Building crack detection method and detection system based on 3D scanning image monitoring

Abstract

The present invention discloses a method and system for detecting cracks in buildings based on 3D scanning image monitoring, and relates to the technical field of building crack detection methods, including the following steps: using a 3D laser scanner to scan the surface of a building to obtain three-dimensional point cloud data of the building surface; filtering the scanned data to obtain a three-dimensional coordinate data file of the building surface, wherein the three-dimensional coordinate data file contains all the collected data points; integrating and analyzing the data points in the three-dimensional coordinate data file and extracting detection data points, and integrating to obtain a set of detection data points. The detection data points extracted in the present invention can cover the distribution area of ​​the original point cloud acquired by the scanner, avoid the occurrence of over-dense or over-sparse areas that affect the expression of the overall geometric features, effectively maintain the geometric features of the original point cloud, and achieve the geometric features and overall shape of the point cloud while reducing the complexity of subsequent calculations, thereby improving the accuracy of subsequent crack detection.

Description

Technical Field

[0001] The present invention relates to the technical field of building crack detection methods, and in particular to a building crack detection method and a detection system based on 3D scanning image monitoring. Background Art

[0002] The method of crack detection in buildings based on 3D scanning image monitoring mainly includes using 3D laser scanning technology to obtain point cloud data of the building, and constructing a three-dimensional model of the building through the point cloud data, and then performing feature extraction to achieve accurate detection of cracks.

[0003] For example, the authorization announcement number is CN118279750B, the authorization announcement date is 2024.07.26, and the name is "Prediction Method and System for Cracks in Building Walls". The method includes performing ground laser scanning on the target building to obtain high-precision three-dimensional point cloud data; performing point cloud data screening on the high-precision three-dimensional point cloud data to obtain a wall point cloud data set and a non-wall point cloud data set; performing triangular irregular network data construction on the wall point cloud data set to obtain a triangular irregular network data set; converting the triangular irregular network data set into grating surface data through an inverse distance weighted algorithm; performing crack feature extraction on the grating surface data through a shape recognition algorithm to obtain a crack feature data set, wherein the crack feature data set obtains crack shape features and crack width features.

[0004] The shortcoming of the existing technology, including the above-mentioned application, is that the point cloud data of the building is obtained through 3D laser scanning technology, and then a three-dimensional model of the building is constructed through the point cloud data. However, since the data points are too redundant, the amount of calculation is too large, which affects the subsequent processing response speed. Although deleting some data points can reduce the number of data points and thus reduce the subsequent calculation amount, it is easy to cause the loss of geometric features of the point cloud, thereby affecting the accuracy of subsequent crack detection and analysis. Summary of the invention

[0005] The purpose of the present invention is to provide a method and system for detecting cracks in buildings based on 3D scanning images, so as to solve the above-mentioned deficiencies in the prior art.

[0006] In order to achieve the above object, the present invention provides the following technical solution: a method for detecting cracks in buildings based on 3D scanning images, comprising the following steps:

[0007] Use a 3D laser scanner to scan the building surface and obtain three-dimensional point cloud data of the building surface;

[0008] Filtering the scanned data to obtain a three-dimensional coordinate data file of the building surface, wherein the three-dimensional coordinate data file contains all the collected data points;

[0009] Integrate and analyze the data points in the three-dimensional coordinate data file and extract the detection data points, and integrate to obtain a set of detection data points;

[0010] Retrieving detection data points in the detection data point set to construct a three-dimensional model of the building, and smoothing the constructed three-dimensional model of the building to eliminate jagged unevenness on the surface;

[0011] The crack area on the building surface is identified through image processing technology, and the identified crack area is segmented from the surrounding normal surface to form an independent crack object and extract the geometric characteristic parameters of the crack.

[0012] As a further description of the above technical solution: integrating and analyzing the data points in the three-dimensional coordinate data file to extract the detection data points, and obtaining the detection data point set is specifically:

[0013] Distribute each data point in the three-dimensional coordinate data file on a two-dimensional plane area based on the two-dimensional plane coordinates to obtain a two-dimensional data point distribution map, and divide the two-dimensional data point distribution map into i squares with equal areas according to preset lengths and widths;

[0014] Get the number of data points in each square and record them as , , ... ;

[0015] Get the depth coordinate difference of the data points in each small square, and integrate to get the depth coordinate difference data set of the data points in each square: , , ... ;

[0016] The distribution extraction coefficient of each square is calculated based on the number of data points in each square. The feature extraction coefficient of each square is calculated based on the depth coordinate difference of the data points in each small square. The corresponding extraction ratio of the data points in each square is calculated based on the integration of the distribution extraction coefficient and the feature extraction coefficient. The number of detected data points extracted in each square is calculated based on the extraction ratio K.

