Building structure earthquake infinitesimal displacement measurement method, design and medium

Abstract

The invention discloses a building structure earthquake infinitesimal displacement measuring method, design and medium, and relates to the technical field of building structure monitoring. The invention aims to solve the problem that the precision and the application scene of the existing video measurement method are restricted. According to the method, feature points of the surface and the ground of a building structure are selected from a pre-earthquake image shot by a monitoring camera, a region of interest is cut according to the feature points, and the monitoring camera is fixed to the building structure; tracking the position of a region of interest in each frame of epicentral image shot by the monitoring camera, and taking the central point of the region of interest in the epicentral image as the feature point of the epicentral image; and according to a camera projection matrix, converting the coordinates of the feature points of the epicenter images from a two-dimensional pixel coordinate system to a three-dimensional world coordinate system, and further generating a displacement time history curve according to the time sequence of each frame of epicenter image.

Description

Technical Field

[0001] This application belongs to the field of building structure monitoring technology, and in particular relates to the measurement of minute displacements. Background Technology

[0002] Structural damage assessment is a crucial step in earthquake disaster evaluation. Under seismic loading, the inter-story drift angle directly correlates with the degree of component damage, while the apex displacement reflects overall stability. China's "Code for Seismic Design of Buildings" uses displacement limits as a safety criterion. However, traditional contact sensors have significant drawbacks in practical applications: they require drilling and pre-embedding cables, making installation difficult on high-rise building roofs and large-span structures; they also suffer from low reliability and poor economic efficiency, making it difficult to meet the needs of rapid post-earthquake response and full-field monitoring.

[0003] Vision-based monitoring technology offers a breakthrough solution. This technology directly reuses existing security cameras, acquiring structural displacement non-contactly, avoiding the challenges of building damage and high-level node monitoring. It requires only conventional electricity and can be remotely maintained via a monitoring platform. However, existing video measurement methods are limited by accuracy and application scenarios. Summary of the Invention

[0004] This application aims to address the limitations of existing video measurement methods in terms of accuracy and application scenarios. It provides a method for measuring minute displacements of building structures during earthquakes, which can overcome camera shake interference and achieve sub-pixel accuracy displacement measurement even in complex vibration environments, thereby promoting the transition of visual measurement technology from the laboratory to practical engineering applications.

[0005] The first aspect of this application provides a method for measuring minute seismic displacements of building structures, including:

[0006] Feature points on the building structure surface and ground are selected from the pre-earthquake images captured by the monitoring camera, and the region of interest is cropped based on the feature points. The monitoring camera is fixed on the building structure.

[0007] The location of the region of interest in each frame of epicenter image captured by the monitoring camera is tracked, and the center point of the region of interest in the epicenter image is used as the feature point of the epicenter image;

[0008] The feature point coordinates of the epicenter image are transformed from a two-dimensional pixel coordinate system to a three-dimensional world coordinate system based on the camera projection matrix, and then a displacement time history curve is generated based on the time sequence of each frame of the epicenter image.

[0009] In one possible design, selecting feature points on the building structure surface and ground from pre-earthquake images captured by monitoring cameras, and cropping the region of interest based on these feature points, includes:

[0010] In the pre-earthquake images captured by the monitoring camera, a pre-selected region of interest is selected for the building structure surface and the ground according to actual needs;

[0011] The Shi-Tomasi corner detection algorithm was used to identify feature points in pre-selected regions of interest on the building structure surface and the ground, respectively.

[0012] Centered on the selected feature points, regions of interest are generated for the building structure surface and the ground according to actual needs.

[0013] In one possible design, the Shi-Tomasi corner detection algorithm is used to identify feature points in pre-selected regions of interest on the building structure surface and the ground, respectively, including:

[0014] The feature value of each pixel in the pre-selected region of interest of the building structure surface and the ground is calculated respectively. The pixel with the largest feature value is taken as the feature point to obtain the feature points of the building structure surface and the ground in the pre-earthquake image.

[0015] In one possible design, tracking the location of the region of interest in each frame of epicenter image captured by the monitoring camera includes:

[0016] Subpixel template matching technology was used to track the regions of interest on the surface of building structures and the ground in each frame of the epicenter image.

