A computer vision-based building foundation settlement detection method

Abstract

The application belongs to the field of foundation settlement detection, and relates to a data analysis technique, in particular to a building foundation settlement detection method based on computer vision, which comprises a detection verification sub-method and an instant decision sub-method; the detection verification sub-method comprises the following steps: performing visual settlement detection analysis on the building foundation; analyzing the effectiveness of the visual settlement detection result of the building foundation observation point; performing visual settlement risk analysis on the building foundation observation point; performing settlement verification analysis on the building foundation observation point; and outputting the settlement detection result of the building foundation observation point; the application can effectively improve the precision and anti-interference capability of foundation settlement detection, and through the double verification mechanism of multi-camera redundant arrangement and coordinate change amount, based on the pixel grid image processing and the coordinate extraction technology of two-dimensional code target, the accurate measurement of millimeter level displacement is realized, and high-precision data support is provided for building structure safety evaluation.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of foundation settlement detection, and relates to a data analysis technique, in particular to a building foundation settlement detection method based on computer vision. BACKGROUND

[0002] Building foundation settlement detection is a key technology for evaluating the structural safety of buildings by measuring their vertical displacement changes in real time or periodically. Traditional methods rely on manual measurements, while modern techniques combine sensors, computer vision, and artificial intelligence to achieve automated, high-precision, and low-cost monitoring.

[0003] The invention patent with publication number CN118095813B discloses a visual monitoring method and system for foundation settlement based on BIM technology. The visual monitoring method can input the extracted indicators into a foundation settlement prediction model to perform settlement prediction and obtain foundation settlement risk indicators. However, this visual monitoring method relies solely on a single algorithm's prediction model for settlement detection and cannot be verified by multiple algorithms. When the collection terminal fails or the collected data is disturbed by the environment, the detection results have large errors. In addition, when the multiple algorithm detection results are inconsistent, the existing technology cannot generate a decision result by combining environmental parameters and historical verification data, resulting in a lag in the output of the results.

[0004] To solve the above technical problems, the present application provides a solution. SUMMARY

[0005] The purpose of the present application is to provide a building foundation settlement detection method based on computer vision, which solves the problem that the existing technology cannot be verified by multiple algorithms.

[0006] The technical problem to be solved by the present application is how to provide a building foundation settlement detection method based on computer vision that can be verified by multiple algorithms.

[0007] The purpose of the present application can be achieved by the following technical solutions:

[0008] A building foundation settlement detection method based on computer vision includes a detection verification sub-method and an instant decision sub-method.

[0009] The detection verification sub-method includes the following steps:

[0010] Step S1: Visual settlement detection analysis of the building foundation;

[0011] Step S2: Analysis of the effectiveness of the visual settlement detection results of the building foundation observation points;

[0012] Step S3: Visual settlement risk analysis of the building foundation observation points;

[0013] Step S4: settlement verification analysis is performed on the building foundation observation point;

[0014] Step S5: the settlement detection result of the building foundation observation point is output;

[0015] The instant decision sub-method comprises the following steps:

[0016] Step P1: statistical analysis is performed on the artificial verification result;

[0017] Step P2: environmental correlation analysis is performed on the artificial verification result;

[0018] Step P3: an instant decision result is generated.

[0019] Further, in step S1, the specific process of visual settlement detection analysis comprises: arranging a camera array at the observation point of the building, regularly taking images of the observation point by the camera, marking the taken images as detection images, enlarging the detection images into pixel grid images, extracting the coordinates of the two-dimensional code targets in the detection images, marking the distance between the two-dimensional code target coordinate points extracted by the same camera in the last two times as the settlement detection value of the camera.

[0020] Further, in step S2, the specific process of visual settlement detection result effectiveness analysis comprises: calculating the variance of the settlement detection values collected by all cameras in the camera array of the same observation point to obtain an effective coefficient of the observation point, comparing the effective coefficient with a preset effective threshold value: if the effective coefficient is less than the effective threshold value, it is determined that the visual settlement detection result effectiveness of the observation point meets the requirements, the settlement detection values collected by all cameras in the camera array of the observation point are summed and averaged to obtain a visual settlement value; if the effective coefficient is greater than or equal to the effective threshold value, it is determined that the visual settlement detection result effectiveness of the observation point does not meet the requirements, a camera maintenance signal is generated and sent to the mobile terminal of the management personnel.