[0017] As a further description of the above technical solution: obtaining the depth coordinate difference number of the data points in each square, and integrating to obtain the depth coordinate difference data set of the data points in each square is specifically as follows:

[0018] Retrieve a data point in the grid and identify the depth coordinate of the data point, import it into the reference library, and record the depth coordinate difference of the current grid as 1;

[0019] Retrieve another data point in the grid and identify the depth coordinate of the data point, compare the current depth coordinate with the depth coordinate in the reference library to see if they are the same, if they are the same, do not record the current data point, if they are different, import the depth coordinate of the current data point into the reference library, and add 1 to the depth coordinate difference of the current grid;

[0020] Repeatedly retrieve all data points in the grid, and finally obtain the depth coordinate difference corresponding to the current grid.

[0021] As a further description of the above technical solution: based on the number of data points in each square, the distribution extraction coefficient of each square is calculated and integrated. The calculation method of the distribution extraction coefficient is: ,in represents the distribution extraction coefficient corresponding to the mth square, represents the number of data points in the mth square, where m is an integer from 1 to i, and the distribution extraction coefficient set corresponding to each square is obtained by integration: { , , ... };

[0022] The feature extraction coefficient of each square is calculated based on the depth coordinate difference of the data points in each small square. The feature extraction coefficient is calculated as follows: ,in represents the feature extraction coefficient corresponding to the m-th square, Represents the depth coordinate difference of the data points in the mth square, where m is an integer from 1 to i, and integrates to obtain the feature extraction coefficient set corresponding to each square: { , , ... }.

[0023] As a further description of the above technical solution: the corresponding extraction ratio of the data points in each grid is calculated based on the integrated distribution extraction coefficient and the feature extraction coefficient:

[0024] The distribution extraction coefficient and feature extraction coefficient of each data point in each square are weighted and summed to obtain the comprehensive extraction coefficient of the corresponding square, where the calculation logic of the comprehensive extraction coefficient is: ,in , are weight factors and , are greater than 0, represents the comprehensive extraction coefficient corresponding to the mth square;

[0025] Integrate and obtain the comprehensive extraction coefficient set corresponding to each grid: , , ... };

[0026] The extraction ratio of data points in each grid is calculated based on the comprehensive extraction coefficient corresponding to each grid, where the calculation logic of the extraction ratio of data points in the grid is: ,in Indicates the corresponding extraction ratio of data points in the mth square.

[0027] As a further description of the above technical solution: the number of extracted feature data points in each square is calculated based on the extraction ratio. Specifically, the number of extracted data points is preset first, and then the number of data points is multiplied by the corresponding extraction ratio of the data points in each square and rounded to obtain the number of extracted feature data points in each square.

[0028] As a further description of the above technical solution: before retrieving the processed data points in the feature data point set to construct the three-dimensional model of the building, the method further includes performing validity analysis on the extracted detection data points, wherein the validity analysis on the extracted detection data points is specifically as follows:

[0029] Obtaining a detection point cloud composed of detection data points and acquiring an original point cloud by scanning;

[0030] The ISS algorithm is used to extract feature data points from the original point cloud and the detection point cloud respectively. The feature data points in the detection point cloud are matched with the feature data points in the original point cloud. The matching number and matching mean of the feature data points in the detection point cloud are collected and associated and integrated for effectiveness evaluation and analysis.

[0031] As a further description of the above technical solution: matching the feature data points in the detection point cloud with the feature data points in the original point cloud is specifically performed as follows:

[0032] Arrange the characteristic data points in the detection point cloud and the characteristic data points in the original point cloud in a corresponding three-dimensional coordinate system;

[0033] Preset a matching distance threshold, select any feature data point in the detection point cloud as the center point, divide the spherical matching area in the three-dimensional coordinate system based on the matching distance threshold, and obtain the feature data points in the original point cloud in the spherical matching area;

[0034] When the feature data points in the original point cloud do not appear in the spherical matching area, it means that there are no matching data points in the feature data points in the current detection point cloud;

[0035] When there are multiple feature data points in the original point cloud in the spherical matching area, the matching value between the feature data point and the center point in each original point cloud in the spherical matching area is calculated, and the feature data point in the original point cloud with the smallest matching value is selected as the matching feature data point of the current center point, and the corresponding matching value is recorded. The calculation logic of the matching value is: , where X, Y, and Z are the three-dimensional coordinate data of the matching feature data point of the current center point.

