WO2024125434A1

WO2024125434A1 - Regional-consistency-based building principal angle correction method

Info

Publication number: WO2024125434A1
Application number: PCT/CN2023/137754
Authority: WO
Inventors: 刘杰; 陈洋涛; 董铱斐; 邹圣兵
Original assignee: 北京数慧时空信息技术有限公司
Priority date: 2022-12-15
Filing date: 2023-12-11
Publication date: 2024-06-20
Also published as: CN115620169A; CN115620169B

Abstract

The present invention relates to the field of remote sensing image processing. Disclosed is a regional-consistency-based building principal angle correction method. The method comprises acquiring a regional principal angle and two candidate principal angles of a building structure, and selecting a real principal angle from the candidate principal angles by means of the relationships between the regional principal angle and the candidate principal angles, so as to correct the principal angle of the building. The present method as a post-processing method for building extraction corrects the principal angle of an irregular building structure according to the alignment consistency of building structures in the range of a region where the irregular building structure is located, thus improving the accuracy of principal angle calculation in post-processing, and thereby greatly improving the post-processing optimization effect of building structure extraction.

Description

Building main angle correction method based on regional consistency

Technical Field

The invention relates to the field of remote sensing image processing, and in particular to a method for correcting a main angle of a building based on regional consistency.

Background technique

At present, with the advent of the era of smart city big data and the continuous development of satellite remote sensing technology, earth observation, satellite remote sensing, ecological assessment, and land supervision are gradually developing in the direction of macro, dynamic, and refined. Therefore, there is a practical need for large-scale and efficient satellite remote sensing image interpretation. Building extraction in high-resolution remote sensing images is of great significance for applications such as illegal construction monitoring, automatic extraction of urban areas, map updates, urban change monitoring, urban planning, three-dimensional modeling, and digital city establishment. While high-resolution remote sensing images improve the spectral characteristics of objects and highlight information such as the structure, texture, and details of objects, they also cause object occlusion due to satellite observation angle problems. The increase in scale brings serious problems of heterogeneous objects with the same spectrum, and increases the noise of the image, thus limiting the accuracy of building extraction. Visual interpretation methods are still the most commonly used interpretation method. Although the accuracy of visual interpretation methods is relatively guaranteed, their low efficiency and time-consuming and labor-intensive shortcomings seriously restrict the large-scale application of high-resolution remote sensing images, resulting in a huge waste of image data.

Deep learning is a new stage in the development of machine learning in artificial intelligence, which effectively solves the problems of characterizing the features of complex objects and analyzing the association of complex scenes. The deep learning method for extracting buildings from high-resolution remote sensing images can automatically extract the characteristic information of buildings and achieve high-precision and high-efficiency building extraction. However, due to the complexity of remote sensing images, buildings are affected by noise, occlusion, shadows, and low contrast. The current automatic building extraction method has low reliability and obvious defects. For example, the spectral characteristics of buildings and roads are similar and cannot be effectively distinguished, and it is impossible to directly obtain high-precision extraction results through fully automatic methods. Another existing method is to combine computer automatic extraction technology with manual interaction, that is, interactive object extraction, and the extraction accuracy is guaranteed to a certain extent. However, this method requires the initial position of the building to be given manually, relies on edge information to extract buildings, has complex interactions, requires high accuracy for the manually given position, and is difficult to implement.

In order to further improve the accuracy of building automatic extraction and apply the extraction results to engineering practice, researchers have carried out various studies in the post-processing direction of building extraction. The extracted results are graphically modified and the optimized results can be directly used in specific engineering projects.

In the post-processing method, the main angle of the building structure obtained by calculation can help improve the accuracy of the building extraction by the deep learning method. However, the existing main angle calculation method still has two difficult problems to overcome. First, due to the diverse shapes of the buildings themselves, coupled with the errors caused by automated extraction, it is difficult to obtain the main angle of the building; second, different main angle calculation methods have certain scopes of application and limitations, especially for buildings with poor extraction effects, the results obtained by different calculation methods are quite different, making it difficult to achieve good post-processing results.

