Different-view-angle geographic entity alignment method and device and storage medium

Abstract

The invention provides a different-view-angle geographic entity alignment method and device and a storage medium, and relates to the technical field of image processing. The method comprises the following steps: for a geographic entity in each remote sensing image, calculating an azimuth angle between a central entity and each surrounding entity by taking a main axis of the central entity as a coordinate axis to obtain an azimuth angle sequence corresponding to the central entity, and calculating a spatial distance between the central entity and each surrounding entity to obtain a spatial distance sequence corresponding to the central entity; determining the similarity between the first geographic entity and the second geographic entity based on the two sequences corresponding to the first geographic entity and the second geographic entity, and rotating the second remote sensing image by taking the difference between the main axes of the first geographic entity and the second geographic entity with the highest similarity as a rotation angle to obtain a third remote sensing image, and aligning the first geographic entity with the third geographic entity. The influence of visual angle deflection and geometric distortion on entity alignment is reduced, and the accuracy of geographic entity alignment is improved.

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and in particular to a method, device and storage medium for aligning geographic entities from different viewing angles. Background Art

[0002] Entity alignment, also known as entity matching or entity resolution, is a core technology in the field of knowledge integration. Its core task is to identify different descriptions of the same real-world entity in different knowledge graphs. At present, entity alignment technology mainly relies on embedding-based models such as TransE (Translating Embeddings) and its variant models, cross-language entity alignment models such as JAPE (Joint Attribute-Preserving Embedding) or graph convolutional neural networks. Through supervised or semi-supervised learning methods, it uses specific entities and corpora for training to achieve higher accuracy.

[0003] In remote sensing applications, due to the various ways of acquiring remote sensing data (such as different sensors and resolutions such as infrared, radar, visible light, and hyperspectral), the extracted geographic entities may have significant differences in visibility and invisibility. In addition, due to differences in image shooting angles, the extracted geographic entities will have varying degrees of spatial distortion, that is, the acquired image data itself and multiple geographic entities in the image have geometric deformation and position offset, which greatly increases the difficulty of the geographic entity alignment task. The traditional entity alignment model is no longer applicable, resulting in a decrease in the accuracy of geographic entity alignment. Summary of the invention

[0004] The present invention provides a different-viewing angle geographic entity alignment method, device and storage medium, which are used to solve the technical problem of low accuracy of geographic entity alignment in the prior art.

[0005] In a first aspect, the present invention provides a method for aligning geographic entities from different perspectives, comprising the following steps.

[0006] For each geographical entity in a remote sensing image, any geographical entity is regarded as a central entity, and all geographical entities around the central entity are regarded as surrounding entities; Taking the main axis of the central entity as the coordinate axis, calculating the azimuth between the central entity and each surrounding entity to obtain an azimuth sequence corresponding to the central entity, and calculating the spatial distance between the central entity and each surrounding entity to obtain a spatial distance sequence corresponding to the central entity; Determine the similarity between the first geographical entity and the second geographical entity based on the azimuth sequence and the spatial distance sequence respectively corresponding to the first geographical entity and the second geographical entity; the first geographical entity is a geographical entity in the first remote sensing image; the second geographical entity is a geographical entity in the second remote sensing image; The second remote sensing image is rotated by taking the difference between the main axes of the first geographic entity and the second geographic entity with the highest similarity as a rotation angle to obtain a third remote sensing image; The first geographic entity is aligned with a third geographic entity; the third geographic entity is a geographic entity in the third remote sensing image.

[0007] In some embodiments, the determining the similarity between the first geographical entity and the second geographical entity based on the azimuth sequence and the spatial distance sequence respectively corresponding to the first geographical entity and the second geographical entity includes: Pairing the first geographic entity with the second geographic entity in pairs to obtain multiple groups of entity pairs; Obtaining a first similarity and a second similarity corresponding to each entity pair; the first similarity is the similarity between an azimuth sequence corresponding to a first geographic entity in the entity pair and an azimuth sequence corresponding to a second geographic entity in the entity pair; the second similarity is the similarity between a spatial distance sequence corresponding to the first geographic entity in the entity pair and a spatial distance sequence corresponding to the second geographic entity in the entity pair; The similarity between the first geographic entity and the second geographic entity in each entity pair is determined based on the first similarity and the second similarity corresponding to the entity pair.

[0008] In some embodiments, the first similarity is represented by a first dynamic time warping distance, which is a dynamic time warping distance between an azimuth sequence corresponding to a first geographic entity in the entity pair and an azimuth sequence corresponding to a second geographic entity in the entity pair; the second similarity is represented by a second dynamic time warping distance, which is a dynamic time warping distance between a spatial distance sequence corresponding to the first geographic entity in the entity pair and a spatial distance sequence corresponding to the second geographic entity in the entity pair; The determining the similarity between the first geographic entity and the second geographic entity in the entity pair based on the first similarity and the second similarity corresponding to each entity pair includes: The first dynamic time warping distance and the second dynamic time warping distance are weightedly summed to obtain a weighted value corresponding to each entity pair; the weighted value is used to characterize the similarity between the first geographic entity and the second geographic entity in the entity pair.

[0009] In some embodiments, aligning the first geographic entity with the third geographic entity includes: Obtaining an azimuth sequence and a spatial distance sequence corresponding to a third geographic entity; Determine the similarity between the first geographical entity and the third geographical entity based on the azimuth sequence and the spatial distance sequence corresponding to the first geographical entity and the azimuth sequence and the spatial distance sequence corresponding to the third geographical entity; An aligned entity pair is determined based on the similarity between the first geographic entity and the third geographic entity.

[0010] In some embodiments, determining the aligned entity pair based on the similarity between the first geographic entity and the third geographic entity includes: Selecting the candidate entity pair with the highest similarity from the candidate entity pairs as the aligned entity pair; the candidate entity pair is obtained by pairing the first geographic entity and the third geographic entity in pairs; Deleting the candidate entity pair including the first geographic entity in the aligned entity pair and the candidate entity pair including the third geographic entity in the aligned entity pair from the candidate entity pairs to obtain updated candidate entity pairs; Iteratively executing the steps of selecting the candidate entity pair with the highest similarity from the updated candidate entity pairs as the aligned entity pair and updating the candidate entity pairs until the updated candidate entity pairs are empty, thereby obtaining all the aligned entity pairs; The first geographic entity or the third geographic entity that is not paired into the aligned entity pair is regarded as a missing entity.

[0011] In some embodiments, the principal axis of the central entity is obtained by the following steps: Obtaining a centralized coordinate point through the boundary coordinates of the central entity; Calculating the covariance matrix of the central coordinate point, and performing eigendecomposition on the covariance matrix to obtain eigenvalues ​​and eigenvectors; The eigenvector corresponding to the largest eigenvalue is selected as the principal axis of the central entity.

[0012] In some embodiments, the method further comprises: extracting geographic entities from the first remote sensing image and the second remote sensing image; The first remote sensing image and the second remote sensing image are obtained by photographing the same type of geographical entities in the same area under different shooting conditions.

[0013] In a second aspect, the present invention provides a device for aligning geographic entities from different perspectives, comprising the following modules.