[0021] Further, in step S3, the specific process of visual settlement risk analysis comprises: comparing the visual settlement value of the observation point with a preset settlement threshold value: if the visual settlement value is less than the settlement threshold value, it is determined that the observation point has no visual settlement risk; if the visual settlement value is greater than or equal to the settlement threshold value, it is determined that the observation point has a visual settlement risk.

[0022] Further, in step S4, the specific process of the settlement verification analysis includes: constructing a detection period by the time points of the latest two image acquisitions by the camera, obtaining the horizontal angle change amount of the observation point in the detection period by the inclinometer and marking it as a settlement verification value, comparing the settlement verification value with a preset settlement verification threshold value: if the settlement verification value is less than the settlement verification threshold value, it is determined that the observation point does not have a verification settlement risk; if the settlement verification value is greater than or equal to the settlement verification threshold value, it is determined that the observation point has a verification settlement risk.

[0023] Further, in step S5, the specific process of the settlement detection result output includes: if the observation point has both visual settlement risk and verification settlement risk, it is determined that the settlement detection result of the observation point is unqualified, a settlement warning signal is generated and sent to the server; if the observation point has neither visual settlement risk nor verification settlement risk, it is determined that the settlement detection result of the observation point is qualified, a detection qualified signal is generated and sent to the server; otherwise, it is determined that the settlement detection result of the observation point is controversial, the corresponding observation point is marked as a verification point, an artificial verification signal is generated and sent to the server; after receiving the artificial verification signal, the server sends the artificial verification signal to the mobile terminal of the manager, and the manager performs artificial verification on the observation point, and marks the visual settlement detection process or the settlement verification process as a correct process.

[0024] Further, in step P1, a statistical period is generated, and the detection value JCi of the environmental parameter i at the time of marking the verification point in the statistical period is obtained, i=1, 2, …, n, n is a positive integer, and the environmental parameter i includes the air temperature value, the air humidity value, the fog concentration value, the rainfall, the wind grade of the verification point.

[0025] Further, in step P2, the specific process of the environmental correlation analysis includes: when the visual settlement detection process is marked as a correct process, the corresponding verification point is marked as a visual correlation point; when the settlement verification process is marked as a correct process, the corresponding verification point is marked as a verification correlation point; data cleaning processing is performed on the visual correlation point to obtain the visual priority range SYi of the environmental parameter i; the same way is used to perform data cleaning processing on the verification correlation point to obtain the verification priority range HYi of the environmental parameter i.

[0026] Further, in step P2, the specific process of data cleaning of the visual correlation points comprises: a visual set i of the environmental parameter i is formed by the detection value JCi of the environmental parameter i of all visual correlation points, variance calculation is performed on all elements of the visual set i to obtain a visual central value SJi of the environmental parameter i, the visual central value SJi is compared with a preset central threshold JZi: if the visual central value SJi is greater than or equal to the central threshold JZi, the maximum element and the minimum element in the visual set i are removed, and then the visual central value SJi is recalculated, and the process is repeated until the visual central value SJi is less than the central threshold JZi; if the visual central value SJi is less than the central threshold JZi, the maximum element and the minimum element retained in the visual set i form a visual priority range SYi of the environmental parameter i.

[0027] Further, in step P3, the generation process of the instant decision result comprises: when the settlement detection result of the subsequent observation point is controversial, the detection value JCi of the environmental parameter i of the observation point is obtained, and it is determined whether the detection value JCi of the environmental parameter i is located in the corresponding visual priority range SYi: if yes, the visual settlement detection result in step S3 is sent to the server; if no, it is determined whether the detection value JCi of the environmental parameter i is located in the corresponding verification priority range HYi: if yes, the settlement verification result in step S4 is sent to the server; if no, an artificial verification signal is generated and sent to the server.