Summary of the invention

The present invention proposes a method for correcting the main angle of a building based on regional consistency. The method corrects the main angle of the irregular building according to the consistency of the orientation of the buildings in the area where the irregular building is located, thereby improving the accuracy of the main angle calculation in post-processing, thereby greatly improving the post-processing optimization effect of building extraction.

To achieve the above technical objectives, the technical solution of the present invention is as follows:

A method for correcting a main angle of a building based on regional consistency, the method comprising the following steps:

S1 obtains a target remote sensing image, and obtains candidate main angles of buildings and structures in the target remote sensing image by a preset method, wherein the candidate main angles include a first candidate value and a second candidate value;

S2 calculates the difference between the first candidate value and the second candidate value, and according to the relationship between the difference and the first preset threshold, divides the building structure into a standard building structure and a building structure to be corrected, and simultaneously obtains the true main angle of the standard building structure;

S3 obtains the regional main angle of the building structure to be corrected through a regional statistical strategy, and the regional statistical strategy includes external expansion main angle statistics and line segment histogram statistics;

S4 calculates the difference between the candidate main angle and the regional main angle, and selects the real main angle of the building structure to be corrected from the candidate main angles according to the relationship between the difference and the second preset threshold.

Further, step S1 comprises:

S11 obtains the angles of the two major axis center lines of the circumscribed rectangle of the minimum area of the building structure as the first candidate value of the building structure;

S12 obtains the vector contour line segment of the building structure by a vector contour extraction method, and obtains the angle of the vector contour line segment;

S13 groups vector contour line segments with similar angles into one group, selects a group of vector contour line segments with the longest total length as a reference vector contour line segment group, and obtains a second candidate value of the building structure by a vertical equivalent method based on the angle of the reference vector contour line segment;

The vector contour segments of similar angles include those with angle differences between [0°, 2°]∪[88°, 90°] Vector contour segments of .

Furthermore, the vector contour extraction method comprises:

By performing semantic segmentation, binarization and mathematical morphology processing on remote sensing images, post-processing images are obtained;

By performing boundary tracing and vector extraction on the post-processed image, the vector contour segments of the building structure are obtained.

Furthermore, the vertical equivalent method includes:

Clustering the reference vector contour line segment group based on the angle of the reference vector contour line segment to obtain a main vector contour line segment group and an auxiliary vector contour line segment group, wherein the total length of the main vector contour line segment group is greater than the total length of the auxiliary vector contour line segment group, and averaging the angles of the main vector contour line segment group and the auxiliary vector contour line segment group to obtain a main vector angle and an auxiliary vector angle;

The auxiliary vector angle is vertically transformed toward the main vector angle to obtain a vertical auxiliary vector angle, and the average of the main vector angle and the vertical auxiliary vector angle is used as the second candidate value.

Further, step S2 includes:

If the difference between the first candidate value and the second candidate value is less than a first preset threshold, the building structure is used as a standard building structure, and the second candidate value is used as the true main angle of the standard building structure;

If the difference between the first candidate value and the second candidate value is greater than the first preset threshold, the building structure is used as the building structure to be corrected.

Furthermore, the regional statistics strategy is outward expansion main angle statistics, and step S3 includes:

The minimum circumscribed rectangle of the building structure to be modified is expanded by a preset distance to obtain an expanded rectangle;

The standard housing structures within the outer expansion rectangle and intersecting with the outer expansion rectangle are used as reference housing structures;

The average of the real main angles of the reference building structures is used as the regional main angle of the building structure to be corrected.

Furthermore, the regional statistical strategy is line segment histogram statistics, and step S3 includes:

The houses and structures in the target remote sensing image are partitioned by hierarchical clustering method based on distance matrix;

Obtaining the angle of the vector contour line segment of the building structure in each partition, wherein the building structure is composed of a plurality of vector contour line segments;

The angle histogram of each partition is constructed by using the angle of the vector contour line segment of the building structure in each partition;

The highest frequency pair in the angle histogram of the partition where the building to be corrected is located The corresponding angle is used as the main angle of the area of the building structure to be modified.

