Building cross-scale semi-supervised merging method and system under scale factor constraint

Abstract

The invention discloses a building cross-scale semi-supervised merging method and system under scale factor constraint. The method comprises the following steps: 1, obtaining vector building data, a synthesized target scale and related scale factors; 2, preprocessing the vector building data according to the related scale factors, and constructing a building relation proximity graph; 3, obtaining a plurality of building bridging surface candidate line segments according to the building relation proximity graph; 4, extracting features of the candidate line segments of the bridging surfaces of the buildings, and inputting the features into a classification support vector machine for screening; and 5, combining the vector buildings according to the screened candidate line segments of the bridge surfaces of the buildings. According to the building cross-scale semi-supervised merging method under the constraint of the scale factors, intelligent merging of the buildings under different scales is realized by introducing the scale factors and a semi-supervised learning technology, and the accuracy of geographic information expression and the practicability of a map are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of cartographic synthesis, and in particular to a method and system for semi-supervised merging of buildings across scales under scale factor constraints. Background Art

[0002] As an important component of geographic spatial vector databases, buildings are one of the foundations for expressing spatial phenomena, supporting spatial analysis and providing spatial services. They are the core elements of large-scale urban maps and have an important impact on the effect of map expression. When the map scale is reduced, due to the crowding and overlap of map content, buildings need to be merged, deleted, typified or collapsed. Among these methods, merging has always been the main comprehensive operator. The merging operation constitutes the most common hierarchical structure between buildings on maps of different scales. It aims to combine separated buildings by filling space when the scale is reduced. By considering the shape structure, spatial distribution and human perception of buildings, a group of visually inseparable adjacent buildings at a larger scale are replaced by a single building at a smaller scale to clearly represent the polygonal features of buildings on small-scale maps.

[0003] With the vigorous development of computer mapping and geographic information systems, building merging has evolved from traditional manual mapping to automatic mapping synthesis, and has made great progress. For a long time, the research results on building polygon merging can be divided into two categories: merging methods based on raster data and merging methods based on vector data. Learning with vector data is more suitable for the generalization process of mapping, and it can be expected that this will become an important research direction for the generalization of intelligent mapping in the future. As an important content of cartographic synthesis, building merging has always been one of the key points and difficulties of cartographic synthesis research at home and abroad.

[0004] Currently, users have put forward higher requirements for the timeliness and consistency of digital spatial information. With the development of geographic information systems (GIS) and remote sensing technology, the representation of maps at different scales has become increasingly important. The scale factor refers to the ratio of a certain distance on the map to the actual distance on the ground, and is a core factor affecting map synthesis. At present, few studies related to cartographic synthesis have considered the impact of scale factors on the rationality of the results. Reasonable map synthesis can effectively convey geographic information at different scales, while unreasonable synthesis may lead to information distortion or loss. Therefore, it is of great significance to study the impact of scale factors on map synthesis. The scale factor is a crucial parameter in the process of map synthesis, which directly affects the representation accuracy of geographic information and the user experience. Summary of the invention

[0005] In order to at least partially solve the problem of geographic information distortion or loss caused by unreasonable selection of scale factors, the present invention provides a method and system for cross-scale semi-supervised merging of buildings under scale factor constraints, which processes vector building data by obtaining scale factors, then constructs candidate line segments of building bridge surfaces, screens the candidate line segments of building bridge surfaces through a classification support vector machine, and uses the screened candidate line segments of building bridge surfaces to perform appropriate building merging. The present invention takes into account the problem of cross-scale intelligent merging of buildings under scale factor constraints, can more effectively convey geographic information, and improve the practicality and readability of maps.

[0006] In order to achieve the above object, the technical solution of the present invention is:

[0007] The first aspect of the present invention proposes a method for semi-supervised merging of buildings across scales under scale factor constraints, comprising:

[0008] Step 1: Obtain vector building data, determine the integrated target scale and the relevant scale factor of the target scale, so as to facilitate the processing of vector building data and the construction of building relationship proximity map;

[0009] Step 2: preprocess the vector building data according to the relevant scale factor, and construct a building relationship proximity map based on the preprocessed vector building data to facilitate subsequent feature extraction;

[0010] Step 3: Obtain candidate line segments of various building bridge surfaces according to the building relationship proximity graph;

[0011] Step 4: extract the features of candidate line segments of bridge surfaces of multiple buildings, input the features into a classification support vector machine for screening, and obtain the screened candidate line segments of bridge surfaces of buildings, so as to facilitate vector building merging;

[0012] Step 5: Merge vector buildings based on the screened candidate line segments of the building bridge surfaces.