[0028] The present application has the following advantages:

[0029] 1. The present application can effectively improve the precision and anti-interference ability of foundation settlement detection, through the redundant arrangement of multiple cameras and the double verification mechanism of coordinate change amount, the problem of insufficient data reliability of traditional methods under equipment failure or environmental interference is solved; at the same time, based on the pixel grid image processing and the coordinate extraction technology of two-dimensional code target, the accurate measurement of millimeter level displacement is realized, which provides high-precision data support for building structure safety evaluation;

[0030] 2. The present application solves the problem of insufficient reliability of single detection means in the prior art, through the data cross verification of visual and physical sensors, the accuracy of the settlement detection result is significantly improved. At the same time, when the detection result is controversial, the artificial verification process can be triggered to further confirm the problem root, such as distinguishing equipment failure and environmental interference, so as to improve the overall robustness of the system;

[0031] 3. The application establishes an environment-related priority range database, makes intelligent decisions by combining real-time environmental parameters when the detection results are controversial, and effectively solves the problem of multi-algorithm verification failure caused by sensor failure or environmental interference. At the same time, through the continuous optimization of historical verification data, the accuracy of the priority range determination is gradually improved with the running time of the system. BRIEF DESCRIPTION OF DRAWINGS

[0032] In order to more clearly illustrate the technical solutions of the embodiments of the present application or the prior art, the drawings needed in the embodiments or prior art description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0033] Figure 1 The method flowchart of the first embodiment of the present application is shown in the figure.

[0034] Figure 2 The method flowchart of the second embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0035] The technical solutions of the present application will be described below in conjunction with the embodiments. Obviously, the described embodiments are only some of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0036] In the prior art, building foundation settlement detection mainly relies on manual measurement or single algorithm model for prediction analysis. Traditional manual measurement has the problems of low efficiency and high cost, while automatic monitoring system based on sensors or computer vision improves efficiency, but generally lacks multi-algorithm cross-validation mechanism. When the acquisition device fails or the monitoring data is disturbed by the environment, a single detection result is prone to error. In addition, when the existing technology encounters contradictory multi-source data, manual intervention is usually required for judgment, and there is a lack of intelligent decision-making mechanism based on historical verification data and environmental parameters, resulting in a lag in the output of the results. For example, in complex environmental conditions such as rain and fog or sudden temperature changes, visual detection data and sensor data may deviate significantly, at which time the system cannot quickly determine the reliable data source, affecting the timeliness of the monitoring.

[0037] To address the aforementioned issues, the inventors identified two core shortcomings in existing settlement monitoring systems: insufficient reliability of single detection methods and a lack of intelligent decision-making mechanisms for disputed data. Regarding the first point, they proposed introducing a dual verification mechanism combining visual inspection and physical sensors, enhancing the reliability of results through cross-validation of multi-source data. For the second point, they discovered a hidden correlation between environmental factors and the applicability of detection methods within manual verification records, leading to the proposal to establish a dynamic mapping relationship between environmental parameters and detection method priorities. Based on this, a technical approach combining phased detection and verification with intelligent real-time decision-making is gradually formed: firstly, the accuracy of basic data is ensured through dual analysis of visual inspection and inclinometer verification; when results are disputed, the optimal detection method is automatically selected by combining environmental correlation characteristics from historical verification data, thereby reducing the frequency of manual intervention.

[0038] A computer vision-based method for detecting building foundation settlement includes a detection and verification sub-method and an instant decision-making sub-method.

[0039] Example 1: As Figure 1 As shown, the detection and verification sub-method includes the following steps:

[0040] Step S1: Visual settlement detection and analysis of building foundation: An array of cameras is set up at the observation points of the building. The cameras periodically take pictures of the observation points and mark the pictures as detection images. The detection images are enlarged into pixel grid images and the coordinates of the QR code targets in the detection images are extracted. The distance between the coordinates of the two most recent QR code targets extracted by the same camera is marked as the settlement detection value of the camera.

[0041] Among these, a camera array refers to a shooting system composed of multiple cameras, which can be implemented using industrial cameras fixedly installed at different angles around the observation point to ensure the integrity of image acquisition through multi-angle coverage. A pixel-grid image refers to a gridded image formed by magnifying the original image, which can be achieved using image interpolation algorithms, such as bilinear interpolation or cubic convolution interpolation. Increasing resolution facilitates precise location of coordinate points. QR code target coordinate extraction refers to obtaining the positional information of the QR code in an image through image recognition algorithms. This can be achieved using edge detection combined with feature matching algorithms, such as feature point detection based on SIFT or ORB, to determine the coordinates of its center point by recognizing the geometric features of the QR code.