[0014] The first determination module is used to determine, for each geographical entity in the remote sensing image, any geographical entity as a central entity and all geographical entities around the central entity as surrounding entities; A first acquisition module is used to calculate the azimuth between the central entity and each surrounding entity with the main axis of the central entity as the coordinate axis to obtain an azimuth sequence corresponding to the central entity, and calculate the spatial distance between the central entity and each surrounding entity to obtain a spatial distance sequence corresponding to the central entity; A second determination module is used to determine the similarity between the first geographical entity and the second geographical entity based on the azimuth sequence and the spatial distance sequence respectively corresponding to the first geographical entity and the second geographical entity; the first geographical entity is a geographical entity in the first remote sensing image; the second geographical entity is a geographical entity in the second remote sensing image; A second acquisition module is used to use the difference between the main axes of the first geographic entity and the second geographic entity with the highest similarity as a rotation angle, and rotate the second remote sensing image to obtain a third remote sensing image; An alignment module is used to align the first geographic entity with a third geographic entity; the third geographic entity is a geographic entity in the third remote sensing image.

[0015] In a third aspect, the present invention provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method for aligning geographic entities from different perspectives as described above is implemented.

[0016] In a fourth aspect, a non-transitory computer-readable storage medium stores a computer program, which, when executed by a processor, implements any of the above-described methods for aligning geographic entities from different perspectives.

[0017] In a fifth aspect, the present invention further provides a computer program product, comprising a computer program, which, when executed by a processor, implements the different-perspective geographic entity alignment method as described in the first aspect above.

[0018] The different-viewing angle geographic entity alignment method, device and storage medium provided by the present invention, for the geographic entities in each remote sensing image, any geographic entity is taken as the central entity, all the geographic entities around the central entity are taken as the surrounding entities, and the main axis of the central entity is taken as the coordinate axis, the azimuth between the central entity and each surrounding entity is calculated to obtain the azimuth sequence corresponding to the central entity, and the spatial distance between the central entity and each surrounding entity is calculated to obtain the spatial distance sequence corresponding to the central entity; then, based on the azimuth sequence and the spatial distance sequence corresponding to the geographic entities in different remote sensing images, the similarity between the geographic entities in different remote sensing images is determined, the difference between the main axes of the first geographic entity and the second geographic entity with the highest similarity is taken as the rotation angle, the second remote sensing image is rotated to obtain the third remote sensing image, and finally the first geographic entity is aligned with the third geographic entity in the third remote sensing image, thereby reducing the influence of the viewing angle deflection and geometric distortion on the entity alignment, ensuring the correct alignment of the geographic entities for the geographic data with large rotation and geometric distortion, and improving the accuracy of the geographic entity alignment. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0020] Figure 1 This is one of the flow charts of the different-viewing angle geographic entity alignment method provided by the present invention.

[0021] Figure 2 It is a schematic diagram of traversal direction for traversing geographical entities based on the main axis provided by the present invention.

[0022] Figure 3 This is the second flow chart of the different-viewing angle geographic entity alignment method provided by the present invention.

[0023] Figure 4 It is a structural schematic diagram of the different-viewing angle geographic entity alignment device provided by the present invention.

[0024] Figure 5 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0025] In the process of obtaining information about the same type of geographic entities in the same area, there are remote sensing images (called source data) obtained from different sensors, different flight altitudes, different shooting angles, and different times, which causes deviations in the extracted geographic entity target information in terms of spatial position, geometric form, number of entities (visible and invisible differences), etc. How to align geographic entity data obtained from different source data one by one - that is, entity a (building, vehicle, etc.) in image A is an entity b in image B, is a problem that needs to be solved.

[0026] Traditional methods often rely on the attribute information and relative spatial relationship of geographic entities to manually annotate or calculate and infer the same entities for entity alignment, but this method has two limitations when performing large-scale entity alignment. First, this method requires the ability to obtain relatively stable and unchanging attribute information of geographic entities. In fact, when extracting multimodal remote sensing targets, even for the same type of targets, the main attribute information (such as color) presented on different modal remote sensing data (such as infrared, radar, visible light, etc.) is not exactly the same. The applicability of methods that rely solely on attribute information for judgment is greatly reduced in this case. Second, traditional methods usually use relative spatial relationships to align based on the premise that two or more data have the same geographic spatial reference. However, in actual application scenarios where multiple sensors are used to shoot and identify the same area, it is often not possible to obtain images at the same angle every time; images shot at different angles usually have large spatial geometric distortions; when the entity has a large deflection and geometric distortion, traditional methods cannot correctly and automatically align the same geographic entity.

[0027] Based on the above technical problems, the present invention proposes a method for aligning geographic entities from different perspectives. For the geographic entities in each remote sensing image, any geographic entity is taken as the central entity, all the geographic entities around the central entity are taken as the surrounding entities, and the main axis of the central entity is taken as the coordinate axis. The azimuth between the central entity and each surrounding entity is calculated to obtain an azimuth sequence corresponding to the central entity, and the spatial distance between the central entity and each surrounding entity is calculated to obtain a spatial distance sequence corresponding to the central entity; then, based on the azimuth sequence and spatial distance sequence corresponding to the geographic entities in different remote sensing images, the similarity between the geographic entities in different remote sensing images is determined, and the difference between the main axes of the first geographic entity and the second geographic entity with the highest similarity is taken as the rotation angle, the second remote sensing image is rotated to obtain a third remote sensing image, and finally the first geographic entity is aligned with the third geographic entity in the third remote sensing image. The problem of constructing distance and azimuth sequences clockwise directly through the horizontal coordinate axis, which results in misalignment of corresponding measurement values ​​of geographic entities in the sequence and thus incorrect calculation of DTW values ​​of corresponding entity distance and azimuth sequences, is avoided. The present invention constructs a coordinate system based on the principal axis of the central entity, and constructs a correct sequence based on the coordinate system, thereby reducing the influence of viewing angle deflection and geometric distortion on entity alignment, and ensuring the correct alignment of geographic entities for geographic data that undergoes significant rotation and geometric distortion.

[0028] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of 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.

[0029] Figure 1 FIG. 1 is one of the flow charts of the different-viewing angle geographic entity alignment method provided by the present invention. Figure 1 As shown, the present invention provides a method for aligning geographical entities from different viewing angles. The method comprises: Step 101: For each geographical entity in a remote sensing image, any geographical entity is regarded as a central entity, and all geographical entities around the central entity are regarded as surrounding entities.

[0030] Specifically, in the embodiment of the present application, the data related to geographic entities extracted from remote sensing images are input in a file format that is a general geographic information file, such as a shapefile file (.shp). The file consists of multiple geographic entity face vectors (Polygons), and its attributes only contain the unique identifier (OID) of each geographic entity and the geographic coordinate position of each entity.

[0031] For each extracted geographic entity, the geographic entity is regarded as the central entity, and all the geographic entities around the geographic entity are regarded as the surrounding entities.

[0032] For example, the first remote sensing image and the second remote sensing image are obtained by shooting the same type of geographical entities in the same area under different shooting conditions. The geographical entities extracted from the first remote sensing image and the second remote sensing image may have deviations in spatial position, geometric shape, number of entities, etc. Entities a, b, and c are extracted from the first remote sensing image, and entities d, e, and f are extracted from the second remote sensing image. For the first remote sensing image, with entity a as the central entity, the surrounding entities of a are entities b and c, with entity b as the central entity, the surrounding entities of b are entities a and c, with entity c as the central entity, the surrounding entities of c are entities a and b; the second remote sensing image can obtain each central entity and its corresponding surrounding entities in the same way.

[0033] Step 102: Taking the principal axis of the central entity as the coordinate axis, calculate the azimuth between the central entity and each surrounding entity to obtain an azimuth sequence corresponding to the central entity, and calculate the spatial distance between the central entity and each surrounding entity to obtain a spatial distance sequence corresponding to the central entity.