[0013] Furthermore, the relevant scale factors include a threshold of details that can be integrated, a spacing threshold for dividing the merged group, an area threshold for expressing the vacant space after integration, and an area threshold for independent buildings that can be integrated, so as to facilitate the merging of vector buildings.

[0014] Furthermore, the preprocessing of the vector building data according to the relevant scale factor specifically includes:

[0015] Delete buildings that are too small to be integrated after the scale change, so that independent buildings can be merged;

[0016] Delete building shared arc sections to merge topologically adjacent buildings to facilitate the merging of topologically adjacent buildings.

[0017] Furthermore, the step of obtaining multiple candidate line segments of building bridge surfaces according to the building relationship proximity graph specifically includes:

[0018] Adjacent buildings in the building relationship proximity graph are connected by connecting edges, and the two buildings connected by the connecting edge are merged into a merged group for easy subsequent processing;

[0019] Construct a minimum enclosing rectangle outside the merged group, and generate adjacent edges of two merged groups under the constraint of the minimum enclosing rectangle;

[0020] The candidate line segments of the building bridge surface are constructed according to the adjacent edges, and the candidate line segments of the building bridge surface include the candidate line segments of the perpendicular lines of the adjacent edge endpoints, the candidate line segments of the adjacent edge extensions, and the candidate line segments of the perpendicular lines of the angle bisectors of the adjacent edge extensions.

[0021] Furthermore, constructing candidate line segments of building bridge surfaces according to adjacent edges specifically includes:

[0022] Extract the endpoints of the adjacent edges of the buildings in the merged group and make a perpendicular line segment to the adjacent edge of another building in the merged group, wherein the perpendicular line is a candidate perpendicular line segment of the adjacent edge endpoints;

[0023] Extending the adjacent edge of the building polygon to intersect the adjacent edge of another building in the merged group, the extended line segment is a candidate line segment for the adjacent edge extension line;

[0024] In the merged group, the adjacent sides of the building polygon are extended and intersected to obtain intersection points, and the intersection points located on the adjacent lines or outside the minimum circumscribed rectangle of the merged group are removed, and the remaining intersection points are retained as valid intersection points. An angle bisector is made with the valid intersection points, and a perpendicular line is made through the angle bisector to intersect the adjacent side. The perpendicular line segment of the angle bisector of the adjacent side is a candidate line segment for the perpendicular line of the angle bisector of the extended angle of the adjacent side.

[0025] Furthermore, the features of extracting candidate line segments of multiple building bridge surfaces include spatial position features of the candidate line segments, geometric features of the candidate line segments, connection features of the candidate line segments with adjacent edges, and intersection features of the candidate line segments with the minimum circumscribed rectangle of the building.

[0026] Furthermore, the spatial position features of the candidate line segment include two end point coordinates and a midpoint coordinate of the candidate line segment;

[0027] The geometric features of the candidate line segments include the length, direction and variation of the candidate line segments; the variation includes the absolute value of the difference between the direction of a candidate line segment and the average direction of all candidate line segments in the building merge group;

[0028] The connection features between the candidate line segment and the adjacent edge include the absolute value of the difference between the direction of the candidate line segment and the direction of the centroid of the minimum circumscribed rectangle of the building and the minimum distance between the two endpoints of the candidate line segment and the endpoint of the adjacent edge;

[0029] The intersection features of the candidate line segment and the minimum circumscribed rectangle of the building include two intersection points of the candidate line segment and the minimum circumscribed rectangle and the minimum distance between the endpoints of the minimum circumscribed rectangle.