[0042] Specifically, multiple cameras are arranged around the observation point to form an array, and each camera collects a foundation image at a preset time interval. The collected image is processed to convert it into a pixel grid image, and an image processing algorithm is used to identify the two-dimensional code target in the image and extract the center coordinates. By comparing the distance change between the coordinate points collected by the same camera at adjacent times, the settlement detection value of the camera is calculated. For example, when the two-dimensional code center coordinates of a camera are (x1, y1) in the first shooting and (x2, y2) in the second shooting, the distance between the two coordinate points is calculated as the settlement amount by the Euclidean distance formula. This method based on multiple camera redundant collection and coordinate comparison can effectively eliminate single-point measurement errors and provide reliable data basis for subsequent analysis.

[0043] Step S2: analyzing the effectiveness of the visual settlement detection result of the building foundation observation point: calculating the variance of the settlement detection values collected by all cameras in the camera array of the same observation point to obtain the effective coefficient of the observation point, and comparing the effective coefficient with the preset effective threshold value: if the effective coefficient is less than the effective threshold value, it is determined that the visual settlement detection result of the observation point meets the requirements, and the settlement detection values collected by all cameras in the camera array of the observation point are summed and averaged to obtain the visual settlement value; if the effective coefficient is greater than or equal to the effective threshold value, it is determined that the visual settlement detection result of the observation point does not meet the requirements, a camera maintenance signal is generated and sent to the mobile terminal of the management personnel;

[0044] The effective coefficient is a quantitative index reflecting the consistency of the camera array data calculated by the variance formula, which can be used to judge the data reliability. The effective threshold value is a critical value for judging the effectiveness of the data, which can be determined by experiments or historical data, for example, it can be a variance threshold value of 0.5 mm². The camera maintenance signal is a command to trigger equipment maintenance, which can be pushed to the terminal through the Internet of Things communication module to realize the rapid positioning of the faulty equipment.

[0045] Specifically, when the settlement detection values collected by multiple devices in the camera array have low dispersion, it indicates that the data consistency of each camera is high, and the visual settlement value is obtained by averaging to improve the detection precision; when the dispersion exceeds the threshold value, it indicates that some cameras may have installation deviation, lens contamination or hardware failure, and the maintenance process is automatically triggered at this time.

[0046] Step S3: Visual settlement risk analysis on the building foundation observation point: compare the visual settlement value of the observation point with the preset settlement threshold value: if the visual settlement value is less than the settlement threshold value, it is determined that the observation point has no visual settlement risk; if the visual settlement value is greater than or equal to the settlement threshold value, it is determined that the observation point has visual settlement risk;

[0047] The visual settlement value refers to the average value of the settlement detection value collected by the camera array after effectiveness verification, which can be calculated by using image processing algorithm to weight and average the coordinate change of the two-dimensional code target. The function is to eliminate the measurement error of single camera and improve the reliability of settlement data.

[0048] The settlement threshold value refers to the maximum allowable settlement amount set according to the building structure safety standard, which can be set as a percentage value of the foundation deformation limit value in the building specification, for example, 5mm. The function is to provide quantitative basis for automatic risk judgment.

[0049] Step S4: Settlement verification analysis on the building foundation observation point: the detection period is composed of the time of the last two image acquisitions by the camera, and the horizontal angle change of the observation point in the detection period is obtained by the inclinometer and marked as the settlement verification value. Compare the settlement verification value with the preset settlement verification threshold value: if the settlement verification value is less than the settlement verification threshold value, it is determined that the observation point has no verification settlement risk; if the settlement verification value is greater than or equal to the settlement verification threshold value, it is determined that the observation point has verification settlement risk;

[0050] The detection period refers to the time interval between adjacent two image acquisitions, which can be determined by the time stamp data of the camera, and is used to limit the calculation period of the settlement verification value to ensure the timeliness of the data.