[0034] Specifically, taking the principal axis of the central entity as the coordinate axis, the azimuth angle between the central entity and each of its corresponding surrounding entities is calculated to obtain an azimuth angle sequence corresponding to the central entity; and the spatial distance between the central entity and each of its surrounding entities is calculated to obtain a spatial distance sequence corresponding to the central entity.

[0035] For example, taking any geographic entity a in the first remote sensing image as an example, the azimuths between the central entity a and its surrounding entities b and c are calculated as follows: , , the azimuth sequence corresponding to the central entity a is obtained as . The spatial distances between the central entity a and its surrounding entities b and c are calculated as D 1. D 2. The spatial distance sequence corresponding to the central entity a is obtained as .

[0036] Specifically, to calculate the azimuth between the central entity and each of its corresponding surrounding entities, it is necessary to first establish a new coordinate system for the central entity, that is, to use the main axis of the central entity as the coordinate axis, and the main axis interval can be divided into and The calculation of azimuth is considered separately in two cases.

[0037] Figure 2 is a schematic diagram of the traversal direction of the geographical entity based on the main axis provided by the present invention, such as Figure 2 As shown, for the main axis interval Geographic entities (such as Figure 2 The main axis angle of the geographical entity is 78 degrees), taking the entity as the center entity, traverse the surrounding geographical entities (i.e. surrounding entities) clockwise. For a geographical entity in an image (i.e. remote sensing image), , the center point coordinates are marked as ( ), for its surrounding entities , the center point coordinates are marked as ( ),entity With geographic entities The azimuth between The calculation formula is as follows: ( i =1,2,3,..., num ; j =1,2,3,..., num ) ( i =1,2,3,..., num ; j =1,2,3,..., num )

[0038]

[0039] in, For remote sensing images geographic entity )’s horizontal and vertical coordinates; num is the total number of geographic entities in the remote sensing image; For remote sensing images geographic entity )’s horizontal and vertical coordinates; Representing geographic entities With geographic entities The difference in the horizontal coordinates of Representing geographic entities With geographic entities The difference in the vertical coordinates. Representing geographic entities With geographic entities The angle between the line and the east direction; Representing geographic entities The angle between the main axis and true north.

[0040] For the main axis interval Geographic entities (such as Figure 2The main axis angle of the geographic entity is 98 degrees), and the surrounding entities are traversed clockwise. The calculation process is the same as the previous case (the main axis interval is geographic entity), the main axis interval is The azimuth between the geographic entity and its surrounding entities The calculation formula is as follows:

[0041] Specifically, the spatial distance between entities can be calculated based on the center coordinates of the geographic entities. Taking the longitude and latitude coordinates as an example, the spatial distance calculation method is as follows:

[0042] in, d Represents the actual distance between points A and B, in km; r Represents the average radius of the Earth, which is 6371.393km; ya 1. x 1 represents the latitude and longitude coordinates of A, yb 1. xb 1 represents the latitude and longitude coordinates of B respectively.

[0043] After obtaining the azimuths and spatial distances between the central entity and all its surrounding entities and normalizing them, the multiple azimuths and multiple spatial distances are organized into azimuth arrays (i.e. azimuth sequence) and distance arrays (i.e. spatial distance sequence), respectively. The two ordered sets can be expressed as: , , which serves as a feature sequence representing the target geographic entity.

[0044] It should be noted that when constructing the spatial distance sequence, normalization is used instead of directly using the calculated distance value. This is because when the ground is photographed at different heights, the distance between objects in the two images is different; when the geographic reference system information is lost, the distance calculation based on the image does not represent the actual physical distance, and the distance between objects in the two images has a significant difference in absolute value, so normalization is required to use their relative spatial relationship for similarity calculation.

[0045] Step 103: determine the similarity between the first geographic entity and the second geographic entity based on the azimuth sequence and spatial distance sequence respectively corresponding to the first geographic entity and the second geographic entity; the first geographic entity is a geographic entity in the first remote sensing image; the second geographic entity is a geographic entity in the second remote sensing image.

[0046] Specifically, after obtaining the azimuth sequence and spatial distance sequence corresponding to each geographic entity, the similarity between the geographic entities can be determined by comparing the similarity of the sequences between the geographic entities from different remote sensing images.

[0047] Step 104: Taking the difference between the main axes of the first geographic entity and the second geographic entity with the highest similarity as a rotation angle, the second remote sensing image is rotated to obtain a third remote sensing image.

[0048] Specifically, after obtaining the similarity between the first geographic entity and the second geographic entity in all entity pairs, the first geographic entity and the second geographic entity with the highest similarity are determined, and the difference between the main axes of the two geographic entities is used as the rotation angle. The second remote sensing image is rotated according to the rotation angle to obtain the rotated second remote sensing image, which is recorded as the third remote sensing image.

[0049] When there are multiple pairs of the first geographic entity and the second geographic entity with the highest similarity, the average of the principal axis differences of the multiple pairs of entities is taken as the rotation angle.

[0050] Step 105: align the first geographic entity with a third geographic entity; the third geographic entity is a geographic entity in the third remote sensing image.

[0051] Specifically, after obtaining the third remote sensing image, the third geographic entity in the third remote sensing image is matched and aligned with the first geographic entity to achieve accurate alignment of the geographic entities.

[0052] The different-perspective geographic entity alignment method provided in the embodiment of the present application uses the principal axis of each entity as the coordinate axis to construct the surrounding entity orientation sequence, and then calculates the similarity between the two entities. After completing the first similarity calculation (similarity calculation between the first geographic entity and the second geographic entity), the principal axis angle difference of the most similar entity pair is used as the rotation angle to rotate the second remote sensing image, thereby aligning the geographic entity in the rotated image with the first geographic entity. This strategy can effectively reduce the impact of perspective deviation on geographic entity alignment, and can improve the accuracy of geographic entity alignment for geographic data that has undergone significant rotation and geometric distortion.

[0053] In some embodiments, the method further comprises: extracting geographic entities from the first remote sensing image and the second remote sensing image; The first remote sensing image and the second remote sensing image are obtained by photographing the same type of geographical entities in the same area under different shooting conditions.

[0054] Specifically, in the embodiment of the present application, taking the alignment of geographic entities in a first remote sensing image and a second remote sensing image as an example, the first remote sensing image and the second remote sensing image are obtained by photographing the same type of geographic entities in the same area under different shooting conditions.

[0055] Among them, different shooting conditions may include different sensors, different flight altitudes, different shooting angles or different times, etc., which may cause deviations in the extracted geographic entity target information in terms of spatial position, geometric shape, number of entities (visibility differences), etc.

[0056] In some embodiments, the principal axis of the central entity is obtained by the following steps: Obtaining a centralized coordinate point through the boundary coordinates of the central entity; Calculating the covariance matrix of the central coordinate point, and performing eigendecomposition on the covariance matrix to obtain eigenvalues ​​and eigenvectors; The eigenvector corresponding to the largest eigenvalue is selected as the principal axis of the central entity.

[0057] Specifically, when performing entity alignment, the principal axis angles of all geographic entities in each image are first calculated based on the principal component analysis method, and then the azimuths of the surrounding entities and the central entity are calculated using the principal axis of each central entity as the coordinate axis.