[0030] The second aspect of the present invention proposes a semi-supervised merging system for buildings across scales under scale factor constraints, comprising:

[0031] A collection module is used to obtain vector building data, determine the integrated target scale and the relevant scale factor of the target scale, so as to facilitate the processing of the vector building data and the construction of the building relationship proximity map;

[0032] A processing module, used for preprocessing the vector building data according to the relevant scale factor, and constructing a building relationship proximity graph according to the preprocessed vector building data to facilitate subsequent feature extraction;

[0033] A candidate line segment module, used for obtaining candidate line segments of bridge surfaces of various buildings according to a building relationship proximity graph;

[0034] A screening module is used to extract features of candidate line segments of bridge surfaces of various buildings, input the features into a classification support vector machine for screening, and obtain the screened candidate line segments of bridge surfaces of buildings, so as to facilitate vector building merging;

[0035] The merging module is used to merge vector buildings according to the filtered candidate line segments of the building bridge surfaces.

[0036] The third aspect of the present invention proposes an electronic device, comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, the method for semi-supervised merging of buildings across scales under scale factor constraints as described in the first aspect above is implemented.

[0037] The fourth aspect of the present invention proposes a computer-readable storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the storage medium is located is controlled to execute a method for semi-supervised merging of buildings across scales under scale factor constraints as described in the first aspect above.

[0038] Beneficial effects of the present invention:

[0039] (1) Improving the accuracy of geographic information expression: By introducing the scale factor as a key constraint in the merging process, the present invention can ensure that the merging results of buildings at different scales are more accurate. This helps to reduce information distortion or loss, more realistically reflect the spatial distribution and morphological characteristics of geographic entities, and improve the accuracy of geographic information expression.

[0040] (2) Enhance the practicality and readability of the map: The present invention makes the representation of buildings in the map more concise and clear by intelligently merging buildings. The merged buildings not only retain key spatial structural information, but also reduce redundant details, thereby improving the readability and practicality of the map. Users can obtain the required information more quickly, thus improving the efficiency of map use.

[0041] (3) Improve data processing efficiency: Compared with the traditional manual merging method, the present invention adopts semi-supervised learning technology to realize the automated processing of building merging. This greatly reduces the need for manual intervention and improves the speed and efficiency of data processing. At the same time, through optimization algorithms and parallel processing technology, the present invention can process large-scale data sets and meet the needs of real-time applications.

[0042] (4) Enhanced flexibility and adaptability of the system: The cross-scale merging method proposed in the present invention has strong flexibility and adaptability. It can process building data at different scales and intelligently merge them according to actual needs. In addition, the system architecture and algorithm design of the present invention are also easy to expand and modify to adapt to new needs and technical challenges that may arise in the future.

[0043] (5) Promoting the sharing and interoperability of geographic information data: By realizing the cross-scale merging of building data, the present invention helps promote the sharing and interoperability of geographic information data. The merged building data is more standardized and unified, which facilitates data exchange and integration between different systems. This helps break down information silos and promotes the integration and utilization of geographic information resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 A flowchart of a method for semi-supervised merging of buildings across scales under scale factor constraints provided in an embodiment of the present invention.

[0045] Figure 2 This is a schematic diagram of the merged vector buildings provided by an embodiment of the present invention.

[0046] Figure 3 A specific flow chart of a method for semi-supervised merging of buildings across scales under scale factor constraints provided in an embodiment of the present invention.

[0047] Figure 4 A schematic diagram of adjacent edges of a merged group provided in an embodiment of the present invention.

[0048] Figure 5 A schematic diagram of constructing three types of candidate line segments provided in an embodiment of the present invention.

[0049] Figure 6 A schematic diagram of features of candidate line segments of a building bridge surface provided by an embodiment of the present invention.

[0050] Figure 7 A schematic diagram of feature extraction of candidate line segments of a building bridge surface provided by an embodiment of the present invention.

[0051] Figure 8 An architecture diagram of a cross-scale semi-supervised merging system of buildings under scale factor constraints provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution in the embodiment of the present invention will be clearly described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is a part of the embodiment of the present invention, not all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0053] Example 1

[0054] like Figure 1 As shown in FIG. 1 , a semi-supervised merging method for buildings across scales under scale factor constraints includes:

[0055] S101: Acquire vector building data, and determine the integrated target scale and related scale factors of the target scale.

[0056] Specifically, by searching the Internet, literature and library books for relevant scale factors of the target scale, including:

[0057] (1) The relevant scaling factors include detail thresholds that can be integrated.