[0051] The inclinometer refers to a sensor for measuring the horizontal angle change of the observation point, which can be realized by using high-precision electronic inclinometer, and provides independent physical parameters for settlement verification through continuous recording of angle change.

[0052] The settlement verification value refers to the cumulative amount of horizontal angle change in the detection period, which can be calculated by the difference value of angle data output by the inclinometer, and is used to reflect the inclination degree of the foundation as an auxiliary verification index of visual settlement detection.

[0053] The settlement verification threshold value refers to the pre-set angle change threshold value, which can be determined according to the foundation structure safety standard or historical data, and is used to judge whether the foundation inclination exceeds the allowable range.

[0054] Specifically, in the settlement verification analysis process, first, the detection period is determined according to the time point of the camera image acquisition, for example, the interval between adjacent two shootings is 24 hours. Then, the horizontal angle change amount in this period is obtained through the tiltmeter installed at the observation point, for example, the cumulative tilt angle is 0.5 degrees. Compare this change with the preset settlement verification threshold, for example, the threshold is 0.3 degrees. If the detection value exceeds the threshold, it is determined that there is a risk of verifying settlement, indicating that the foundation may be tilted; if it does not exceed the threshold, it is determined that the risk is controllable. This process introduces tiltmeter data and forms a double verification with visual detection results, which can effectively identify visual detection errors caused by camera failure or environmental interference.

[0055] Compared with the prior art, the existing method relies on a single visual detection algorithm, which is prone to misjudgment when the device is abnormal or the environment is disturbed. However, the present scheme combines tiltmeter physical measurement data to construct a multi-source verification mechanism, which can distinguish between real settlement and device error. For example, when the camera lens is contaminated and the visual detection is abnormal, the tiltmeter data can still provide a reliable basis for judgment, thereby reducing the false positive rate.

[0056] Step S5: Output the settlement detection result of the building foundation observation point: If the observation point has both visual settlement risk and verification settlement risk, it is determined that the settlement detection result of the observation point is unqualified, a settlement warning signal is generated and sent to the server; If the observation point does not have both visual settlement risk and verification settlement risk, it is determined that the settlement detection result of the observation point is qualified, a detection qualified signal is generated and sent to the server; Otherwise, it is determined that the settlement detection result of the observation point is controversial, and the corresponding observation point is marked as a verification point, an artificial verification signal is generated and sent to the server; After the server receives the artificial verification signal, the artificial verification signal is sent to the mobile terminal of the management personnel, and the management personnel verifies the observation point manually, and marks the visual settlement detection process or the settlement verification process as the correct process;

[0057] Among them, the visual settlement risk refers to the situation that the settlement value calculated by the two-dimensional code target coordinate change amount collected by the camera array exceeds the preset threshold, which can be realized by pixel-level analysis of the detection image using image processing algorithm, and is used to reflect the degree of structure deformation based on visual data. The verification settlement risk refers to the situation that the horizontal angle change measured by the tiltmeter exceeds the verification threshold, which can be realized by real-time data collection using high-precision tilt sensor, and is used to verify the physical accuracy of the visual detection result. The artificial verification signal refers to the recheck instruction triggered when the two detection results are different, which can be realized through the communication interface between the server and the mobile terminal, and is used to start the artificial intervention mechanism to ensure the detection reliability.

[0058] Specifically, when both visual settlement detection and tilt angle verification results exceed the threshold, a settlement risk is directly identified and an early warning is triggered; when neither exceeds the threshold, the structural safety status is confirmed; when conflicting detection results occur, the system automatically marks the disputed observation point and pushes it to management personnel for on-site verification. Management personnel confirm which process in visual detection or tilt angle verification is erroneous through on-site measurements and feeds back the correct detection method to the system for recording. This multimodal detection result cross-verification mechanism effectively avoids misjudgments caused by single sensor failure or environmental interference.

[0059] Example 2: Figure 2 As shown, the instant decision-making sub-method includes the following steps:

[0060] Step P1: Perform statistical analysis on the manual verification results: generate a statistical period, and obtain the detection value JCi of the environmental parameter i when the verification point is marked in the statistical period, i=1,2,...,n, where n is a positive integer. The environmental parameter i includes the air temperature value, air humidity value, fog concentration value, rainfall, wind force level, etc. of the verification point.