[0036] Repeat the calculation until all feature data points in the detection point cloud are matched, and integrate to obtain the matching value data set of feature data points in the detection point cloud: , , }, Represents the matching value of the feature data point in the jth feature point cloud.

[0037] As a further description of the above technical solution: collecting the matching number and matching mean of feature data points in the feature point cloud and performing correlation integration for effectiveness evaluation analysis is as follows:

[0038] Setting the effectiveness evaluation threshold ;

[0039] Calculate the validity coefficient G of the feature data points in the feature point cloud, where the calculation logic of the validity coefficient G is , where F represents the number of feature data points in the feature point cloud, j represents the number of corresponding matching data points of feature data points in the feature point cloud, j≤F, , are weight factors and , are greater than 0, is a preset matching distance threshold;

[0040] Comparison of effectiveness coefficient G and effectiveness assessment threshold , when G> , it means that the integration analysis of the data points in the three-dimensional coordinate data file and the extraction of feature data points are valid.

[0041] The building crack detection system based on 3D scanning image monitoring is used to implement the above-mentioned building crack detection method based on 3D scanning image monitoring, including: a scanning acquisition unit, which is used to scan the surface of the building to obtain three-dimensional point cloud data of the building surface;

[0042] An information processing module, which is used to filter and denoise the three-dimensional point cloud data;

[0043] An integration and extraction module is used to integrate and analyze the three-dimensional point cloud data and extract detection data points, and integrate to obtain a set of detection data points;

[0044] A model building module, which is used to retrieve detection data points in the detection data point set to build a three-dimensional model of the building;

[0045] The image processing module is used to identify the crack area on the surface of the building, segment the identified crack area from the surrounding normal surface, form an independent crack object and extract the geometric characteristic parameters of the crack.

[0046] In the above technical solution, the present invention provides a method and system for detecting cracks in buildings based on 3D scanning images:

[0047] The method and system for detecting cracks in buildings based on 3D scanning image monitoring obtains the distribution extraction coefficient and feature extraction coefficient of the data points in each square through distribution calculation, integrates them to calculate the comprehensive extraction coefficient, and calculates the extraction ratio of the data points in each square based on the comprehensive extraction coefficient, so that when extracting detection data points from the original data points in each square, the two dimensions of data point distribution and data point geometric feature richness are associated and integrated and extracted, so that the extracted detection data points can cover the distribution area of ​​the original point cloud acquired by the scanner, avoid the appearance of over-dense or over-sparse areas, and affect the expression of the overall geometric features, and effectively maintain the geometric features of the original point cloud, so as to reduce the complexity of subsequent calculations while maintaining the geometric features and overall shape of the point cloud, thereby improving the accuracy of subsequent crack detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiments of the present application 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 recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0049] Figure 1 A schematic flow chart of a method for detecting cracks in buildings based on 3D scanning images provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0050] In order to enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings.

[0051] See also Figure 1 The embodiment of the present invention provides a technical solution: a method for detecting cracks in buildings based on 3D scanning images, comprising the following steps:

[0052] Use a 3D laser scanner to scan the building surface and obtain three-dimensional point cloud data of the building surface;

[0053] The scanning data is filtered to obtain a three-dimensional coordinate data file of the building surface, wherein the three-dimensional coordinate data file contains all the collected data points; wherein the scanning data is filtered to remove noise points and outliers from the collected three-dimensional point cloud data by using bilateral filtering or Gaussian filtering, thereby improving the quality of the collected three-dimensional point cloud data;

[0054] Integrate and analyze the data points in the three-dimensional coordinate data file and extract the detection data points to obtain a set of detection data points. Integrate and analyze the data points in the three-dimensional coordinate data file and extract the detection data points to achieve data simplification, reduce the amount of point cloud data, improve processing speed and ensure the accuracy of subsequent detection;

[0055] Retrieving detection data points in the detection data point set to construct a three-dimensional model of the building, converting the obtained discrete detection data points into a continuous three-dimensional surface model through a grid generation method, and smoothing the constructed three-dimensional model of the building to eliminate jagged unevenness on the surface;

[0056] The crack area on the building surface is identified through image processing technology, and the identified crack area is segmented from the surrounding normal surface by using threshold segmentation algorithm to form an independent crack object and extract the geometric characteristic parameters of the crack.