Further, step S4 includes:

Calculate the difference between the first candidate value and the main angle of the region, and the difference between the second candidate value and the main angle of the region, and select the smaller difference as the target difference;

If the target difference is less than a second preset threshold, a candidate main angle close to the regional main angle is selected as the true main angle of the building structure to be corrected;

If the target difference is greater than a second preset threshold, the second candidate value is selected as the true main angle of the building structure to be corrected.

The beneficial effects of the present invention are as follows: the present invention proposes a new method for correcting the main angle of a building based on regional consistency. The regional main angle of the building structure to be corrected is calculated by a regional statistical strategy, and the optimal value of the candidate main angle is selected based on the regional main angle to obtain the true main angle. As a post-processing method for building extraction, the present invention solves the problem of difficulty in calculating the main angle of irregular buildings, thereby greatly improving the post-processing optimization effect of building extraction in remote sensing images.

BRIEF DESCRIPTION OF THE DRAWINGS

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only 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.

FIG1 is a flow chart of a method for correcting a main angle of a building based on regional consistency according to the present invention;

FIG2 is a schematic diagram of the original distribution of buildings provided in Example 1 of the present invention;

FIG3 is a schematic diagram of a correction result of a main angle of a building in Embodiment 1 of the present invention;

FIG. 4 is a schematic diagram of the original distribution of buildings provided in the third embodiment of the present invention.

Detailed ways

The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field belong to the scope of protection of the present invention.

Please refer to FIG. 1 , which is a flow chart of an embodiment of a method for correcting a main angle of a building based on regional consistency according to the present invention. The method comprises the following steps:

S1 obtains a candidate main angle of a building structure by a preset method, wherein the candidate main angle includes a first candidate value and a second candidate value;

S2 classifies the housing structures into standard housing structures and housing structures to be corrected according to the difference between the first candidate value and the second candidate value;

S3: obtaining the real main angle of the standard building structure within the preset range, and calculating the regional main angle of the building structure to be corrected according to the real main angle of the standard building structure within the preset range;

S4 selects the value of the main angle of the region closest to the building structure to be corrected from the candidate main angles of the building structure to be corrected, and obtains the real main angle of the building structure to be corrected.

The technical idea of the present invention is as follows: 1. The current building extraction method based on remote sensing images can only extract relatively regular buildings with obvious features, and its versatility is poor. When the buildings are dense or irregular buildings are encountered, the extraction effect is often not ideal, and appropriate post-processing steps are required to optimize the accuracy of the extracted buildings. Among them, obtaining the main angle of the building is of great help to post-processing optimization. The method of obtaining the main angle of the building usually includes: 1) obtaining the minimum circumscribed rectangle of the building, and the direction of the two major axis center lines of the minimum circumscribed rectangle is used as the first candidate value of the main angle of the building; 2) obtaining the contour line segment of the building, calculating the total length of all similar angle contour line segments, and the direction with the longest total length is used as the second candidate value of the main angle of the building. For irregular buildings, different methods for calculating the main angle may also obtain different results. How to further determine the most realistic main angle is emphasized in the subsequent steps of the present invention; 2. The present invention introduces the concept of regional main angle to help determine the real main angle of the building. The regional main angle is the average value of the main angle of a group of buildings with similar main angles within a certain area where the building is located. The regional main angle can reflect the regional consistency of the building area and provide guidance for the main angle of the building to be corrected that is consistent with the regional consistency; 3. Select the candidate main angle of the building to be corrected that is closest to the regional main angle as the true main angle. This method is guided by the regional consistency of the building area, can determine the true main angle of the irregular building to be corrected, and further improve the effect of building extraction post-processing, thereby achieving high-precision building extraction.