[0061] The statistical period refers to the time interval used for data collection, which can be implemented using a fixed duration or dynamically adjusted methods, such as setting it to 24 hours or automatically adjusting the period length based on the number of verification points. Environmental parameter i refers to the set of external factors that may affect the operating status of the detection equipment, which can be collected through devices such as temperature sensors, humidity sensors, optical sensors, rain gauges, and anemometers. The measured value JCi refers to the actual measured value of each environmental parameter at the time the verification point is marked, which can be transmitted to the data processing system in real time via IoT devices.

[0062] Specifically, during the manual verification result statistics phase, a time-dimensional statistical period is first established as the data acquisition window. Within this period, when an observation point is marked as a verification point, multi-dimensional environmental parameter data corresponding to that moment are collected simultaneously. For example, when an observation point is marked due to a dispute over the test results, the system automatically records the environmental conditions such as temperature and humidity at that time. By establishing a time correlation between environmental parameters and verification events, a dataset containing environmental factors is formed, providing basic data support for subsequent analysis of the degree to which different testing methods are affected by the environment.

[0063] Step P2: environmental correlation analysis on artificial verification result: when the visual settlement detection process is marked as a correct process, the corresponding verification point is marked as a visual correlation point; when the settlement verification process is marked as a correct process, the corresponding verification point is marked as a verification correlation point; data cleaning processing is performed on the visual correlation point: the detection value JCi of the environmental parameter i of all visual correlation points forms a visual set i of the environmental parameter i, the variance of all elements of the visual set i is calculated to obtain the visual central value SJi of the environmental parameter i, the visual central value SJi is compared with the preset central threshold JZi: if the visual central value SJi is greater than or equal to the central threshold JZi, the maximum element and the minimum element in the visual set i are removed, and then the visual central value SJi is recalculated, and the process is repeated until the visual central value SJi is less than the central threshold JZi; if the visual central value SJi is less than the central threshold JZi, the maximum element and the minimum element retained in the visual set i form a visual priority range SYi of the environmental parameter i; the same method is used to perform data cleaning processing on the verification correlation point to obtain a verification priority range HYi of the environmental parameter i;

[0064] The visual correlation point refers to the corresponding verification point when the visual settlement detection process is marked as a correct process, which can be realized by correlating the correct detection result with the verification point coordinates, and is used for subsequent analysis of the environmental parameter range suitable for the visual detection method. The verification correlation point refers to the corresponding verification point when the settlement verification process is marked as a correct process, which can be realized by correlating the correct verification result with the verification point coordinates, and is used for analyzing the environmental parameter range suitable for the verification method. The data cleaning processing refers to filtering the detection data by variance calculation and outlier removal, which can be realized by using an iterative variance calculation method, and is used for eliminating noise data in the environmental parameter. The visual priority range SYi refers to the reasonable interval of the environmental parameter corresponding to the visual correlation point after data cleaning, which can be realized by retaining the maximum and minimum detection values after data cleaning to form a range interval, and is used to represent the environmental conditions suitable for the visual detection method. The verification priority range HYi refers to the reasonable interval of the environmental parameter corresponding to the verification correlation point after data cleaning, which can be realized by using the same construction method as the visual priority range, and is used to represent the environmental conditions suitable for the settlement verification method.

[0065] Specifically, when the visual settlement detection process is verified to be correct, the corresponding verification point is classified as a visual correlation point, and its environmental parameter data is included in the visual set for cleaning. By calculating the variance of the environmental parameters in the visual set, if the variance exceeds the preset threshold, the maximum and minimum detection values are removed, and the process is repeated until the variance meets the requirements. The final retained data interval is defined as the visual priority range SYi. For the verification correlation point, the same data cleaning logic is used to generate the verification priority range HYi. Thus, in subsequent decision-making processes, the corresponding detection results can be automatically selected as valid basis based on the priority range of the environmental parameters.