[0017] In one possible design, the use of subpixel template matching technology to track the regions of interest on the building structure surface and the ground in each frame of the epicenter image includes:

[0018] Each frame of the epicenter image is converted to grayscale.

[0019] A normalized cross-correlation algorithm is used for template matching to obtain the integer pixel coordinates of the region of interest;

[0020] Subpixel-level coordinates of the region of interest are calculated using quadratic polynomial interpolation.

[0021] In one possible design, the method for obtaining the camera projection matrix includes:

[0022] The monitoring camera was calibrated using the direct linear transformation calibration method to obtain the camera projection matrix.

[0023] In one possible design, the transformation of the feature point coordinates of the epicenter image from a two-dimensional pixel coordinate system to a three-dimensional world coordinate system based on the camera projection matrix includes:

[0024] The following formula is used to transform the feature points of the ground in the epicenter image from the two-dimensional pixel coordinate system to the three-dimensional world coordinate system:

[0025] ;

[0026] The following formula is used to transform the feature points of building structures in the epicenter image from a two-dimensional pixel coordinate system to a three-dimensional world coordinate system:

[0027] ;

[0028] In the formula, and The epicenters are respectively in a three-dimensional world coordinate system. Homogeneous coordinates of feature points on the ground and building structure surfaces at any given time. and The epicenters are respectively in a two-dimensional world coordinate system. Homogeneous coordinates of feature points on the ground and building structure surfaces at any given time. As a scale factor, For camera projection matrix, and These represent the number of rows and columns of the camera projection matrix, respectively.

[0029] In one possible design, the generation of displacement time history curves based on the temporal sequence of each frame of epicenter image includes:

[0030] Calculate the vibration response of the ground and building structure surfaces:

[0031] ,

[0032] ,

[0033] In the formula, and Pre-earthquake conditions in three-dimensional world coordinate system The coordinates of feature points on the ground and building structure surfaces at any given time. For the ground vibration response, This refers to the vibration response of the building structure relative to the ground.

[0034] Displacement time history curves are generated by arranging the ground and building surface vibration responses corresponding to each frame of the epicenter image in chronological order.

[0035] The second aspect of this application provides a building structure seismic micro-displacement measurement device, which includes a processor and a memory. The memory stores at least one instruction, which is loaded and executed by the processor to implement the building structure seismic micro-displacement measurement method described above.

[0036] A third aspect of this application provides a computer storage medium storing at least one instruction, which is loaded and executed by a processor to implement the above-described method for measuring micro-displacement of building structures during earthquakes.

[0037] The beneficial effects of this application are:

[0038] 1. Solving the challenge of accurate measurement of minute displacements under camera shaking conditions: By establishing a relative motion calculation model of "camera-ground-building", the effective separation of the camera's own motion and the building structure's response is achieved. This method does not require the assumption of an absolutely stationary camera and can be directly applied to real-world scenarios where the camera and structure shake synchronously under seismic conditions.

[0039] 2. Breakthrough multi-target synchronous processing and measurement accuracy: By tracking ground static reference points and building feature points separately, and integrating subpixel interpolation and DLT calibration technology, the measurement accuracy of building displacement is improved to 0.1 pixel level while solving camera motion, solving the problem of accurate measurement of subpixel-level micro-displacements in long-distance videos.

[0040] 3. Innovative Technology Integration and Automated Processing: The system integrates core technologies such as DLT calibration, dual-region feature detection, and sub-pixel template matching, constructing a technical system of "calibration-detection-displacement separation-precise tracking." This system can automatically select optimal feature points, simultaneously calculate camera and structural displacements, and complete precise mapping in three-dimensional physical space, forming a complete solution from complex image sequences to engineering-usable data.