[0058] Figure 3 FIG. 2 is a flow chart of the method for aligning geographic entities from different perspectives provided by the present invention. Figure 3 As shown, after executing the geographic entity data file input (including the geographic entity data file f1 of the first remote sensing image and the geographic entity data file f2 of the second remote sensing image), the principal axis of each geographic entity in the input file is calculated one by one, and the principal component analysis method can be used for calculation.

[0059] First, the centralized coordinate point matrix, that is, the centralized coordinate point, is obtained through the boundary coordinates of the central entity; then, the covariance matrix of the centralized coordinate point is calculated, and the covariance matrix is ​​eigendecomposed to obtain eigenvalues ​​and eigenvectors, from which the eigenvector corresponding to the largest eigenvalue is selected as the principal axis of the central entity.

[0060] For example, extract coordinate points from a central entity (such as a polygon, line, or point set). For a polygon, these are usually the vertex coordinates of the outer ring (i.e., boundary coordinates). All the extracted boundary coordinate points form a m ×2 matrix P ,in m is the number of boundary coordinate points of the central entity ( m >0, 2 means the coordinate dimension of each point is 2 ( x and yCalculate the mean of all boundary coordinate points of the central entity and , find the difference between each coordinate point and the mean to get the centralized coordinate point matrix X , X The calculation formula is as follows:

[0061] in, Represents the central coordinate point matrix; Represents the matrix composed of boundary coordinate points; and are the mean abscissa and ordinate of all boundary coordinate points of the central entity respectively.

[0062] Then, calculate the covariance matrix of the centered coordinate points:

[0063] in, represents the covariance matrix of the centered coordinate points, X is the centered coordinate point matrix, m The number of boundary coordinate points of the center entity.

[0064] Next, the covariance matrix Perform characteristic decomposition and solve the formula as follows:

[0065] in, represents the feature vector, Represents the eigenvalue.

[0066] Finally, sort the eigenvalues ​​in descending order and select the eigenvector corresponding to the largest eigenvalue. As the main axis, , indicating the direction in which the data changes the most. The angle of the principal axis is the angle between the eigenvector and the coordinate axis, and the calculation formula is as follows:

[0067] in, Indicates the main axis angle; , are the eigenvectors corresponding to the largest eigenvalues. of The direction and y Directional weight.

[0068] In the embodiment of the present application, the calculated main axis angle Convert to Get the final spindle angle within the interval , The calculation formula is as follows:

[0069] The different-perspective geographic entity alignment method provided in the embodiment of the present application calculates the principal axis angle of all geographic entities in each image according to the principal component analysis method, so that the azimuth between the central entity and the surrounding entities can be calculated using the principal axis as the coordinate axis, and the azimuth similarity judgment and the similarity judgment of the geographic entities are performed based on the azimuth, thereby reducing the impact of perspective deviation on the geographic entity alignment and improving the accuracy of geographic entity alignment.

[0070] In some embodiments, the determining the similarity between the first geographical entity and the second geographical entity based on the azimuth sequence and the spatial distance sequence respectively corresponding to the first geographical entity and the second geographical entity includes: Pairing the first geographic entity with the second geographic entity in pairs to obtain multiple entity pairs; the first geographic entity is a geographic entity in the first remote sensing image; the second geographic entity is a geographic entity in the second remote sensing image; Obtaining a first similarity and a second similarity corresponding to each entity pair; the first similarity is the similarity between an azimuth sequence corresponding to a first geographic entity in the entity pair and an azimuth sequence corresponding to a second geographic entity in the entity pair; the second similarity is the similarity between a spatial distance sequence corresponding to the first geographic entity in the entity pair and a spatial distance sequence corresponding to the second geographic entity in the entity pair; The similarity between the first geographic entity and the second geographic entity in each entity pair is determined based on the first similarity and the second similarity corresponding to the entity pair.

[0071] Specifically, the geographic entities in different remote sensing images are paired two by two to obtain multiple groups of entity pairs.

[0072] For example, the geographic entities in the first remote sensing image (i.e., the first geographic entity) include a, b, and c, and the geographic entities in the second remote sensing image (i.e., the second geographic entity) include d, e, and f. Then, the first geographic entity and the second geographic entity are paired in pairs to obtain multiple entity pairs, including (a, d), (a, e), (a, f), (b, d), (b, e), (b, f), (c, d), (c, e), and (c, f).

[0073] The first similarity and the second similarity corresponding to each entity pair are obtained, that is, the similarity between the azimuth sequence corresponding to the first geographic entity in the entity pair and the azimuth sequence corresponding to the second geographic entity in the entity pair, and the similarity between the spatial distance sequence corresponding to the first geographic entity and the spatial distance sequence corresponding to the second geographic entity in the entity pair are obtained. Then, based on the first similarity and the second similarity corresponding to each entity pair, the similarity between the corresponding first geographic entity and the second geographic entity is determined.

[0074] For example, for the entity pair (a, d), obtain the similarity Sim-1 between the azimuth sequence T1 corresponding to the first geographic entity a and the azimuth sequence T2 corresponding to the second geographic entity d, obtain the similarity Sim-2 between the spatial distance sequence S1 corresponding to a and the spatial distance sequence S2 corresponding to d, and determine the similarity between a and d based on Sim-1 and Sim-2.

[0075] The different-viewing angle geographic entity alignment method provided in the embodiment of the present application determines the similarity between geographic entities by analyzing the similarity of the azimuth angle sequence and the spatial distance sequence corresponding to each geographic entity, and takes into account the influence of the deflection angle and geometric distortion on the geographic entity alignment, thereby improving the accuracy of the geographic entity alignment.

[0076] In some embodiments, the first similarity is represented by a first dynamic time warping distance, which is a dynamic time warping distance between an azimuth sequence corresponding to a first geographic entity in the entity pair and an azimuth sequence corresponding to a second geographic entity in the entity pair; the second similarity is represented by a second dynamic time warping distance, which is a dynamic time warping distance between a spatial distance sequence corresponding to the first geographic entity in the entity pair and a spatial distance sequence corresponding to the second geographic entity in the entity pair; The determining the similarity between the first geographic entity and the second geographic entity in the entity pair based on the first similarity and the second similarity corresponding to each entity pair includes: The first dynamic time warping distance and the second dynamic time warping distance are weightedly summed to obtain a weighted value corresponding to each entity pair; the weighted value is used to characterize the similarity between the first geographic entity and the second geographic entity in the entity pair.

[0077] Specifically, since the image data is from different sources, the number of geographic entities may be different, and information such as location coordinates may also be biased. Traditional methods for judging vector similarity, such as directly calculating the Euclidean distance, are not feasible. In time series analysis, Dynamic Time Warping (DTW) is a method specifically used to calculate the similarity of two time series, especially for time series of different lengths and rhythms.

[0078] In an embodiment of the present application, a dynamic time warping distance is used to characterize the similarity between azimuth sequences and the similarity between spatial distance sequences, and a weighted value of the dynamic time warping distance of the two sequences is used to characterize the similarity of corresponding entity pairs.

[0079] For example, Figure 3 As shown in the figure, after obtaining the principal axis of each geographic entity, the azimuth sequence and spatial distance sequence corresponding to each geographic entity are calculated. Specifically, each geographic entity is calculated one by one: the geographic entity is taken as the central entity, the principal axis of the central entity is taken as the coordinate axis, and the azimuths of multiple surrounding geographic entities (i.e., surrounding entities) and the central entity are calculated to form an azimuth sequence (i.e., Figure 3 ; and calculate the spatial distances between multiple surrounding geographic entities and the central entity, and normalize them to form a spatial distance sequence (i.e. Figure 3 distance sequence in .