[0066] Step P3: generating an immediate decision result: when the settlement detection result of the subsequent observation point is controversial, the detection value JCi of the environmental parameter i of the observation point is obtained, and it is determined whether the detection value JCi of the environmental parameter i is located within the corresponding visual priority range SYi: if yes, the visual settlement detection result in step S3 is sent to the server; if no, it is determined whether the detection value JCi of the environmental parameter i is located within the corresponding verification priority range HYi: if yes, the settlement verification result in step S4 is sent to the server; if no, an artificial verification signal is generated and sent to the server;

[0067] When the visual detection and verification results of the observation point are inconsistent, the system automatically extracts the current environmental parameters and compares them with the priority range formed by the historical verification. If all the environmental parameters are within the visual priority range, it indicates that the current environment meets the applicable conditions of visual detection, and the visual detection result is preferred. If the visual conditions are not met but the verification priority range is met, the verification result is used as the final decision. When the environmental parameters exceed both priority ranges, the system automatically triggers the artificial verification process to avoid misjudgment that may be caused by using a single detection result in an abnormal environment.

[0068] A building foundation settlement detection method based on computer vision, in which, during operation, a detection and verification sub-method performs visual settlement detection and analysis on the building foundation, evaluates the effectiveness of the detection result, performs risk judgment and verification analysis, and finally outputs a comprehensive detection result. An immediate decision sub-method establishes an association model of detection methods and environmental conditions by statistically analyzing the environmental parameters in the artificial verification records, and automatically selects the optimal detection result output method in subsequent controversial scenarios.

[0069] The above content is merely an example and description of the structure of the present application. Those skilled in the art can make various modifications or supplements to the described specific embodiments or use similar ways to replace them, as long as they do not deviate from the structure of the present application or exceed the scope defined by the present claims, and they should belong to the protection scope of the present application.

[0070] In the description of the specification, reference to "one embodiment", "an example", "a specific example" or the like means that a particular feature, structure, material or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the application. The appearances of the phrases "in one embodiment", "an example", "a specific example" or the like in various places in the specification are not necessarily referring to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0071] The preferred embodiments of the application disclosed above are only to help explain the application. The preferred embodiments do not describe all the details of the application and limit the application to the specific embodiments. Obviously, many modifications and variations can be made in light of the contents of the specification. The specification selects and specifically describes these embodiments in order to better explain the principles and practical application of the application, so that those skilled in the art can well understand and utilize the application. The application is limited only by the claims and their full scope and equivalents.

Claims

1. A computer vision-based building foundation settlement detection method, characterized by, The method comprises a detection and verification sub-method and an instant decision sub-method. The detection and verification sub-method comprises the following steps: Step S1: visually detecting and analyzing the settlement of the building foundation; Step S2: analyzing the validity of the visual settlement detection result of the building foundation observation point; Step S3: analyzing the visual settlement risk of the building foundation observation point; Step S4: analyzing the settlement verification of the building foundation observation point; Step S5: outputting the settlement detection result of the building foundation observation point; The instant decision sub-method comprises the following steps: Step P1: statistically analyzing the artificial verification result; Step P2: analyzing the environmental correlation of the artificial verification result; Step P3: generating an instant decision result; In step S1, the specific process of the visual settlement detection analysis comprises: arranging a camera array at the observation point of the building, regularly taking images of the observation point by the camera, marking the taken images as detection images, enlarging the detection images into pixel grid images, extracting the coordinates of the two-dimensional code targets in the detection images, marking the distance between the two-dimensional code target coordinate points extracted by the same camera in the last two times as the settlement detection value of the camera, and marking the distance between the two-dimensional code target coordinate points extracted by the same camera in the last two times as the settlement detection value of the camera. In step S2, the specific process of the visual settlement detection result validity analysis comprises: calculating the variance of the settlement detection values collected by all cameras in the camera array of the same observation point to obtain the effective coefficient of the observation point, comparing the effective coefficient with a preset effective threshold value, determining that the visual settlement detection result validity of the observation point meets the requirements if the effective coefficient is less than the effective threshold value, summing and averaging the settlement detection values collected by all cameras in the camera array of the observation point to obtain the visual settlement value, and determining that the visual settlement detection result validity of the observation point does not meet the requirements if the effective coefficient is greater than or equal to the effective threshold value, generating a camera maintenance signal and sending the camera maintenance signal to the mobile terminal of the management personnel. In step P2, the specific process of the environmental correlation analysis comprises: marking the corresponding verification point as a visual correlation point when the visual settlement detection process is marked as a correct process, marking the corresponding verification point as a verification correlation point when the settlement verification process is marked as a correct process, performing data cleaning processing on the visual correlation point to obtain the visual priority range SYi of the environmental parameter i, and performing data cleaning processing on the verification correlation point in the same way to obtain the verification priority range HYi of the environmental parameter i. In step P2, the specific process of data cleaning of the visual correlation point includes: constructing a visual set i of the environmental parameter i by the detection value JCi of the environmental parameter i of all visual correlation points, calculating the visual central value SJi of the visual set i by variance calculation of all elements of the visual set i, comparing the visual central value SJi with the preset central threshold JZi: if the visual central value SJi is greater than or equal to the central threshold JZi, the maximum element and the minimum element in the visual set i are removed, and then the visual central value SJi is recalculated, and the process is repeated until the visual central value SJi is less than the central threshold JZi; if the visual central value SJi is less than the central threshold JZi, the maximum element and the minimum element retained in the visual set i are used to construct a visual priority range SYi of the environmental parameter i.