[0058] (2) The spacing threshold for dividing the merged groups.

[0059] (3) The area threshold of the open space after integration.

[0060] (4) Area thresholds of independent buildings that can be integrated.

[0061] Taking the requirements of 1:1000 or 1:5000 topographic maps as an example: convex and concave parts less than 1mm on the map can generally be comprehensively represented; house spacing less than 1.5mm on the map can be comprehensively represented; minor streets and lanes can be appropriately selected; vacant land can be selected according to the characteristics of southern and northern residential areas, and the selection index is generally 4mm on the map.2 -9mm 2 ; For independent houses, those with a length less than 1.0 mm and a width less than 0.7 mm on the drawing shall be represented by non-scale symbols.

[0062] S102: pre-processing the vector building data according to relevant scale factors, and constructing a building relationship proximity graph according to the pre-processed vector building data.

[0063] S103: Obtain multiple types of building bridge surface candidate line segments according to the building relationship proximity graph.

[0064] S104: extracting features of multiple candidate line segments of building bridge surfaces, and inputting the features into a classification support vector machine for screening to obtain screened candidate line segments of building bridge surfaces.

[0065] Specifically, the candidate line segments of the bridge surfaces of various buildings are divided into 60% training set, 20% validation set and 20% test set, and 15% of the training set is manually labeled. The features of the extracted candidate line segments of the bridge surfaces of various buildings are standardized and input into a classification support vector machine (OCSVM). The kernel function is used to construct a classification support vector machine, and the training set consisting of partial labeled data and unlabeled data is used for training. After the training is completed, the trained OCSVM model is used to classify the test set to predict the candidate line segment data class that needs to be retained. The cross-validation method is used to select the optimal parameters of the OCSVM classifier.

[0066] In order to evaluate the classification effect of the algorithm, it is also necessary to confirm the classification effect by comparing it with the manual classification results. The classification accuracy P, recall rate R and F value are calculated according to the following formula:

[0067]

[0068] Among them, P is the classification accuracy, R is the recall rate, F is the F value, TP is the number of correct classifications, FP is the number of wrong classifications, and FN is the number of missed classifications.

[0069] S105: Merging vector buildings according to the screened building bridge surface candidate line segments.

[0070] Specifically, the building bridging surface is formed according to the screened building bridging surface candidate line segments and the adjacent edges of the building, and it is extracted together with the edges of the original merged group as the outer contour of the building to realize the merging of vector buildings.

[0071] When merging, the convex and concave details of the extracted outer contours are processed according to the scale factor, and the extracted outer contours are expressed in space, so that the merging result is closer to the true value, and the unreasonable shapes in the extracted outer contours, such as the bridge surface intersection, are reconstructed. Specifically, after obtaining the features of the candidate line segments of the building bridge surface, such as Figure 2 As shown in (a), the minimum concave and convex details of the extracted outer contour are integrated, as shown in Figure 2 As shown in (b), the extracted outer contour is expressed in a clear space to make the merged result closer to the true value, as shown in Figure 2 As shown in (c), unreasonable shapes in the extracted outer contour, such as bridge surface intersections, are post-processed by bridge surface reconstruction.

[0072] When generating candidate line segments for bridge surfaces, the present invention not only considers the right-angle features, geometric features, and spatial structural relationships of building polygons, but also extracts multiple features of candidate line segments for comprehensive evaluation. These features include spatial position features, geometric features, connection features with adjacent edges, and intersection features with the minimum circumscribed rectangle of the building. Through multi-feature fusion and comprehensive evaluation, the present invention can more accurately screen out the optimal candidate line segments for merging, thereby improving the quality and reliability of the merging results.

[0073] like Figure 3 As shown, the present invention pre-processes the vector building data by obtaining the target scale and the relevant scale factor of the target scale, then constructs a building relationship proximity map of the vector building, finds the candidate line segments of the building bridge surface, screens the candidate line segments of the building bridge surface through a classification support vector machine, and finally merges the vector buildings according to the screened candidate line segments of the building bridge surface. The problem of intelligent merging of buildings across scales under the constraint of scale factors is considered, which can more effectively convey geographic information and improve the practicality and readability of the map.