[0041] 4. Outstanding engineering practicality and economy: This application makes full use of the existing urban security surveillance camera network, with no stringent requirements on camera installation conditions, and no need for additional shockproof reinforcement or special sensors. It enables widely deployed ordinary monitoring equipment to have structural health monitoring functions, providing a universal, economical and efficient solution for the security monitoring of large building complexes. Attached Figure Description

[0042] Figure 1 Diagram of the DLT calibration method calculation model

[0043] Figure 2 A flowchart for a method of measuring minute displacements of building structures during earthquakes;

[0044] Figure 3 The flowchart for the Blender-based simulation experiment;

[0045] Figure 4 Modeling layout and renderings for Blender simulation experiments;

[0046] Figure 5The figure shows the displacement calculation results under the large PGA condition in the Blender simulation experiment;

[0047] Figure 6 The figure shows the displacement calculation results under small PGA conditions in the Blender simulation experiment;

[0048] Figure 7 This is a graph showing the error coefficients of the displacement calculation results from the Blender simulation experiment. Detailed Implementation

[0049] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application. It should be noted that, unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other.

[0050] Precision limitations. Although the displacement amplitude of a building structure in actual physical space may reach the centimeter level, due to the geometric attenuation effect caused by the imaging distance, its projected displacement on the image plane often manifests as a sub-pixel level change (<1 pixel). This displacement-pixel conversion relationship at the imaging scale makes it difficult for traditional vision algorithms based on integer pixel matching to achieve accurate measurement.

[0051] Idealized assumptions for application scenarios. Most existing visual measurement methods rely on a crucial premise: the monitoring camera must remain absolutely stationary during the measurement process. This assumption is extremely difficult to meet in real-world earthquake monitoring scenarios. Cameras are typically mounted on the ground or adjacent structures, and during an earthquake, these supports themselves vibrate, causing the camera to sway accordingly. This coupled motion between the camera and the structure means that the pixel displacement in the image sequence simultaneously includes the displacement of the building target and the camera's own displacement, leading to significant deviations or even complete failure of the measurement results.

[0052] The essence of the camera imaging process is to convert points in the three-dimensional world coordinate system. Points projected into a two-dimensional pixel coordinate system The camera projection matrix establishes this projection relationship of point coordinates from three dimensions to two dimensions, as shown in equation (1):

[0053] (1),

[0054] In the formula, For point Homogeneous coordinates in a two-dimensional pixel coordinate system; Scale factor; For camera projection matrix; For point Homogeneous coordinates in a three-dimensional world coordinate system; For camera projection matrix The first in Yuan, This indicates the row number of the camera projection matrix. This indicates the number of columns in the camera projection matrix.

[0055] Figure 1 The text provides details about the events before the earthquake. Time and Earthquake Action At any given moment, the positional relationship between the ground and the building relative to the outdoor monitoring camera is determined. The reference object is set as the monitoring camera, rigidly connected to the exterior and structure of the monitored building. A three-dimensional coordinate system XYZ is established, ensuring it remains stationary relative to the camera coordinate system, with the ground plane coinciding with the XY plane. At a certain moment under seismic action... The building exhibits a two-way displacement response. Assume the height of the key point on the top floor relative to the ground plane (XY). Approximately unchanged.

[0056] At any given time, feature points on the ground in the three-dimensional XYZ coordinate system Key points of the building's top floor After being imaged by the camera, the pixel coordinates of the projected points in the two-dimensional UV pixel coordinate system are as follows: , .

[0057] According to equation (1), the projection relationship between ground feature points in the two-dimensional pixel coordinate system and the three-dimensional coordinate system at any time can be expressed by equation (2), and the projection relationship of key points on the top floor of the building can be expressed by equation (3):

[0058] (2),

[0059] (3).

[0060] From equations (2) and (3), the coordinates of the ground feature points and the key points on the top floor of the building in the three-dimensional coordinate system can be obtained, specifically expressed as equations (4) and (5):

[0061] (4),

[0062] (5),

[0063] In the formula, This is a second-order minor of the camera matrix. , and Each represents the row number of the camera projection matrix. and Both represent the column number of the camera projection matrix. , .

[0064] Time and At any given moment, the difference between the X-axis and Y-axis coordinates of the ground feature points in the three-dimensional coordinate system represents the displacement response of the camera relative to the ground in the X and Y directions, respectively, as shown in Equation (6):

[0065] (6),

[0066] Similarly, the seismic response of buildings relative to the ground and It can be expressed as equation (7):

[0067] (7).