2. The computer vision-based building foundation settlement detection method of claim 1, wherein, In step S3, the specific process of visual subsidence risk analysis includes: comparing the visual subsidence value of the observation point with the preset subsidence threshold: if the visual subsidence value is less than the subsidence threshold, it is determined that the observation point has no visual subsidence risk; if the visual subsidence value is greater than or equal to the subsidence threshold, it is determined that the observation point has visual subsidence risk.

3. The computer vision-based building foundation settlement detection method of claim 2, wherein, In step S4, the specific process of subsidence verification analysis includes: constructing a detection period by the time of the last two image acquisitions of the camera, obtaining the horizontal angle change amount of the observation point in the detection period by the inclinometer and marking it as a subsidence verification value, comparing the subsidence verification value with the preset subsidence verification threshold: if the subsidence verification value is less than the subsidence verification threshold, it is determined that the observation point has no verification subsidence risk; if the subsidence verification value is greater than or equal to the subsidence verification threshold, it is determined that the observation point has verification subsidence risk.

4. The computer vision-based building foundation settlement detection method of claim 3, wherein, In step S5, the specific process of subsidence detection result output includes: if the observation point has both visual subsidence risk and verification subsidence risk, it is determined that the subsidence detection result of the observation point is unqualified, a subsidence warning signal is generated and sent to the server; if the observation point has no visual subsidence risk and verification subsidence risk, it is determined that the subsidence detection result of the observation point is qualified, a detection qualified signal is generated and sent to the server; otherwise, it is determined that the subsidence detection result of the observation point is controversial, the corresponding observation point is marked as a verification point, an artificial verification signal is generated and sent to the server; after receiving the artificial verification signal, the server sends the artificial verification signal to the mobile terminal of the management personnel, and the management personnel performs artificial verification on the observation point, and marks the visual subsidence detection process or the subsidence verification process as a correct process.

5. The computer vision-based building foundation settlement detection method of claim 4, wherein, In step P1, a statistical period is generated, and the detection value JCi of the environmental parameter i of the verification point at the time of marking in the statistical period is obtained, i=1, 2, …, n, n is a positive integer, and the environmental parameter i includes the air temperature value, the air humidity value, the fog concentration value, the rainfall, the wind grade of the verification point.

6. The computer vision-based building foundation settlement detection method of claim 5, wherein, In step P3, the generation process of the instant decision result comprises: when the settlement detection result of the subsequent observation point is controversial, obtaining the detection value JCi of the environmental parameter i of the observation point, determining whether the detection value JCi of the environmental parameter i is all located within the corresponding visual priority range SYi: if yes, the visual settlement detection result in step S3 is sent to the server; if not, determining whether the detection value JCi of the environmental parameter i is all located within the corresponding verification priority range HYi: if yes, the settlement verification result in step S4 is sent to the server; if not, an artificial verification signal is generated and sent to the server.

Images