[0074] Example 2

[0075] Based on the above embodiment, the embodiment of the present invention provides a specific process of constructing candidate line segments of building bridge surfaces and extracting features, which specifically includes:

[0076] Use the preprocessed vector building data to construct a Delaunay triangulation network, generate skeleton lines, and finally generate a building relationship proximity map. In the building relationship proximity map, two buildings are connected by a connecting edge. The length of the connecting edge is extracted, and according to the scale factor constraint (the spacing threshold for dividing the merged group), the connecting edge between the buildings that need to be integrated is retained, and the two buildings connected by the connecting edge are grouped into a merged group. Figure 4 As shown, adjacent edges are constructed based on the merged groups.

[0077] Three types of building bridge surface candidate segments are constructed according to the adjacent edges: the candidate segment of the perpendicular line of the adjacent edge endpoint, the candidate segment of the adjacent edge extension line, and the candidate segment of the perpendicular line of the angle bisector of the adjacent edge extension line.

[0078] like Figure 5 As shown in (a), the endpoints of the adjacent edges of the buildings in the merged group are extracted to make a perpendicular line segment to the adjacent edge of another building in the merged group. The perpendicular line is a candidate perpendicular line segment for the endpoints of the adjacent edges.

[0079] like Figure 5 As shown in (b), the adjacent edge of the extended building polygon intersects with the adjacent edge of another building in the merged group, and the extended line segment is a candidate line segment for the adjacent edge extension line.

[0080] like Figure 5 As shown in (c), the adjacent sides of the building polygon in the merged group are extended and intersected to obtain intersection points. The intersection points located on the adjacent lines or outside the minimum circumscribed rectangle of the merged group are removed, and the remaining intersection points are retained as valid intersection points. An angle bisector is drawn with the valid intersection points, and a perpendicular line is drawn through the angle bisector to intersect the adjacent side. The perpendicular line segment of the angle bisector of the adjacent side is the candidate line segment of the perpendicular line of the angle bisector of the extended angle of the adjacent side.

[0081] like Figure 6 As shown, the features of the candidate line segments of the bridge surfaces of multiple buildings are extracted, including the spatial position features of the candidate line segments, the geometric features of the candidate line segments, the connection features of the candidate line segments with adjacent edges, and the intersection features of the candidate line segments with the minimum circumscribed rectangle of the building.

[0082] The spatial position features of the candidate line segments include the coordinates of the two endpoints and the midpoint of the candidate line segments. Specifically, the spatial position coordinates of the two endpoints of each candidate line segment and the midpoint coordinates of the line segment are extracted, and the midpoint coordinates of each candidate line segment are calculated by calculating the average value of the two endpoint coordinates. If the endpoint coordinates are (x1, y1) and (x2, y2), the coordinates of the midpoint are:

[0083]

[0084] Among them, x m is the x coordinate of the midpoint, y m is the y-axis coordinate of the midpoint, (x1, y1) and (x2, y2) are the coordinates of the two endpoints of the candidate line segment.

[0085] The geometric features of the candidate line segments include the length, direction and variation of the candidate line segments.

[0086] The degree of variation includes the absolute value of the difference between the direction of a candidate line segment and the average direction of all candidate line segments in the building merge group.

[0087] Specifically, the direction of the candidate line segment is expressed by the following formula:

[0088] θ=atan2(y2-y1,x2-x1)

[0089] Among them, θ is the direction of the candidate line segment, atan2 is the inverse tangent function, which takes into account the signs of the numerator and denominator, and the return value range is [-π, π], that is, [-180°, 180°], (x1, y1) and (x2, y2) are the coordinates of the two endpoints of the candidate line segment respectively.

[0090] The connection features between the candidate line segment and the adjacent edge include the absolute value of the difference between the direction of the candidate line segment and the direction of the centroid of the minimum circumscribed rectangle of the building and the minimum distance between the two endpoints of the candidate line segment and the endpoint of the adjacent edge;

[0091] The intersection features of the candidate line segment and the minimum enclosing rectangle of the building include two intersection points of the candidate line segment and the minimum enclosing rectangle and the minimum distance between the endpoints of the minimum enclosing rectangle.