[0080] After obtaining the azimuth sequence and spatial distance sequence, the following operations are performed on the geographic entities in f1 (i.e., the first geographic entity) one by one: pair them with all geographic entities in f2 (i.e., the second geographic entity) to form entity pairs, and use the dynamic time warping algorithm to calculate and weight the DTW distances of the azimuth sequence and distance sequence between the entity pairs; obtain the entity pair with the smallest DTW distance weight value as the most similar entity pair, and record the DTW distance weight value dw and the difference between the main axes of the two geographic entities. α is considered as the rotation angle between the two images.

[0081] The specific process of obtaining the weighted value corresponding to each entity pair is as follows: For each geographic entity, the constructed direction and distance sequence formula is as follows:

[0082]

[0083] in, Indicates The azimuth sequence corresponding to the geographical entities, Respectively represent The azimuth of a geographic entity and its surrounding entities, Indicates the total number of geographic entities; Indicates The spatial distance sequence corresponding to each geographic entity; Respectively represent The spatial distance between a geographic entity and its surrounding entities.

[0084] Assume that the first The azimuth sequence corresponding to the first geographic entity is , the spatial distance sequence is , the lengths are , m represents the total number of geographic entities in f1; The azimuth sequence corresponding to the geographic entity (i.e., the second geographic entity) is , the spatial distance sequence is , the lengths are , n Represents the total number of geographic entities in f2.

[0085] Generate respectively ( m -1)×( n -1), wherein the distance metric matrix M is used to store the local difference calculation between the distance vectors, and the orientation metric matrix N is used to store the local difference calculation between the orientation vectors.

[0086] From the lower left cell of the distance metric matrix M To the upper right cell , calculate each pair of distance vector elements and Difference:

[0087] Similarly, from the lower left cell of the orientation metric matrix N To the upper right cell , calculate each orientation vector element and Difference:

[0088] in, represents the element at the corresponding position in the distance metric matrix M, Represents the element at the corresponding position in the orientation measurement matrix N.

[0089] When calculating the distance, find the minimum distance path from the starting point of the two sequences to the end point of the two sequences (that is, from one end of the diagonal to the other end). Starting from the lower left corner, you can go right, up, or to the upper right. Compare the distances after going in these three directions and choose the one with the minimum distance. If the distances are equal, choose to go in the upper right direction first, and so on.

[0090] Start matching the two sequences from the lower left corner of the metric matrix. At each point, the distances calculated for all previous points are accumulated. This dynamic programming is used to recursively calculate the minimum cumulative value for each position in the matrix until the entire matrix is ​​filled. For the distance metric matrix M, calculate the minimum cumulative distance corresponding to each matrix cell :

[0091] in, Represents the distance from the 0th row and 0th column in the distance metric matrix M to Line Minimum cumulative distance of column positions; Represents the distance metric matrix M Line Elements of column positions; Represents the distance from the 0th row and 0th column in the distance metric matrix M to Line The minimum cumulative distance of column positions, It means from row 0, column 0 to Line The minimum cumulative distance of column positions, It means from row 0, column 0 to Line The final spatial distance sequence DTW distance is the distance metric matrix M in the upper right corner (i.e. m -1 line m -1 column position) corresponding to the minimum cumulative distance .

[0092] For the orientation metric matrix N, calculate the minimum cumulative distance corresponding to each matrix cell :

[0093] in, Represents the position from the 0th row and 0th column in the orientation measurement matrix N to the Line Minimum cumulative distance of column positions; Represents the first Line Elements of column positions; Represents the position from the 0th row and 0th column in the orientation measurement matrix N to the Line The minimum cumulative distance of column positions, It means from row 0, column 0 to Line The minimum cumulative distance of column positions, It means from row 0, column 0 to Line The final azimuth sequence DTW distance is the position of the upper right corner of the azimuth metric matrix N (i.e. m -1 line m -1 column position) corresponding to the minimum cumulative distance .

[0094] For the obtained relationship sequence, the above algorithm is used to calculate the similarity between the sequences. In the embodiment of the present application, the difference between the main axes of the most similar entity pair is selected as the image rotation angle .

[0095] Pair the geographic entities in the two images one by one, and calculate the DTW value between the azimuth and distance organization sequences of the geographic entities using the above algorithm. The azimuth sequence DTW value is recorded as ( ), the spatial distance sequence DTW value is recorded as ( ), the two distances are weighted to obtain the weighted DTW value F :

[0096] in, is the weight corresponding to the DTW value of the azimuth sequence; is the weight corresponding to the DTW value of the spatial distance sequence. Weighted DTW value F The smaller the value, the higher the similarity between geographic entities. The similarity can be expressed as .

[0097] The different-perspective geographic entity alignment method provided in the embodiment of the present application introduces a dynamic time warping method commonly used in time series analysis to analyze the similarity of two sequences and perform weighted processing, obtains a similarity weighted value to characterize the similarity between geographic entities, and improves the effectiveness and accuracy of obtaining the similarity of geographic entities.

[0098] In some embodiments, aligning the first geographic entity with the third geographic entity includes: Acquire an azimuth sequence and a spatial distance sequence corresponding to a third geographic entity; the third geographic entity is a geographic entity in the third remote sensing image; Determine the similarity between the first geographical entity and the third geographical entity based on the azimuth sequence and the spatial distance sequence corresponding to the first geographical entity and the azimuth sequence and the spatial distance sequence corresponding to the third geographical entity; An aligned entity pair is determined based on the similarity between the first geographic entity and the third geographic entity.

[0099] Specifically, after the difference between the main axes of the first geographic entity and the second geographic entity with the highest similarity is used as the rotation angle, the second remote sensing image is rotated to obtain the third remote sensing image, the azimuth sequence and spatial distance sequence corresponding to all geographic entities in the third remote sensing image are recalculated, and the similarity between the first geographic entity and the third geographic entity in the third remote sensing image is determined, and the alignment entity pair is determined based on the similarity between the first geographic entity and the third geographic entity.

[0100] For example, after obtaining the weighted DTW values ​​of the first geographic entity and the second geographic entity, the entity with the smallest DTW (highest similarity) in the geographic entity is regarded as an "aligned entity pair" and the similarity is recorded. The difference between the principal axes of the geographic entity with the smallest DTW distance (highest similarity) in the "aligned entity pair" is taken. As the rotation angle of the second image.

[0101] Then, similarity calculation and result output are performed on the entity pairs after the perspective deflection correction, such as Figure 3 As shown, f2 is set according to Rotate to complete the perspective deflection correction, and then recalculate the azimuth sequence and distance sequence of all geographic entities in the rotated f2 (i.e., the third remote sensing image). The calculation method of the distance sequence of the third remote sensing image is the same as the calculation method of the distance sequence of the geographic entities in the first remote sensing image and the second remote sensing image, but the calculation of the azimuth sequence is slightly different. The calculation of the azimuth sequence of the geographic entities in the first remote sensing image and the second remote sensing image is based on the establishment of the coordinate axis based on the principal axis of the central entity. However, after the perspective deflection correction, the calculation of the azimuth sequence of the geographic entities in the third remote sensing image only needs to establish a coordinate system in the horizontal direction.