Embodiment 1

This embodiment provides a method for obtaining the regional main angle of the building structure to be corrected based on the statistics of the external expansion main angle. At the same time, this embodiment provides a method for obtaining the real main angle of the building structure to be corrected when the target difference is less than the second preset threshold:

S13 groups vector contour segments with similar angles into a group and selects the segment with the longest total length. A set of vector contour line segments is used as a reference vector contour line segment set, and based on the angles of the reference vector contour line segments, a second candidate value of the building structure is obtained by a vertical equivalent method;

The vector contour line segments of similar angles are vector contour line segments whose angle difference is [0°, 2°]∪[88°, 90°].

To provide additional explanation for step S12, the vector contour extraction method includes:

Perform semantic segmentation on the target remote sensing image through deep learning methods to obtain segmentation data;

This embodiment uses mask-RCNN to perform semantic segmentation on the target remote sensing image. Mask-RCNN follows the idea of Faster RCNN, and uses the architecture of ResNet-FPN for feature extraction. In addition, a Mask prediction branch is added, which can complete high-quality semantic segmentation tasks. It has a good extraction effect on ordinary buildings, and the effect is significantly reduced only when facing complex irregular buildings. This embodiment uses ArcGIS to make samples by manual sample outlining. During the training process, pixel blocks cropped to 512×512 are used as samples, and about 7,000 instances are used as training data sets and about 3,000 instances are used as verification data sets. Computer configuration: CPU: Xeon(R)CPU E5-2620 v4@2.10GHz×32; Graphics card: Quadro M4000; Memory: 128G; Operating system: Ubuntu 16.04. The training process took more than 40 hours and 90,000 iterations.

The segmented data is binarized, and 0.5 is used as the binarization threshold of the building target probability map to obtain a segmented binary image;

The segmentation binary image of the building structure is used to highlight the outline of the target building and compress the overall image data volume;

The segmented binary image is post-processed by mathematical morphology method to obtain the post-processed image:

Among them, the mathematical morphology method consists of a set of morphological algebraic operators, and the basic operations include: dilation, erosion, opening and closing;

The segmented binary image is denoised by mathematical morphology methods to remove useless information such as classification noise and small non-building structures. Image shape and structure analysis and processing can be performed based on mathematical morphology methods, including image segmentation, feature extraction, edge detection, image filtering, image enhancement and restoration.

The boundary of the post-processed image is traced by Moore's field boundary tracing method to generate the contour line segments of the building structure;

Based on the contour line segments of the building structures, vector extraction is performed on the post-processed image to obtain the vector contour line segments of the building structures.

To provide additional explanation for step S12, the vertical equivalent method includes:

Clustering the reference vector contour line segment group based on the angle of the reference vector contour line segment to obtain a main vector contour line segment group and an auxiliary vector contour line segment group;

The total length of the line segments of the main vector contour line segment group is greater than that of the auxiliary vector contour line segment group. The total length of the line segments of the segment group;

Calculate the angle mean of the main vector contour line segment group and the auxiliary vector contour line segment group respectively to obtain the main vector angle and the auxiliary vector angle;

The angle difference between the main vector contour line segment group and the auxiliary vector contour line segment group is [0°, 2°], and the angle difference between the groups is [88°, 90°], so the main vector angle and the auxiliary vector angle are close to a vertical relationship;

Transform the auxiliary vector angle vertically toward the direction of the main vector angle to obtain the vertical auxiliary vector angle;

The vertical transformation is expressed as follows:

Among them, xi is the main vector angle, yi is the auxiliary vector angle, and f(xi) is the vertical auxiliary vector angle;

After the vertical change operation, the angle difference between the vertical auxiliary vector angle and the main vector angle is [0°, 2°];

The average of the main vector angle and the vertical auxiliary vector angle is used as the second candidate value b2.

S2 calculates the difference between the first candidate value and the second candidate value, and divides the building structures into standard building structures and building structures to be corrected according to the relationship between the difference and the first preset threshold.

To provide additional explanation for step S2, in this specific embodiment, the preset threshold is set to 3°. If the difference between the first candidate value and the second candidate value is less than 3°, it is considered that the main direction of the building structure does not need to be corrected, and the building structure is taken as a standard building structure, and the second candidate value is taken as the true main angle of the standard building structure.