[0092] The connection features between the candidate line segment and the adjacent edge include the consistency between the direction of the candidate line segment and the main direction of the building and the minimum distance between the two endpoints of the line segment and the endpoints of the adjacent edge.

[0093] The consistency between the candidate line segment direction and the main direction of the building is calculated based on the generated minimum enclosing rectangle, and the absolute value of the difference between the line segment direction and the direction of the centroid of the minimum enclosing rectangle of the building.

[0094] like Figure 7 As shown in (a), the minimum distance between the endpoints of the adjacent edge is the shortest distance between the two endpoints of the candidate line segment and the endpoints of the adjacent edge respectively extracted according to the generated adjacent edge.

[0095] The intersection features of the minimum enclosing rectangle of the candidate line segment building include the minimum distance between the two intersection points of the line segment and the minimum enclosing rectangle and the endpoints of the minimum enclosing rectangle. Based on the constructed merged group minimum enclosing rectangle, the minimum distance between the two intersection points of the candidate line segment and the minimum enclosing rectangle and the endpoints of the minimum enclosing rectangle is calculated. Figure 7 (b), the length of dis1 and dis2 is the minimum distance between the two intersection points of the red candidate line and the minimum enclosing rectangle and the endpoints of the minimum enclosing rectangle.

[0096] Example 3

[0097] Corresponding to the above method, such as Figure 8 As shown in FIG. 1 , a semi-supervised merging system for buildings across scales under scale factor constraints includes:

[0098] The collection module is used to obtain vector building data, determine the integrated target scale and the related scale factors of the target scale.

[0099] The processing module is used to pre-process the vector building data according to the relevant scale factor, and construct a building relationship proximity map according to the pre-processed vector building data.

[0100] The candidate line segment module is used to obtain multiple candidate line segments of building bridge surfaces according to the proximity of building relationships.

[0101] The screening module is used to extract the features of candidate line segments of multiple building bridge surfaces, input the features into a classification support vector machine for screening, and obtain the screened candidate line segments of the building bridge surfaces.

[0102] The merging module is used to merge vector buildings according to the filtered candidate line segments of the building bridge surfaces.

[0103] It should be noted that the system for semi-supervised merging of buildings across scales under scale factor constraints provided in an embodiment of the present invention is to implement a method for semi-supervised merging of buildings across scales under scale factor constraints in the above-mentioned embodiment. Its specific functions can be referred to the above-mentioned method embodiments, which will not be repeated here.

[0104] Example 4

[0105] Corresponding to the above method, an embodiment of the present invention provides an electronic device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements a method for semi-supervised merging of buildings across scales under a scale factor constraint as in the above embodiment.

[0106] An embodiment of the present invention provides a computer-readable storage medium, the storage medium including a stored computer program, wherein when the computer program is running, the device where the storage medium is located is controlled to execute a method for semi-supervised merging of buildings across scales under scale factor constraints as described in the above embodiment.

[0107] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A semi-supervised merging method for buildings across scales under scale factor constraints, characterized in that: include: Step 1: Obtain vector building data, determine the integrated target scale and the relevant scale factors of the target scale; Step 2: preprocessing the vector building data according to the relevant scale factor, and constructing a building relationship proximity map according to the preprocessed vector building data; Step 3: Obtain candidate line segments of various building bridge surfaces according to the building relationship proximity graph; Step 4: extracting features of candidate line segments of multiple building bridge surfaces, inputting the features into a classification support vector machine for screening, and obtaining screened candidate line segments of building bridge surfaces; Step 5: Merge vector buildings based on the screened candidate line segments of the building bridge surfaces.

2. According to the method of claim 1, the semi-supervised merging method of buildings across scales under scale factor constraints is characterized by: The relevant scale factors include a threshold of details that can be integrated, a threshold of spacing for dividing and merging groups, a threshold of area expressed by vacant land after integration, and a threshold of area of ​​independent buildings that can be integrated.

3. According to the method of claim 1, the semi-supervised merging method of buildings across scales under scale factor constraints is characterized by: The preprocessing of the vector building data according to the relevant scale factors specifically includes: Delete buildings whose area is too small to be integrated after the scale change; Delete building-shared arcs to merge topologically adjacent buildings.