[0102] For example, rotate the second image around the center of the image After the angle is determined, a coordinate system is established in the horizontal direction to calculate the distance and orientation of the central entity and the surrounding entities. The orientation calculation formula is as follows: ( i =1,2,3,..., num ; j =1,2,3,..., num ) ( i=1,2,3,..., num ; j =1,2,3,..., num )

[0103] in, For remote sensing images geographic entity )’s horizontal and vertical coordinates; num is the total number of geographic entities in the remote sensing image; For remote sensing images geographic entity )’s horizontal and vertical coordinates; Representing geographic entities With geographic entities The difference in the horizontal coordinates of Representing geographic entities With geographic entities The difference in the vertical coordinates. Representing geographic entities With geographic entities The angle between the line and the east direction.

[0104] After the calculation is completed, it is organized into an azimuth sequence and a spatial distance sequence, and then the DTW distance weighted value of all geographic entities in f1 and all geographic entities in f2 is calculated using the dynamic time warping method (i.e., F ), select the entity pair with the smallest DTW distance weighted value (highest similarity) as the aligned entity pair and record it.

[0105] The method for determining the similarity between the first geographic entity and the third geographic entity is the same as the method for determining the similarity between the first geographic entity and the second geographic entity, and will not be described in detail herein.

[0106] In some embodiments, determining the aligned entity pair based on the similarity between the first geographic entity and the third geographic entity includes: Selecting the candidate entity pair with the highest similarity from the candidate entity pairs as the aligned entity pair; the candidate entity pair is obtained by pairing the first geographic entity and the third geographic entity in pairs; Deleting the candidate entity pair including the first geographic entity in the aligned entity pair and the candidate entity pair including the third geographic entity in the aligned entity pair from the candidate entity pairs to obtain updated candidate entity pairs; Iteratively executing the steps of selecting the candidate entity pair with the highest similarity from the updated candidate entity pairs as the aligned entity pair and updating the candidate entity pairs until the updated candidate entity pairs are empty, thereby obtaining all the aligned entity pairs; The first geographic entity or the third geographic entity that is not paired into the aligned entity pair is regarded as a missing entity.

[0107] Specifically, multiple candidate entity pairs are obtained by pairing the first geographic entity and the third geographic entity in pairs, and the candidate entity pair with the highest similarity is selected as the aligned entity pair, and the aligned entity pair is output, and the candidate entity pair is updated as follows: the candidate entity pair containing any geographic entity in the aligned entity pair is deleted, and the remaining candidate entity pairs are used as new candidate entity pairs. Then a new round of iteration is started, that is, the candidate entity pair with the highest similarity is selected from the updated candidate entity pairs as the aligned entity pair of this round, and the candidate entity pairs are updated in the same way as above, and the aligned entity pairs of each round are determined and output iteratively continuously until the updated candidate entity pairs are empty, then the iteration is stopped, and all aligned entity pairs are obtained.

[0108] If there is a first geographic entity or a third geographic entity that is not paired to form an aligned entity pair, the geographic entity that is not paired to form an aligned entity pair is regarded as a missing entity and is output. Figure 3 As shown, after traversing all geographic entities and obtaining aligned entity pairs and missing entities, all aligned entity pairs and unaligned missing entities are output.

[0109] For example, the geographic entities in the first remote sensing image (ie, the first geographic entity) include a, b, and c, and the geographic entities in the third remote sensing image (ie, the third geographic entity) include d, e, and f. Then the first geographic entity is paired with the third geographic entity in pairs to obtain multiple groups of candidate entity pairs, including (a, d), (a, e), (a, f), (b, d), (b, e), (b, f), (c, d), (c, e), (c, f), and the similarities of the two geographic entities in these candidate entity pairs are determined to be F1 to F9, respectively. The candidate entity pair with the highest similarity (such as the highest F5) is selected, that is, (b, e), and (b, e) is output as the aligned entity pair. Then, the candidate entity pairs containing b or e are deleted from the remaining candidate entity pairs, that is, (b, d), (b, f), (a, e), and (c, e) are deleted, and the new candidate entity pairs are: (a, d), (a, f), (c, d), and (c, f), with similarities of F1, F3, F7, and F9, respectively. The candidate entity pair with the highest similarity (such as the highest F1) (a, d) is selected from these four candidate entity pairs, and (a, d) is output as an aligned entity pair, and then the candidate entity pairs containing a or d are deleted from the remaining candidate entity pairs, that is, (a, f) and (c, d) are deleted, and the new candidate entity pair is (c, f). The candidate entity pair with the highest similarity is selected from the new candidate entity pairs, that is, (c, f), as the aligned entity pair and output. After deleting (c, f), the candidate entity pair is empty, and the iteration is stopped to obtain all the aligned entity pairs, that is, (b, e), (a, d), and (c, f).

[0110] For another example, the geographic entities in the first remote sensing image (i.e., the first geographic entity) include a, b, and c, and the geographic entities in the third remote sensing image (i.e., the third geographic entity) include d and e. Pair the first geographic entity with the third geographic entity in pairs to obtain multiple candidate entity pairs, including (a, d), (a, e), (b, d), (b, e), (c, d), and (c, e). Determine the similarities between the two geographic entities in these candidate entity pairs as F1 to F6. Select the candidate entity pair with the highest similarity (e.g., F4 is the highest), that is, (b, e), and output (b, e) as the aligned entity pair. Then, delete the candidate entity pairs containing b or e from the remaining candidate entity pairs, and obtain new candidate entity pairs: (a, d), (c, d), with similarities of F1 and F5, respectively. Select the candidate entity pair (a, d) with the highest similarity (such as the highest F1) from these two candidate entity pairs, and output (a, d) as the aligned entity pair. Then delete the candidate entity pairs containing a or d from the remaining candidate entity pairs. At this time, (c, d) is deleted, the candidate entity pairs are cleared, and the iteration is stopped to obtain all the aligned entity pairs, namely (b, e) and (a, d). At this time, the first geographic entity c is left unmatched into an aligned entity pair. c is output as a missing entity.

[0111] The different-viewing angle geographic entity alignment method provided in the embodiment of the present application selects one or more most similar entity pairs after completing the first similarity calculation (similarity calculation between the first geographic entity and the second geographic entity), and takes the average of their main axis angle differences as the rotation angle of the second image relative to the first image. The advantage of this is that for those geographic entities whose main axis features are not obvious (such as circles, squares, equilateral triangles, etc.), or geographic entities whose shapes change due to image distortion (the change in their main axis direction may not be caused by image rotation). The difference in main axis direction between entity pairs cannot accurately reflect the rotation relationship between the two images; and the most similar entity pairs selected are most likely consistent with the rotation direction between the images.

[0112] Due to the deformation of geographic entities, the entity pairs with deviations in the calculation of the principal axis are inconsistent in the order of entities obtained by scanning and traversing the same surrounding entities. There are large differences in the constructed orientation sequence and distance sequence, which will lead to errors in similarity calculation. Therefore, after the second image is rotated relative to the first image, the orientation sequence of the geographic entity is recalculated with the horizontal line of the two images as the horizontal axis, and the similarity between the geographic entity pairs (the first geographic entity and the third geographic entity) is recalculated. The accuracy of similarity calculation is further improved, thereby improving the accuracy of geographic entity alignment.

[0113] In addition, the different-viewing angle geographic entity alignment method provided in the embodiment of the present application reduces the time and effort consumed by the current manual method for aligning remote sensing image entities, and only two shapefiles need to be input to complete the rapid geographic entity alignment. By combining the principal axis calculation with the sequence comparison method in time series analysis, the image can be corrected, and the problem of only constructing a sequence for geographic entity alignment causing the constructed sequence to be disordered is solved. It is suitable for geographic entity alignment in the case of large rotation amplitude and geometric distortion between source data.