[0068] In view of this, in order to address the challenge of accurately measuring the subpixel-level changes in structural displacement on the image plane under conditions of camera self-shaking in outdoor long-distance video monitoring, this application provides a method for measuring the micro-displacement of building structures in seismic events based on subpixel template matching. By establishing a relative motion solution model of camera-ground-building, and separating the camera's own displacement, the method achieves subpixel-level accurate measurement of the actual relative displacement of the building structure.

[0069] Specific implementation method one: Refer to Figure 2 This embodiment specifically describes the method for measuring minute seismic displacements of building structures, which includes:

[0070] Feature points on the building structure surface and ground are selected from the pre-earthquake images captured by the monitoring camera, and the region of interest is cropped based on the feature points. The monitoring camera is fixed on the building structure.

[0071] The location of the region of interest in each frame of epicenter image captured by the monitoring camera is tracked, and the center point of the region of interest in the epicenter image is used as the feature point of the epicenter image;

[0072] The feature point coordinates of the epicenter image are transformed from a two-dimensional pixel coordinate system to a three-dimensional world coordinate system based on the camera projection matrix, and then a displacement time history curve is generated based on the time sequence of each frame of the epicenter image.

[0073] In one implementation, the step of selecting feature points on the building structure surface and ground in the pre-earthquake image captured by the monitoring camera, and cropping the region of interest based on the feature points, includes:

[0074] In the pre-earthquake images captured by the monitoring camera, a pre-selected region of interest is selected for the building structure surface and the ground according to actual needs;

[0075] The Shi-Tomasi corner detection algorithm was used to identify feature points in pre-selected regions of interest on the building structure surface and the ground, respectively.

[0076] Centered on the selected feature points, regions of interest are generated for the building structure surface and the ground according to actual needs.

[0077] In one embodiment, the step of identifying feature points using the Shi-Tomasi corner detection algorithm in pre-selected regions of interest on the building structure surface and the ground includes:

[0078] The feature value of each pixel in the pre-selected region of interest of the building structure surface and the ground is calculated respectively. The pixel with the largest feature value is taken as the feature point to obtain the feature points of the building structure surface and the ground in the pre-earthquake image.

[0079] In one implementation, tracking the location of the region of interest in each frame of epicenter image captured by the monitoring camera includes:

[0080] Subpixel template matching technology was used to track the regions of interest on the surface of building structures and the ground in each frame of the epicenter image.

[0081] In one implementation, the step of using sub-pixel template matching technology to track the regions of interest on the surface of building structures and the ground in each frame of the epicenter image includes:

[0082] Each frame of the epicenter image is converted to grayscale.

[0083] A normalized cross-correlation algorithm is used for template matching to obtain the integer pixel coordinates of the region of interest;

[0084] Subpixel-level coordinates of the region of interest are calculated using quadratic polynomial interpolation.

[0085] In one embodiment, the method for obtaining the camera projection matrix includes:

[0086] The monitoring camera was calibrated using the direct linear transformation calibration method to obtain the camera projection matrix.

[0087] In one implementation, the step of transforming the feature point coordinates of the epicenter image from a two-dimensional pixel coordinate system to a three-dimensional world coordinate system based on the camera projection matrix includes:

[0088] The following formula is used to transform the feature points of the ground in the epicenter image from the two-dimensional pixel coordinate system to the three-dimensional world coordinate system:

[0089] ;

[0090] The following formula is used to transform the feature points of building structures in the epicenter image from a two-dimensional pixel coordinate system to a three-dimensional world coordinate system:

[0091] ;

[0092] In the formula, and The epicenters are respectively in a three-dimensional world coordinate system. Homogeneous coordinates of feature points on the ground and building structure surfaces at any given time. and The epicenters are respectively in a two-dimensional world coordinate system. Homogeneous coordinates of feature points on the ground and building structure surfaces at any given time. As a scale factor, For camera projection matrix, and These represent the number of rows and columns of the camera projection matrix, respectively.

[0093] In one implementation, generating the displacement time history curve based on the temporal sequence of each frame of epicenter image includes:

[0094] Calculate the vibration response of the ground and building structure surfaces:

[0095] ,

[0096] ,

[0097] In the formula, and Pre-earthquake conditions in three-dimensional world coordinate system The coordinates of feature points on the ground and building structure surfaces at any given time. For the ground vibration response, This refers to the vibration response of the building structure relative to the ground.