[0057] The data points in the three-dimensional coordinate data file are integrated and analyzed to extract the detection data points, and the detection data point set is obtained as follows:

[0058] Each data point in the three-dimensional coordinate data file is distributed on a two-dimensional plane area based on the two-dimensional plane coordinates to obtain a two-dimensional data point distribution map, and the two-dimensional data point distribution map is divided into i squares with equal areas according to preset lengths and widths; it should be noted that the data point has three coordinate values ​​of X, Y, and Z, wherein the X coordinate value represents the position of the point in the horizontal direction, the Y coordinate value represents the position of the point in the vertical direction, and the Z coordinate value represents the position of the point in the depth direction, and the data points are distributed on the two-dimensional plane area through the X coordinate position and the Y coordinate position of each data point.

[0059] Get the number of data points in each square and record them as , , ... ;in represents the number of data points in the i-th square;

[0060] Get the depth coordinate difference of the data points in each small square, and integrate to get the depth coordinate difference data set of the data points in each square: , , ... ;in Represents the depth coordinate difference dataset of the data points in the i-th square.

[0061] The depth coordinate difference of the data points in each grid is obtained, and the depth coordinate difference data set of the data points in each grid is obtained by integration:

[0062] Retrieve a data point in the grid and identify the depth coordinate of the data point, import it into the reference library, and record the depth coordinate difference of the current grid as 1;

[0063] Retrieve another data point in the grid and identify the depth coordinate of the data point, compare the current depth coordinate with the depth coordinate in the reference library to see if they are the same, if they are the same, do not record the current data point, if they are different, import the depth coordinate of the current data point into the reference library, and add 1 to the depth coordinate difference of the current grid;

[0064] Repeatedly retrieve all data points in the grid, and finally obtain the depth coordinate difference number corresponding to the current grid. By collecting the depth coordinate difference number of all data points in each grid, the richness of the geometric features of the data points in each grid can be quantitatively represented, where the greater the depth coordinate difference number of the data points in the grid, the richer the collective features of the data points in the grid.

[0065] The distribution extraction coefficient of each square is calculated based on the number of data points in each square. The feature extraction coefficient of each square is calculated based on the depth coordinate difference of the data points in each small square. The corresponding extraction ratio of the data points in each square is calculated based on the integration of the distribution extraction coefficient and the feature extraction coefficient. The number of detected data points extracted in each square is calculated based on the extraction ratio K.

[0066] The distribution extraction coefficient of each square is calculated based on the number of data points in each square. The calculation method of the distribution extraction coefficient is: ,in represents the distribution extraction coefficient corresponding to the mth square, represents the number of data points in the mth square, where m is an integer from 1 to i, and the distribution extraction coefficient set corresponding to each square is obtained by integration: { , , ... };

[0067] The feature extraction coefficient of each square is calculated based on the depth coordinate difference of the data points in each small square. The feature extraction coefficient is calculated as follows: ,in represents the feature extraction coefficient corresponding to the m-th square, Represents the depth coordinate difference of the data points in the mth square, where m is an integer from 1 to i, and integrates to obtain the feature extraction coefficient set corresponding to each square: { , , ... }.

[0068] Based on the distribution extraction coefficient and feature extraction coefficient, the corresponding extraction ratio of the data points in each grid is calculated as follows:

[0069] The distribution extraction coefficient and feature extraction coefficient of each data point in each square are weighted and summed to obtain the comprehensive extraction coefficient of the corresponding square, where the calculation logic of the comprehensive extraction coefficient is: ,in , are weight factors and , are greater than 0, Indicates the comprehensive extraction coefficient corresponding to the mth square, optional , ;

[0070] Integrate and obtain the comprehensive extraction coefficient set corresponding to each grid: , , ... };

[0071] The extraction ratio of data points in each grid is calculated based on the comprehensive extraction coefficient corresponding to each grid, where the calculation logic of the extraction ratio of data points in the grid is: ,in Indicates the corresponding extraction ratio of data points in the mth square.