[0114] Figure 4 is a schematic diagram of the structure of the different-viewing angle geographic entity alignment device provided by the present invention, such as Figure 4 As shown, the present invention provides a different-viewing angle geographic entity alignment device, including a first determination module 401, a first acquisition module 402, a second determination module 403, a second acquisition module 404 and an alignment module 405.

[0115] The first determination module 401 is used to determine, for each geographical entity in a remote sensing image, any geographical entity as a central entity and all geographical entities around the central entity as surrounding entities.

[0116] The first acquisition module 402 is used to calculate the azimuth between the central entity and each surrounding entity with the main axis of the central entity as the coordinate axis to obtain an azimuth sequence corresponding to the central entity, and calculate the spatial distance between the central entity and each surrounding entity to obtain a spatial distance sequence corresponding to the central entity.

[0117] The second determination module 403 is used to determine the similarity between the first geographic entity and the second geographic entity based on the azimuth sequence and spatial distance sequence corresponding to the first geographic entity and the second geographic entity respectively; the first geographic entity is a geographic entity in the first remote sensing image; the second geographic entity is a geographic entity in the second remote sensing image.

[0118] The second acquisition module 404 is used to use the difference between the main axes of the first geographic entity and the second geographic entity with the highest similarity as a rotation angle, and rotate the second remote sensing image to obtain a third remote sensing image.

[0119] The alignment module 405 is used to align the first geographic entity with a third geographic entity; the third geographic entity is a geographic entity in the third remote sensing image.

[0120] Specifically, the above-mentioned different-perspective geographic entity alignment device provided by the present invention can implement all the method steps implemented by the above-mentioned different-perspective geographic entity alignment method embodiment, and can achieve the same technical effect. The parts and beneficial effects that are the same as the method embodiment in this embodiment will not be described in detail here.

[0121] It should be noted that the division of units / modules in the above-mentioned embodiments of the present invention is schematic and is only a logical function division. There may be other division methods in actual implementation. In addition, the functional units in the various embodiments of the present application may be integrated into a processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated units may be implemented in the form of hardware or in the form of software functional units.

[0122] Figure 5 is a schematic diagram of the structure of the electronic device provided by the present invention, such as Figure 5 As shown, the electronic device may include: a processor 501, a communications interface 502, a memory 503 and a communication bus 504, wherein the processor 501, the communications interface 502 and the memory 503 communicate with each other via the communication bus 504. The processor 501 may call the logic instructions in the memory 503 to execute the different-viewing angle geographic entity alignment method, which includes: For each geographical entity in a remote sensing image, any geographical entity is regarded as a central entity, and all geographical entities around the central entity are regarded as surrounding entities; Taking the main axis of the central entity as the coordinate axis, calculating the azimuth between the central entity and each surrounding entity to obtain an azimuth sequence corresponding to the central entity, and calculating the spatial distance between the central entity and each surrounding entity to obtain a spatial distance sequence corresponding to the central entity; The similarity between the first geographic entity and the second geographic entity is determined based on the azimuth sequence and spatial distance sequence respectively corresponding to the first geographic entity and the second geographic entity; the first geographic entity is a geographic entity in a first remote sensing image; the second geographic entity is a geographic entity in a second remote sensing image; the difference between the main axes of the first geographic entity and the second geographic entity with the highest similarity is used as a rotation angle, and the second remote sensing image is rotated to obtain a third remote sensing image; the first geographic entity is aligned with the third geographic entity; the third geographic entity is a geographic entity in the third remote sensing image.

[0123] Specifically, the processor 501 may be a central processing unit (CPU), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or a complex programmable logic device (CPLD), and the processor may also adopt a multi-core architecture.

[0124] The logic instructions in the memory 503 can be implemented in the form of software functional units and can be stored in a processor-readable storage medium when sold or used as an independent product. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) or a processor to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.

[0125] In some embodiments, a computer program product is further provided, the computer program product comprising a computer program, the computer program may be stored on a non-transitory computer-readable storage medium, and when the computer program is executed by a processor, the computer can perform the different-perspective geographic entity alignment method provided by the above-mentioned method embodiments, the method comprising: For each geographical entity in a remote sensing image, any geographical entity is regarded as a central entity, and all geographical entities around the central entity are regarded as surrounding entities; Taking the main axis of the central entity as the coordinate axis, calculating the azimuth between the central entity and each surrounding entity to obtain an azimuth sequence corresponding to the central entity, and calculating the spatial distance between the central entity and each surrounding entity to obtain a spatial distance sequence corresponding to the central entity; The similarity between the first geographic entity and the second geographic entity is determined based on the azimuth sequence and spatial distance sequence respectively corresponding to the first geographic entity and the second geographic entity; the first geographic entity is a geographic entity in a first remote sensing image; the second geographic entity is a geographic entity in a second remote sensing image; the difference between the main axes of the first geographic entity and the second geographic entity with the highest similarity is used as a rotation angle, and the second remote sensing image is rotated to obtain a third remote sensing image; the first geographic entity is aligned with the third geographic entity; the third geographic entity is a geographic entity in the third remote sensing image.

[0126] Specifically, the above-mentioned computer program product provided in the embodiment of the present application can implement all the method steps implemented by the above-mentioned method embodiments, and can achieve the same technical effect. The parts and beneficial effects of this embodiment that are the same as the method embodiment will not be described in detail here.

[0127] In some embodiments, a computer-readable storage medium is further provided, wherein the computer-readable storage medium stores a computer program, wherein the computer program is used to enable a computer to execute the different-perspective geographic entity alignment method provided by the above-mentioned method embodiments, and the method comprises: For each geographical entity in a remote sensing image, any geographical entity is regarded as a central entity, and all geographical entities around the central entity are regarded as surrounding entities; Taking the main axis of the central entity as the coordinate axis, calculating the azimuth between the central entity and each surrounding entity to obtain an azimuth sequence corresponding to the central entity, and calculating the spatial distance between the central entity and each surrounding entity to obtain a spatial distance sequence corresponding to the central entity; The similarity between the first geographic entity and the second geographic entity is determined based on the azimuth sequence and spatial distance sequence respectively corresponding to the first geographic entity and the second geographic entity; the first geographic entity is a geographic entity in a first remote sensing image; the second geographic entity is a geographic entity in a second remote sensing image; the difference between the main axes of the first geographic entity and the second geographic entity with the highest similarity is used as a rotation angle, and the second remote sensing image is rotated to obtain a third remote sensing image; the first geographic entity is aligned with the third geographic entity; the third geographic entity is a geographic entity in the third remote sensing image.

[0128] Specifically, the above-mentioned computer-readable storage medium provided by the present invention can implement all the method steps implemented by the above-mentioned method embodiments, and can achieve the same technical effects. The parts and beneficial effects that are the same as the method embodiments in this embodiment will not be described in detail here.

[0129] It should be noted that the computer-readable storage medium can be any available medium or data storage device that can be accessed by the processor, including but not limited to magnetic storage (such as floppy disks, hard disks, magnetic tapes, magneto-optical disks (MO), etc.), optical storage (such as CD, DVD, BD, HVD, etc.), and semiconductor storage (such as ROM, EPROM, EEPROM, non-volatile memory (NAND FLASH), solid-state drive (SSD)), etc.

[0130] It should also be noted that the terms "first", "second", etc. in the present invention are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the terms used in this way can be interchangeable under appropriate circumstances, so that the embodiments of the present invention can be implemented in an order other than those illustrated or described herein, and the objects distinguished by "first" and "second" are usually of the same type, and the number of objects is not limited. For example, the first object can be one or more.