[0098] Displacement time history curves are generated by arranging the ground and building surface vibration responses corresponding to each frame of the epicenter image in chronological order.

[0099] To further illustrate the implementation scheme of this application, Figure 2 A method for measuring minute seismic displacements of building structures is provided, comprising steps one through five. The numbering of these steps does not necessarily preclude their sequential execution. Each step is described in detail below:

[0100] Step 1: Camera calibration and spatial matrix determination.

[0101] During the pre-earthquake preparation phase, the Direct Linear Transform (DLT) calibration method was used to calibrate the monitoring cameras. Multiple sets of calibration points with known spatial coordinates were placed on the building structure surface and nearby ground, ensuring that these points were uniformly distributed and non-coplanar within the camera's field of view. Images of the calibration points were captured by the monitoring cameras, and their two-dimensional pixel coordinates were extracted. The DLT algorithm was used to calculate the camera's projection matrix, ensuring that the projection error was less than a 0.1 pixel threshold, thereby establishing a precise mapping relationship between the camera's imaging plane and the actual spatial coordinates.

[0102] In this embodiment, the DLT calibration method establishes a mapping relationship between the camera imaging plane and the actual spatial coordinates through at least six known spatial coordinates that are not coplanar, thereby completing the camera calibration.

[0103] Step 2: Dual-region feature point detection and ROI (Region of Interest) generation

[0104] In the surveillance footage captured before the earthquake, pre-selected ROI regions were chosen for the building structure surface and the ground. The Shi-Tomasi corner detection algorithm was used to identify high-stability feature points on the building structure surface and the ground. Two independent ROI regions were generated by cropping the selected feature points as the center. These regions were used to track the displacement of the building target and the displacement of the ground reference point reflecting the camera's own motion, respectively.

[0105] In this embodiment, the Shi-Tomasi corner detection algorithm calculates the feature value of each pixel in the image and selects the point with the largest feature value as the most representative feature point. The size of the ROI region is determined according to the actual monitoring requirements, and it must be ensured that it contains sufficient feature information to improve matching accuracy.

[0106] Step 3: Epicenter Video Acquisition and Composite Motion Recording

[0107] When an earthquake occurs, monitoring cameras are activated to record video. The video, recorded by the cameras which may shake, fully documents the combined motion of the building structure and ground reference points. Considering that the cameras may shake along with their installation foundations, the video will simultaneously acquire a sequence of combined motion images of feature points on the building structure surface and on the ground, providing a data basis for subsequent separation of camera motion and structural response.

[0108] Step 4: Dual-target matching and sub-pixel coordinate calculation

[0109] In the post-earthquake analysis phase, the monitoring video was converted into an image sequence frame by frame. The post-earthquake video was analyzed frame by frame. Subpixel template matching technology was used to track the positional changes of the building structure surface and the ground ROI area in each frame of the image. Integer pixel coordinates of building feature points and ground feature points were obtained to obtain the precise subpixel coordinate sequence of feature points.

[0110] Subpixel template matching technology is achieved through the following steps:

[0111] Each frame of the image is converted to grayscale. The integer pixel coordinates are obtained by template matching using the normalized cross-correlation (NCC) algorithm. Then, the sub-pixel coordinates of the matching position are calculated using the quadratic polynomial interpolation method.

[0112] Step 5: Relative motion calculation and displacement separation

[0113] Accurately calculate the pixel displacement of ground feature points (reflecting the camera's own motion) and the pixel displacement of building feature points (reflecting the combined motion of the building and the camera).

[0114] Based on the calibrated camera matrix, the camera motion is calculated using the displacement data of ground feature points.

[0115] The true two-way seismic response (offset) of the building relative to the ground is separated from the composite motion of the building feature points. The two-way seismic response of the top floor of the building is obtained, displacement time history curves are generated, and structural seismic response data are output.

[0116] This embodiment proposes a three-stage technical solution: First, a projection matrix is ​​established through DLT calibration to achieve accurate conversion of pixel displacement to three-dimensional physical quantities. Then, the Shi-Tomasi algorithm is used to extract highly stable feature points from the building surface and the ground to ensure the traceability of features during vibration. Finally, the sub-pixel template matching algorithm is combined to improve the tracking accuracy to the 0.1 pixel level.