S3 obtains the real main angle of the standard building structure within the preset range, and obtains the regional main angle of the building structure to be corrected according to the shape information of the standard building structure within the preset range.

In the first embodiment, the regional statistical strategy is external expansion main angle statistics, and step S3 includes:

Obtaining the real main angles a ₂ , a ₃ , a ₄ , a ₅ , a ₆ , and a ₇ of the reference building structure;

Please refer to Figure 2, A is the building to be corrected, and other houses within a certain distance d are reference buildings. The houses involved in the main angle calculation of the area include Includes all houses within a distance d from A and those intersecting the range line.

Based on the real main angle, the real main angle of the reference building structure is clustered by the nearest neighbor clustering method:

(1) Select any one of the true main angles of the above six reference building structures as the first cluster center, such as z ₁ = a ₂ , where z ₁ is the cluster center of reference group A ₁ ;

(2) The current cluster center is z ₁ , and the Euclidean distance from a ₃ to the cluster center z ₁ is calculated as dist(a ₃ ,z ₁ );

If dist(a3,z1)∈[0,3] or ∪[87,93], then a ₂ ∈A ₁ ;

(3) The current cluster center is z ₁ , and the Euclidean distance from a ₅ to the cluster center z ₁ is calculated as dist(a ₄ ,z ₁ );

If dist(a ₄ ,z ₁ )∈(3,87), then a ₄ is defined as the second cluster center z ₂ , z ₂ =a ₄ , where z ₂ is the cluster center of the reference group A ₂ ;

(1) This process is deduced in this way until the clustering of the true main angles of the above eight reference building structures is completed. The clustering results are A1 = {a ₂ , a ₃ , a ₅ , a ₆ , a ₇ }}, A ₂ = {a ₄ }.

The nearest neighbor clustering method is classified according to the Euclidean distance of the true main angle, and the Euclidean distance is expressed as:

Wherein, dist(a _i ,a _j ) is the Euclidean distance from the true main angle a _i to the true main angle a _j ;

A reference group with the largest number of reference buildings and structures is selected, and the mean of the true main angles of the reference buildings and structures in the group is calculated to obtain the regional main angle of the building and structure to be corrected;

The reference group with the largest number of reference buildings and structures is _A1 . The mean of the main angles of the buildings and structures in this group is calculated to obtain the main angle of the area of the buildings and structures to be corrected.

In the first embodiment, the target difference is less than the second preset threshold, and step S4 includes:

The target difference is smaller than the second preset threshold, and the candidate main angle corresponding to the target difference is selected as the true main angle of the building structure to be corrected.

The acquisition of the main angle of the area in this embodiment is shown in Figure 2. In this embodiment, the target difference is less than the second preset threshold, which is visually manifested as the building structure to be corrected. The candidate main angle of the object is consistent with its regional main angle, that is, it is possible to obtain a better main angle correction result by using the regional main angle through the method of the present invention.

Then, the value of the regional main angle closest to the building structure to be corrected is selected from the two candidate main angles of the building structure to be corrected to obtain the real main angle of the building structure to be corrected.

The result of the main angle correction in the embodiment is shown in FIG. 3 .

Embodiment 2

This embodiment provides a method for obtaining the main angle of the area of the building structure to be corrected based on line segment histogram statistics.

In the second embodiment, the regional statistical strategy is segment histogram statistics, and step S3 includes:

The hierarchical clustering method based on the distance matrix is used to partition the standard housing structures within the preset range:

The values of the elements in the distance matrix are obtained by weighting the spatial distance matrix and the angular distance matrix between buildings. The formula is as follows:
d _i,j =W ₁ Δρ _i,j +w ₂ Δθ _i,j

Among them, d _i,j is the element value corresponding to the i-th building and the j-th building in the distance matrix, Δρ _i,j is the spatial distance between the i-th building and the j-th building, Δθ _i,j is the angular distance between the i-th building and the j-th building, w ₁ is the weight value of the spatial distance, w ₂ is the weight value of the angular distance, and w ₁ +w ₂ =1.