[0131] In the present invention, the term "plurality" refers to two or more than two, and other quantifiers are similar to them.

[0132] In the present invention, "determine B based on A" means that the factor A should be considered when determining B. It is not limited to "B can be determined based on A alone", but should also include: "determine B based on A and C", "determine B based on A, C and E", "determine C based on A, and further determine B based on C", etc. In addition, it can also include taking A as a condition for determining B, for example, "when A meets the first condition, use the first method to determine B"; for another example, "when A meets the second condition, determine B", etc.; for another example, "when A meets the third condition, determine B based on the first parameter", etc. Of course, it can also be a condition that takes A as a factor for determining B, for example, "when A meets the first condition, use the first method to determine C, and further determine B based on C", etc.

[0133] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage and optical storage, etc.) containing computer-usable program codes.

[0134] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer executable instructions. These computer executable instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0135] These processor executable instructions may also be stored in a processor readable memory that can direct a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the processor readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0136] These processor-executable instructions may also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for executing on the computer or other programmable device to implement the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0137] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for aligning geographic entities from different perspectives, characterized in that: include: For each geographical entity in a remote sensing image, any geographical entity is regarded as a central entity, and all geographical entities around the central entity are regarded as surrounding entities; Taking the main axis of the central entity as the coordinate axis, calculating the azimuth between the central entity and each surrounding entity to obtain an azimuth sequence corresponding to the central entity, and calculating the spatial distance between the central entity and each surrounding entity to obtain a spatial distance sequence corresponding to the central entity; Determine the similarity between the first geographical entity and the second geographical entity based on the azimuth sequence and the spatial distance sequence respectively corresponding to the first geographical entity and the second geographical entity; The first geographical entity is a geographical entity in a first remote sensing image; the second geographical entity is a geographical entity in a second remote sensing image; The second remote sensing image is rotated by taking the difference between the main axes of the first geographic entity and the second geographic entity with the highest similarity as a rotation angle to obtain a third remote sensing image; aligning the first geographic entity with a third geographic entity; The third geographic entity is a geographic entity in the third remote sensing image.

2. The different-viewing angle geographic entity alignment method according to claim 1, characterized in that: The determining the similarity between the first geographical entity and the second geographical entity based on the azimuth sequence and the spatial distance sequence respectively corresponding to the first geographical entity and the second geographical entity comprises: Pairing the first geographic entity with the second geographic entity in pairs to obtain multiple groups of entity pairs; Obtaining a first similarity and a second similarity corresponding to each entity pair; the first similarity is the similarity between an azimuth sequence corresponding to a first geographic entity in the entity pair and an azimuth sequence corresponding to a second geographic entity in the entity pair; the second similarity is the similarity between a spatial distance sequence corresponding to the first geographic entity in the entity pair and a spatial distance sequence corresponding to the second geographic entity in the entity pair; The similarity between the first geographic entity and the second geographic entity in each entity pair is determined based on the first similarity and the second similarity corresponding to the entity pair.

3. The different-viewing angle geographic entity alignment method according to claim 2, characterized in that: The first similarity is characterized by a first dynamic time warping distance, which is a dynamic time warping distance between an azimuth sequence corresponding to a first geographic entity in the entity pair and an azimuth sequence corresponding to a second geographic entity in the entity pair; The second similarity is characterized by a second dynamic time warping distance, where the second dynamic time warping distance is a dynamic time warping distance between a spatial distance sequence corresponding to a first geographic entity in the entity pair and a spatial distance sequence corresponding to a second geographic entity in the entity pair; The determining the similarity between the first geographic entity and the second geographic entity in the entity pair based on the first similarity and the second similarity corresponding to each entity pair includes: Performing a weighted summation on the first dynamic time warping distance and the second dynamic time warping distance to obtain a weighted value corresponding to each entity pair; The weighted value is used to represent the similarity between the first geographic entity and the second geographic entity in the entity pair.

4. The different-viewing angle geographic entity alignment method according to claim 1, characterized in that: The aligning the first geographic entity with the third geographic entity comprises: Obtaining an azimuth sequence and a spatial distance sequence corresponding to a third geographic entity; Determine the similarity between the first geographical entity and the third geographical entity based on the azimuth sequence and the spatial distance sequence corresponding to the first geographical entity and the azimuth sequence and the spatial distance sequence corresponding to the third geographical entity; An aligned entity pair is determined based on the similarity between the first geographic entity and the third geographic entity.

5. The different-viewing angle geographic entity alignment method according to claim 4, characterized in that: The determining the aligned entity pair based on the similarity between the first geographic entity and the third geographic entity comprises: Selecting the candidate entity pair with the highest similarity from the candidate entity pairs as the aligned entity pair; the candidate entity pair is obtained by pairing the first geographic entity and the third geographic entity in pairs; Deleting the candidate entity pair including the first geographic entity in the aligned entity pair and the candidate entity pair including the third geographic entity in the aligned entity pair from the candidate entity pairs to obtain updated candidate entity pairs; Iteratively executing the steps of selecting the candidate entity pair with the highest similarity from the updated candidate entity pairs as the aligned entity pair and updating the candidate entity pairs until the updated candidate entity pairs are empty, thereby obtaining all the aligned entity pairs; The first geographic entity or the third geographic entity that is not paired into the aligned entity pair is regarded as a missing entity.

6. The different-viewing angle geographic entity alignment method according to claim 1, characterized in that: The principal axis of the central entity is obtained by the following steps: Obtaining a centralized coordinate point through the boundary coordinates of the central entity; Calculating the covariance matrix of the central coordinate point, and performing eigendecomposition on the covariance matrix to obtain eigenvalues ​​and eigenvectors; The eigenvector corresponding to the largest eigenvalue is selected as the principal axis of the central entity.

7. The different-viewing angle geographic entity alignment method according to claim 1, characterized in that: The method further comprises: extracting geographic entities from the first remote sensing image and the second remote sensing image; The first remote sensing image and the second remote sensing image are obtained by photographing the same type of geographical entities in the same area under different shooting conditions.

8. A device for aligning geographical entities from different viewing angles, characterized in that: include: The first determination module is used to determine, for each geographical entity in the remote sensing image, any geographical entity as a central entity and all geographical entities around the central entity as surrounding entities; A first acquisition module is used to calculate the azimuth between the central entity and each surrounding entity with the main axis of the central entity as the coordinate axis to obtain an azimuth sequence corresponding to the central entity, and calculate the spatial distance between the central entity and each surrounding entity to obtain a spatial distance sequence corresponding to the central entity; A second determination module, configured to determine the similarity between the first geographic entity and the second geographic entity based on the azimuth sequence and the spatial distance sequence respectively corresponding to the first geographic entity and the second geographic entity; The first geographical entity is a geographical entity in a first remote sensing image; the second geographical entity is a geographical entity in a second remote sensing image; A second acquisition module is used to use the difference between the main axes of the first geographic entity and the second geographic entity with the highest similarity as a rotation angle, and rotate the second remote sensing image to obtain a third remote sensing image; an alignment module, configured to align the first geographic entity with a third geographic entity; The third geographic entity is a geographic entity in the third remote sensing image.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the different-perspective geographic entity alignment method as described in any one of claims 1 to 7 is implemented.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the different-perspective geographic entity alignment method as described in any one of claims 1 to 7 is implemented.

Images