[0072] The specific method of calculating the number of extracted feature data points in each grid based on the extraction ratio is to first preset the number of extracted data points, then multiply the number of data points by the corresponding extraction ratio of the data points in each grid and round it to obtain the number of extracted feature data points in each grid.

[0073] It should be noted that the distribution extraction coefficient and feature extraction coefficient of the data points in each square are obtained through distribution calculation, and the comprehensive extraction coefficient is integrated to calculate the comprehensive extraction coefficient, and the data point extraction ratio in each square is calculated based on the comprehensive extraction coefficient, so that when extracting detection data points from the original data points in each square, the two dimensions of data point distribution and data point geometric feature richness are correlated and integrated and extracted, so that the extracted detection data points can cover the distribution area of ​​the original point cloud acquired by the scanner, avoiding overly dense or sparse areas that affect the expression of the overall geometric features, and effectively maintaining the geometric features of the original point cloud, thereby reducing the complexity of subsequent calculations while maintaining the geometric features and overall shape of the point cloud, thereby improving the accuracy of subsequent crack detection.

[0074] Furthermore, before retrieving the processed data points in the feature data point set to construct the three-dimensional building model, the method further includes performing validity analysis on the extracted detection data points, wherein the validity analysis on the extracted detection data points is specifically as follows:

[0075] Obtaining a detection point cloud composed of detection data points and acquiring an original point cloud by scanning;

[0076] The ISS algorithm is used to extract feature data points from the original point cloud and the detection point cloud respectively. The feature data points in the detection point cloud are matched with the feature data points in the original point cloud. The matching number and matching mean of the feature data points in the detection point cloud are collected and associated and integrated for effectiveness evaluation and analysis.

[0077] The matching process of the feature data points in the detection point cloud and the feature data points in the original point cloud is specifically as follows: the feature data points in the detection point cloud and the feature data points in the original point cloud are arranged in a three-dimensional coordinate system in correspondence;

[0078] Preset a matching distance threshold, select any feature data point in the detection point cloud as the center point, divide the spherical matching area in the three-dimensional coordinate system based on the matching distance threshold, and obtain the feature data points in the original point cloud in the spherical matching area;

[0079] When the feature data points in the original point cloud do not appear in the spherical matching area, it means that there are no matching data points in the feature data points in the current detection point cloud;

[0080] When there are multiple feature data points in the original point cloud in the spherical matching area, the matching value between the feature data point and the center point in each original point cloud in the spherical matching area is calculated, and the feature data point in the original point cloud with the smallest matching value is selected as the matching feature data point of the current center point, and the corresponding matching value is recorded. The calculation logic of the matching value is: , where X, Y, and Z are the three-dimensional coordinate data of the matching feature data point of the current center point.

[0081] Repeat the calculation until all feature data points in the detection point cloud are matched, and integrate to obtain the matching value data set of feature data points in the detection point cloud: , , }, Represents the matching value of the feature data point in the jth feature point cloud.

[0082] The matching number and matching mean of feature data points in the feature point cloud are collected and associated and integrated for effectiveness evaluation and analysis as follows:

[0083] Setting the effectiveness evaluation threshold ;

[0084] Calculate the validity coefficient G of the feature data points in the feature point cloud, where the calculation logic of the validity coefficient G is , where F represents the number of feature data points in the feature point cloud, j represents the number of corresponding matching data points of feature data points in the feature point cloud, j≤F, , are weight factors, and , are greater than 0, is a preset matching distance threshold;

[0085] Comparison of effectiveness coefficient G and effectiveness assessment threshold , when G> When , it means that the validity of the extracted feature data points is qualified by integrating and analyzing the data points in the three-dimensional coordinate data file. It should be noted that by matching the feature data points in the detection point cloud with the feature data points in the original point cloud, and evaluating the validity of the associated integration based on the matching number and matching mean of the feature data points in the detection point cloud, the validity of the extracted detection data points can be evaluated and analyzed to avoid the loss of important information of the extracted detection data points.

[0086] A crack detection system for monitoring buildings based on 3D scanning images is used to implement the above-mentioned crack detection method for monitoring buildings based on 3D scanning images, including: a scanning acquisition unit, which is used to scan the surface of a building and obtain three-dimensional point cloud data of the building surface; an information processing module, which is used to filter and denoise the three-dimensional point cloud data; an integration and extraction module, which is used to integrate and analyze the three-dimensional point cloud data and extract detection data points, and integrate to obtain a set of detection data points; a model construction module, which is used to call the detection data points in the detection data point set to construct a three-dimensional model of the building; an image processing module, which is used to identify crack areas on the surface of the building, segment the identified crack areas from the surrounding normal surfaces, form independent crack objects and extract geometric feature parameters of the cracks.