Among them, the shortest distance between the corner points of the boundary contours of the two buildings is taken as the spatial distance between the two buildings. The formula is as follows:

Where x _p,i , y _p,i represent the X and Y coordinates of the pth corner point in the i-th building contour curve; x _q,j , y _q,j represent the X and Y coordinates of the qth corner point in the j-th building contour curve; Ni represents the number of line segments of the i-th building contour curve, and N _j represents the number of line segments of the j-th building contour curve;

Among them, the difference between the mean angles of the boundary contours of the two buildings is taken as the angular distance between the two buildings, and the formula is as follows:

Wherein, θk _,i represents the angle of the kth line segment in the outer contour curve of the i-th building; θl _,j represents the angle of the lth line segment in the outer contour curve of the j-th building; S _i represents the number of line segments of the i-th building, and S _j represents the number of line segments of the contour curve of the j-th building.

Get the angle of the vector contour line segment of the reference house structure. House structures are composed of multiple vector contour segments;

The angle histograms of each partition are constructed using the angles of the vector contour segments of the reference buildings and structures in each partition. The angle corresponding to the highest frequency in the histogram of each partition is taken as the regional main angle of the corresponding building.

Embodiment three:

This embodiment provides a method for obtaining the true main angle of the building structure to be corrected when the target difference is greater than the second preset threshold:

Please refer to FIG. 4 , in the third embodiment, the target difference is greater than the second preset threshold:

S3 obtains the regional main angle of the building structure to be corrected by expanding the main angle statistics.

A is the house to be corrected, and the real main angles of other houses within a certain distance d are a ₁ , a ₂ , …, a ₈ . The houses participating in the regional main angle calculation include all houses within the distance d of A and those intersecting with the range line. The main angles of the houses and structures within the range of the house and structure to be corrected are counted, and the houses and structures with similar main angles are grouped together to obtain 4 groups of results: A ₁ = {a ₁ , a ₂ , a ₃ , a ₄ , a ₇ }, A ₂ = {a ₅ }, A ₃ = {a ₆ }, A ₄ = {a ₈ }, and the group of houses and structures with the largest number is selected as A1.

Calculate the average of the main angles of the group of buildings and structures to obtain the main angle of the area of the building and structure to be corrected

The target difference is greater than a second preset threshold value, and the second candidate value is selected as the true main angle of the building structure to be corrected.

As can be seen from Figure 4, the main angle candidate values _b1 and _b2 of house A are both significantly different from the regional main angle a, so the regional main angle of the house structure to be corrected cannot be used as a guiding parameter for correcting the main angle of the house structure to be corrected. Therefore, the second candidate value _b2 of the main angle of the house structure to be corrected is finally used as the final true main angle.

The beneficial effects of the present invention are as follows: the present invention proposes a new method for correcting the main angle of a building based on regional consistency. The main angle of the area where the building to be corrected is located is calculated by measuring the main angle of standard buildings within a certain range around the building to be corrected. The regional main angle is used as a reference to select the optimal value among the candidate main angles to obtain the true main angle. As a post-processing method for building extraction, the present invention solves the problem of difficulty in calculating the main angle of irregular buildings, thereby greatly improving the post-processing optimization effect of building extraction in remote sensing images.

It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

The method for correcting the main angle of a building based on regional consistency is characterized by comprising the following steps:

S1 obtains a target remote sensing image, and obtains candidate main angles of buildings and structures in the target remote sensing image by a preset method, wherein the candidate main angles include a first candidate value and a second candidate value;

S2 calculates the difference between the first candidate value and the second candidate value, and according to the relationship between the difference and the first preset threshold, divides the building structure into a standard building structure and a building structure to be corrected, and simultaneously obtains the true main angle of the standard building structure;

S3 obtains the regional main angle of the building structure to be corrected through a regional statistical strategy, and the regional statistical strategy includes external expansion main angle statistics and line segment histogram statistics;

