[0068] The mapping data update completeness rate of the method of this invention reaches 95.7%, which is higher than the 85.1% of conventional distance matching. This indicates that the present invention can update the boundary of the land parcel, land type attributes, elevation changes, and adjacency status of the land parcels according to the change identification results, and write local anomalies into the verification markers for joint output. Therefore, the present invention can form a complete processing chain in terms of mapping point matching, boundary continuity maintenance, change location, anomaly control, and result updating, thereby improving the accuracy and completeness of land and resources mapping results updates.

[0069] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for processing land and resources surveying and mapping data based on spatial computing, characterized in that, Includes the following steps: Acquire land and resources surveying and mapping data, perform spatial benchmark unification and point objectification processing, and match the current surveying and mapping points with historical surveying and mapping points; A boundary chain is constructed based on the matched current mapping points. Preservation sampling is performed on key boundary points in the boundary chain, and random sampling is performed on points outside the boundary chain to obtain a continuous set of boundary sampling points. Write the continuous set of boundary sampling points into the RandLA-Net structure, determine the neighborhood point set of the sampling points, extract the mapping neighborhood relationship between the sampling points and the neighborhood points, and generate mapping neighborhood coding features. The local feature aggregation structure in the RandLA-Net structure is called to calculate the neighborhood relationship score. The neighborhood relationship score is normalized to obtain the neighborhood point aggregation weight. Based on the neighborhood point aggregation weight, the mapping neighborhood coding features are fused to generate the patch boundary enhancement features. Based on the enhanced features of the map patch boundaries, the differences between the current survey point and the historical survey points are decomposed to form survey difference residuals; Extract the residual type of the mapping difference residual, select the backhaul path, perform split backhaul on the mapping difference residual, and generate change identification results; Based on the change identification results, update the mapping data associated with the current mapping point and output the land and resources mapping data processing results.

2. The land and resources surveying and mapping data processing method based on spatial computing according to claim 1, characterized in that, The spatial benchmark unification and point objectification process includes the following steps: Transform land and resources surveying and mapping data into the same spatial coordinate system, and distinguish the current surveying point from the previous surveying point according to the surveying time phase; For the current survey point, select previous survey points that are spatially adjacent, have the same map patch affiliation, and have the same boundary affiliation from the previous survey points; Calculate the planar distance between the current survey point and the selected previous survey points, and take the previous survey point with the smallest planar distance as the historical survey point corresponding to the current survey point.

3. The land and resources surveying and mapping data processing method based on spatial computing according to claim 1, characterized in that, The construction of the address chain and random sampling includes the following steps: Arrange the current survey points according to the endpoint connection relationship of the matched current survey points in the same patch boundary to obtain the boundary chain; The starting and ending points of the boundary chain, the locations where the boundary direction changes, the intersections of map patch boundaries, and the boundary points whose locations differ from historical survey points are identified as key boundary points. Preservation sampling is performed on critical boundary points, and random sampling is performed on points outside the boundary chain. The critical boundary points and the randomly sampled points outside the boundary chain are combined to form a continuous set of boundary sampling points.

4. The land and resources surveying and mapping data processing method based on spatial computing according to claim 1, characterized in that, The generation of the mapping neighborhood coding features includes the following steps: Each sampling point in the continuous sampling point set at the boundary is written into the RandLA-Net structure. Based on the spatial proximity relationship of the sampling points in the continuous sampling point set at the boundary, neighborhood points are selected for each sampling point to determine the neighborhood point set. The sampling points are compared item by item with the neighboring points in the neighborhood point set. The comparison includes the patch number, the position of the boundary chain, the direction of the boundary extension, and the positional differences relative to the historical survey points, thus forming the survey neighborhood relationship. The mapping neighborhood relationships are encoded to form mapping neighborhood coding features.

5. The land and resources surveying and mapping data processing method based on spatial computing according to claim 1, characterized in that, The calculation of the neighborhood relationship score includes the following steps: For each neighboring point under the same sampling point, calculate the planar distance between the neighboring point and the sampling point, and convert the planar distance into a distance score; Compare the connection status of neighboring points and sampling points in the boundary chain, and convert the connection status into boundary continuous scores; Based on the same affiliation, shared edge adjacency, non-shared edge adjacency, and no adjacency status between the neighboring points and the sampled points, a map patch affiliation score is generated. The positional difference direction between the neighboring points and the sampling points and the corresponding historical mapping points is matched with the current boundary chain extension direction to generate a temporal consistency score; The neighborhood relationship score is obtained by summing the distance score, boundary continuity score, patch affiliation score, and temporal consistency score.

6. The land and resources surveying and mapping data processing method based on spatial computing according to claim 5, characterized in that, The generation of the patch boundary enhancement features includes the following steps: For the same sampling point, the neighborhood relationship scores corresponding to each neighboring point in the neighborhood point set are summarized to form a total score; Divide the neighborhood relationship score of each neighboring point by the sum of the scores to determine the neighborhood aggregation weight corresponding to each neighboring point; The mapping neighborhood coding features are weighted and fused according to the neighborhood point aggregation weight to form the patch boundary enhancement features.

7. The land and resources surveying and mapping data processing method based on spatial computing according to claim 1, characterized in that, The formation of the mapping difference residuals includes the following steps: Using the current survey point and the corresponding historical survey point corresponding to the enhanced feature of the map patch boundary as the difference object, extract the planar location, land type attribute, elevation value and map patch adjacency status of the difference object; The residual type is determined based on the plane position offset, land type attribute difference, elevation difference, and difference in the adjacency status of map patches; The residual type is associated with the corresponding difference content to form the surveying difference residual.

8. The land and resources surveying and mapping data processing method based on spatial computing according to claim 1, characterized in that, The splitting and backhauling of the mapping difference residuals includes the following steps: Select the return path according to the residual type in the surveying difference residual; When the residual type is boundary position residual, the current mapping points with continuous planar position offsets are back-transmitted along the same boundary chain; When the residual type is land use attribute residual, the current mapping points with consistent land use attribute differences are back-transmitted along the adjacent state of the map patches; When the residual type is elevation change residual, the current mapping points with continuous elevation differences are back-transmitted along the spatial proximity relationship corresponding to the neighboring point set. When the residual type is the patch adjacency residual, the current mapping point with continuous differences in the adjacency state of the patch is back-transmitted along the shared boundary chain; The change identification result is generated based on the residual distribution in each return path.

9. A method for processing land and resources surveying and mapping data based on spatial computing according to claim 1, characterized in that, The updating of the mapping data includes the following steps: When the change identification result is a change in the boundary position, the boundary of the patch associated with the current survey point is corrected according to the feedback segment corresponding to the change in the boundary position; When the change identification result is a change in land category attribute, the land category attribute of the current mapping point to which the map patch belongs will be updated to the land category attribute corresponding to the change identification result; When the change identification result is an elevation change, the elevation change content of the current survey point is written into the survey data associated with the current survey point; When the change identification result is a change in the adjacency of a patch, the adjacency status of the patch to which the current survey point belongs is updated to the adjacency status of the patch corresponding to the change identification result.

10. A method for processing land and resources surveying and mapping data based on spatial computing according to claim 9, characterized in that, The output of the land and resources surveying and mapping data processing results includes the following steps: The updated map patch boundaries, land use attributes, elevation changes, and map patch adjacency status are summarized to form updated surveying and mapping data. Write the current mapping point corresponding to the local anomaly result into the verification mark; The updated surveying and mapping data and the verification marks will be output together as the land and resources surveying and mapping data processing results.

Images