[0087] In addition, the parts of the above-mentioned technical solutions provided in the embodiments of the present application that are consistent with the corresponding technical solutions in the prior art in terms of implementation principle are not described in detail to avoid excessive elaboration. The above only describes certain exemplary embodiments of the present invention by way of illustration. It is undoubted that for those of ordinary skill in the art, the described embodiments can be modified in various different ways without departing from the spirit and scope of the present invention. Therefore, the above-mentioned drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A method for detecting cracks in buildings based on 3D scanning images, characterized in that: The following steps are involved: Use a 3D laser scanner to scan the building surface and obtain three-dimensional point cloud data of the building surface; Filtering the scanned data to obtain a three-dimensional coordinate data file of the building surface, wherein the three-dimensional coordinate data file contains all the collected data points; Integrate and analyze the data points in the three-dimensional coordinate data file and extract the detection data points, and integrate to obtain a set of detection data points; Retrieving detection data points in the detection data point set to construct a three-dimensional model of the building, and smoothing the constructed three-dimensional model of the building to eliminate jagged unevenness on the surface; The crack area on the building surface is identified through image processing technology, and the identified crack area is segmented from the surrounding normal surface to form an independent crack object and extract the geometric characteristic parameters of the crack; The data points in the three-dimensional coordinate data file are integrated and analyzed to extract the detection data points, and the detection data point set is obtained as follows: Distribute each data point in the three-dimensional coordinate data file on a two-dimensional plane area based on the two-dimensional plane coordinates to obtain a two-dimensional data point distribution map, and divide the two-dimensional data point distribution map into i squares with equal areas according to preset lengths and widths; Get the number of data points in each square and record them as S1, S2, S3...S i ; Get the depth coordinate difference of the data points in each small square, and integrate to get the depth coordinate difference data set of the data points in each square: C1, C2, C3...C i ; Based on the number of data points in each square, the distribution extraction coefficient of each square is calculated, and the feature extraction coefficient of each square is calculated based on the depth coordinate difference of the data points in each small square. The corresponding extraction ratio of the data points in each square is calculated based on the integration of the distribution extraction coefficient and the feature extraction coefficient, and the number of detection data points extracted in each square is calculated based on the extraction ratio K; The distribution extraction coefficient of each square is calculated based on the number of data points in each square. The calculation method of the distribution extraction coefficient is: Where T am represents the distribution extraction coefficient corresponding to the mth square, S m represents the number of data points in the mth square, where m is an integer from 1 to i, and the distribution extraction coefficient set corresponding to each square is obtained by integration: {T a1 , T a2 , T a3 ...T ai }; The feature extraction coefficient of each square is calculated based on the depth coordinate difference of the data points in each small square. The feature extraction coefficient is calculated as follows: Where T bm represents the feature extraction coefficient corresponding to the mth square, C m represents the depth coordinate difference of the data points in the mth square, where m is an integer from 1 to i, and the feature extraction coefficient set corresponding to each square is obtained by integration: {T b1 、T b2 、T b3 ...T bi }; The corresponding extraction ratio of the data points in each grid is calculated based on the integrated distribution extraction coefficient and the feature extraction coefficient: The distribution extraction coefficient and feature extraction coefficient of each data point in each square are weighted and summed to obtain the comprehensive extraction coefficient of the corresponding square, where the calculation logic of the comprehensive extraction coefficient is: in are weight factors and Both are greater than 0, T cm represents the comprehensive extraction coefficient corresponding to the mth square; Integrate and obtain the comprehensive extraction coefficient set corresponding to each grid: {T c1 、T c2 、T c3 ...T ci }; The extraction ratio of data points in each grid is calculated based on the comprehensive extraction coefficient corresponding to each grid, where the calculation logic of the extraction ratio of data points in the grid is: Where K m Indicates the corresponding extraction ratio of data points in the mth square.