S4 calculates the difference between the candidate main angle and the regional main angle, and selects the real main angle of the building structure to be corrected from the candidate main angles according to the relationship between the difference and the second preset threshold.
The method for correcting the main angle of a building based on regional consistency according to claim 1, characterized in that step S1 comprises:

S11 obtains the angles of the two major axis center lines of the circumscribed rectangle of the minimum area of the building structure as the first candidate value of the building structure;

S12 obtains the vector contour line segment of the building structure by a vector contour extraction method, and obtains the angle of the vector contour line segment;

S13 groups vector contour line segments with similar angles into one group, selects a group of vector contour line segments with the longest total length as a reference vector contour line segment group, and obtains a second candidate value of the building structure by a vertical equivalent method based on the angle of the reference vector contour line segment;

The vector contour line segments with similar angles include vector contour line segments with an angle difference of [0°, 2°]∪[88°, 90°].
The method for correcting the main angle of a building based on regional consistency according to claim 2, characterized in that the vector contour extraction method comprises:

By performing semantic segmentation, binarization and mathematical morphology processing on remote sensing images, post-processing images are obtained;

By performing boundary tracing and vector extraction on the post-processed image, the vector contour segments of the building structure are obtained.
The method for correcting the main angle of a building based on regional consistency according to claim 2 is characterized in that the vertical equivalence method comprises:

The reference vector contour line segment group is clustered based on the angle of the reference vector contour line segment to obtain a main vector contour line segment group and an auxiliary vector contour line segment group, wherein the total length of the line segments of the main vector contour line segment group is greater than the total length of the line segments of the auxiliary vector contour line segment group, and the angles of the main vector contour line segment group and the angles of the auxiliary vector contour line segment group are respectively Find the mean and get the main vector angle and auxiliary vector angle;

The auxiliary vector angle is vertically transformed toward the main vector angle to obtain a vertical auxiliary vector angle, and the average of the main vector angle and the vertical auxiliary vector angle is used as the second candidate value.
The method for correcting the main angle of a building based on regional consistency according to claim 1, characterized in that step S2 comprises:

If the difference between the first candidate value and the second candidate value is less than a first preset threshold, the building structure is used as a standard building structure, and the second candidate value is used as the true main angle of the standard building structure;

If the difference between the first candidate value and the second candidate value is greater than the first preset threshold, the building structure is used as the building structure to be corrected.
The method for correcting the main angle of a building based on regional consistency according to claim 1 is characterized in that the regional statistical strategy is external expansion main angle statistics, and step S3 comprises:

The minimum circumscribed rectangle of the building structure to be modified is expanded by a preset distance to obtain an expanded rectangle;

The standard housing structures within the outer expansion rectangle and intersecting with the outer expansion rectangle are used as reference housing structures;

The average of the real main angles of the reference building structures is used as the regional main angle of the building structure to be corrected.
The method for correcting the main angle of a building based on regional consistency according to claim 1 is characterized in that the regional statistical strategy is line segment histogram statistics, and step S3 comprises:

The houses and structures in the target remote sensing image are partitioned by hierarchical clustering method based on distance matrix;

Obtaining the angle of the vector contour line segment of the building structure in each partition, wherein the building structure is composed of a plurality of vector contour line segments;

The angle histogram of each partition is constructed by using the angle of the vector contour line segment of the building structure in each partition;

The angle corresponding to the highest frequency in the angle histogram of the partition where the building or structure to be corrected is located is taken as the regional main angle of the building or structure to be corrected.
The method for correcting the main angle of a building based on regional consistency according to claim 1, characterized in that step S4 comprises:

Calculate the difference between the first candidate value and the main angle of the region, and the difference between the second candidate value and the main angle of the region, and select the smaller difference as the target difference;

If the target difference is less than a second preset threshold, a candidate main angle close to the regional main angle is selected as the true main angle of the building structure to be corrected;

If the target difference is greater than the second preset threshold, the second candidate value is selected as the value to be corrected. The true main angle of the building structure.