4. The method for semi-supervised merging of buildings across scales under scale factor constraints according to claim 1, characterized in that: The step of obtaining multiple candidate line segments of building bridge surfaces according to the building relationship proximity graph specifically includes: Adjacent buildings in the building relationship proximity graph are connected by connecting edges, and the two buildings connected by the connecting edge are merged into a merged group; Construct a minimum enclosing rectangle outside the merged group, and generate adjacent edges of two merged groups under the constraint of the minimum enclosing rectangle; The candidate line segments of the building bridge surface are constructed according to the adjacent edges, and the candidate line segments of the building bridge surface include the candidate line segments of the perpendicular lines of the adjacent edge endpoints, the candidate line segments of the adjacent edge extensions, and the candidate line segments of the perpendicular lines of the angle bisectors of the adjacent edge extensions.

5. The method for semi-supervised merging of buildings across scales under scale factor constraints according to claim 4, characterized in that: The step of constructing candidate line segments of building bridge surfaces according to adjacent edges specifically includes: Extract the endpoints of the adjacent edges of the buildings in the merged group and make a perpendicular line segment to the adjacent edge of another building in the merged group, wherein the perpendicular line is a candidate perpendicular line segment of the adjacent edge endpoints; Extending the adjacent edge of the building polygon to intersect the adjacent edge of another building in the merged group, the extended line segment is a candidate line segment for the adjacent edge extension line; In the merged group, the adjacent sides of the building polygon are extended and intersected to obtain intersection points, and the intersection points located on the adjacent lines or outside the minimum circumscribed rectangle of the merged group are removed, and the remaining intersection points are retained as valid intersection points. An angle bisector is made with the valid intersection points, and a perpendicular line is made through the angle bisector to intersect the adjacent side. The perpendicular line segment of the angle bisector of the adjacent side is a candidate line segment for the perpendicular line of the angle bisector of the extended angle of the adjacent side.

6. The method for semi-supervised merging of buildings across scales under scale factor constraints according to claim 1, characterized in that: The features of the candidate line segments of the bridge surfaces of various buildings are extracted, specifically including the spatial position features of the candidate line segments, the geometric features of the candidate line segments, the connection features of the candidate line segments with adjacent edges, and the intersection features of the candidate line segments with the minimum circumscribed rectangle of the building.

7. The method for semi-supervised merging of buildings across scales under scale factor constraints according to claim 6, characterized in that: The spatial position features of the candidate line segment include the coordinates of two endpoints and the midpoint of the candidate line segment; The geometric features of the candidate line segments include the length, direction and variation of the candidate line segments; the variation includes the absolute value of the difference between the direction of a candidate line segment and the average direction of all candidate line segments in the building merge group; The connection features between the candidate line segment and the adjacent edge include the absolute value of the difference between the direction of the candidate line segment and the direction of the centroid of the minimum circumscribed rectangle of the building and the minimum distance between the two endpoints of the candidate line segment and the endpoint of the adjacent edge; The intersection features of the candidate line segment and the minimum circumscribed rectangle of the building include two intersection points of the candidate line segment and the minimum circumscribed rectangle and the minimum distance between the endpoints of the minimum circumscribed rectangle.

8. A semi-supervised merging system for buildings across scales under scale factor constraints, characterized by: include: A collection module is used to obtain vector building data, determine the integrated target scale and the related scale factors of the target scale; A processing module, used for preprocessing the vector building data according to the relevant scale factor, and constructing a building relationship proximity map according to the preprocessed vector building data; A candidate line segment module, used for obtaining candidate line segments of bridge surfaces of various buildings according to a building relationship proximity graph; A screening module is used to extract features of candidate line segments of multiple building bridge surfaces, input the features into a classification support vector machine for screening, and obtain screened candidate line segments of building bridge surfaces; The merging module is used to merge vector buildings according to the filtered candidate line segments of the building bridge surfaces.

9. An electronic device, characterized in that: The method comprises a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, a method for semi-supervised merging of buildings across scales under scale factor constraints as described in any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that: The storage medium includes a stored computer program, wherein when the computer program is running, the device where the storage medium is located is controlled to execute the cross-scale semi-supervised merging method of buildings under scale factor constraints as described in any one of claims 1 to 7.

Images