2. The method for detecting cracks in buildings based on 3D scanning images according to claim 1, characterized in that: The depth coordinate difference of the data points in each grid is obtained, and the depth coordinate difference data set of the data points in each grid is obtained by integration: Retrieve a data point in the grid and identify the depth coordinate of the data point, import it into the reference library, and record the depth coordinate difference of the current grid as 1; Retrieve another data point in the grid and identify the depth coordinate of the data point, compare the current depth coordinate with the depth coordinate in the reference library to see if they are the same, if they are the same, do not record the current data point, if they are different, import the depth coordinate of the current data point into the reference library, and add 1 to the depth coordinate difference of the current grid; Repeatedly retrieve all data points in the grid, and finally obtain the depth coordinate difference corresponding to the current grid.

3. The method for detecting cracks in buildings based on 3D scanning images according to claim 1, characterized in that: The specific method of calculating the number of extracted feature data points in each grid based on the extraction ratio is to first preset the number of extracted data points, then multiply the number of data points by the corresponding extraction ratio of the data points in each grid and round it to obtain the number of extracted feature data points in each grid.

4. The method for detecting cracks in buildings based on 3D scanning image monitoring according to claim 1, characterized in that: Before retrieving the processed data points in the feature data point set to construct the three-dimensional model of the building, the validity analysis of the extracted detection data points is also included, wherein the validity analysis of the extracted detection data points is specifically as follows: Obtaining a detection point cloud composed of detection data points and acquiring an original point cloud by scanning; The ISS algorithm is used to extract feature data points from the original point cloud and the detection point cloud respectively. The feature data points in the detection point cloud are matched with the feature data points in the original point cloud. The matching number and matching mean of the feature data points in the detection point cloud are collected and associated and integrated for effectiveness evaluation and analysis.

5. The method for detecting cracks in buildings based on 3D scanning images according to claim 4 is characterized in that: The matching process of the feature data points in the detection point cloud and the feature data points in the original point cloud is as follows: Arrange the characteristic data points in the detection point cloud and the characteristic data points in the original point cloud in a corresponding three-dimensional coordinate system; Preset a matching distance threshold, select any feature data point in the detection point cloud as the center point, divide the spherical matching area in the three-dimensional coordinate system based on the matching distance threshold, and obtain the feature data points in the original point cloud in the spherical matching area; When the feature data points in the original point cloud do not appear in the spherical matching area, it means that there are no matching data points in the feature data points in the current detection point cloud; When there are multiple feature data points in the original point cloud in the spherical matching area, the matching value between the feature data point and the center point in each original point cloud in the spherical matching area is calculated, and the feature data point in the original point cloud with the smallest matching value is selected as the matching feature data point of the current center point, and the corresponding matching value is recorded. The calculation logic of the matching value is: Among them, X, Y, and Z are the three-dimensional coordinate data of the matching feature data point of the current center point. Repeat the calculation until all feature data points in the detection point cloud are matched, and integrate to obtain the matching value data set of feature data points in the detection point cloud: P1 , D P2 , D P3 ...D Pj }, D Pj Represents the matching value of the feature data point in the jth feature point cloud.

6. The method for detecting cracks in buildings based on 3D scanning images according to claim 5, characterized in that: The matching number and matching mean of feature data points in the feature point cloud are collected and associated and integrated for effectiveness evaluation and analysis as follows: Set the effectiveness evaluation threshold G e ; Calculate the validity coefficient G of the feature data points in the feature point cloud, where the calculation logic of the validity coefficient G is Where F represents the number of feature data points in the feature point cloud, j represents the number of corresponding matching data points of feature data points in the feature point cloud, j≤F, are weight factors and Both are greater than 0, L e is a preset matching distance threshold; Comparison of the effectiveness coefficient G and the effectiveness assessment threshold G e , when G>G e , it means that the integration analysis of the data points in the three-dimensional coordinate data file and the extraction of feature data points are valid.

7. A building crack detection system based on 3D scanning images, which is used to implement the building crack detection method based on 3D scanning images as described in any one of claims 1 to 6, characterized in that: include: A scanning acquisition unit, which is used to scan the surface of the building and obtain three-dimensional point cloud data of the building surface; An information processing module, which is used to filter and denoise the three-dimensional point cloud data; An integration and extraction module is used to integrate and analyze the three-dimensional point cloud data and extract detection data points, and integrate to obtain a set of detection data points; A model building module, which is used to retrieve detection data points in the detection data point set to build a three-dimensional model of the building; The image processing module is used to identify the crack area on the surface of the building, segment the identified crack area from the surrounding normal surface, form an independent crack object and extract the geometric characteristic parameters of the crack.

Images