[0117] To verify the accuracy of the proposed method, this embodiment planned a simulation experiment based on Blender software for verification:

[0118] First, two outdoor building structure models, 28m apart, were created in Blender software. A monitoring device was strategically installed on one of the buildings, with a focal length of 9mm, a frame rate of 50fps, and a resolution of 2K (2560×1440). One pixel in the image represents approximately 0.05m of actual distance. Next, Ansys software was used to generate seismic data to obtain displacement information for the building and monitoring device. Then, the acquired displacement data was accurately imported into Blender using the Python-API interface. Afterward, Blender realistically simulated the motion of the monitoring device and building structure under earthquake trajectories. Finally, the video recorded by the monitoring device during the motion was exported for subsequent analysis and research. Since cameras typically only capture a single side of a building when shooting from a distance, and structural damage assessments often focus on lateral displacement, the simulation experiment only demonstrated lateral displacement.

[0119] Reference Figure 2 The specific steps of the simulation experiment are explained in detail, including:

[0120] Simulation test step one: Build a 3D building model in the Blender environment and set detailed parameters for the camera, including key indicators such as focal length, frame rate and resolution;

[0121] Step 2 of the simulation experiment: Simulate a real outdoor environment, assign texture materials to the ground and structure, and apply texture maps to the buildings and the ground.

[0122] Step 3 of the simulation test: Use Ansys software to generate ground motion data, input the ground motion data and camera displacement data into Blender, and simulate the seismic motion trajectory of the camera and building structure.

[0123] Simulation test step four: Render the video using Blender and output the outdoor simulated surveillance video;

[0124] Step 5 of the simulation test: Export the monitoring video recorded during the movement of the camera and structure.

[0125] The layout and camera view of the Blender simulation experiment are as follows: Figure 3 As shown; to highlight the effectiveness of the proposed method, three groups of larger PGAs are first set so that the peak pixel displacement is greater than 1 pixel, and the LK optical flow method and integer pixel matching method are compared with this algorithm, as shown. Figure 5 As shown. Since the accuracy of the integer pixel matching method is low at larger PGAs, while the LK optical flow method and this algorithm maintain good accuracy, four more smaller PGAs are set to ensure the structural peak pixel displacement is less than 1 pixel, further testing the accuracy of this algorithm, and using the optical flow method for comparison. A coefficient of determination is used. Peak relative error Root mean square error Correlation coefficient with Pearson The monitoring error was quantitatively assessed and calculated using formulas (8)-(11):

[0126] (8),

[0127] (9),

[0128] (10)

[0129] (11),

[0130] In the formula, The number of sampling points for the earthquake response time history. and They are respectively True values ​​and visual measurements of the seismic response at any given time. This represents the average of the true values ​​of the seismic response. and These are the true value and visual measurement of the peak seismic response, respectively.

[0131] The displacement time history comparison diagram between this embodiment and the traditional optical flow method obtained from the simulation experiment is shown below. Figure 4 The error coefficient comparison chart is as follows: Figure 5 As shown.

[0132] The results show that when the peak pixel displacement of the structure is less than 1 pixel, the relative error of the peak structural displacement obtained in this embodiment is... Stable within 5%, coefficient of determination Correlation coefficient with Pearson The accuracy remained stable above 0.9, and the root mean square error consistently outperformed the traditional optical flow method. This indicates that the proposed embodiment is feasible for monitoring minute displacements in the seismic response of building structures at the sub-pixel level, and exhibits good monitoring accuracy in both the time domain and at the peak of the seismic response.

[0133] Specific Implementation Method Two: The building structure seismic micro-displacement measurement device described in this embodiment includes a processor and a memory. The memory stores at least one instruction, which is loaded and executed by the processor to implement the building structure seismic micro-displacement measurement method as described in Specific Implementation Method One.

[0134] Specific Implementation Method 3: A computer storage medium storing at least one instruction, which is loaded and executed by a processor to implement the building structure seismic micro-displacement measurement method as described in Specific Implementation Method 1.

[0135] While specific embodiments of this application have been described herein with reference to them, it should be understood that these embodiments are merely examples of the principles and applications of this application. Therefore, it should be understood that many modifications can be made to the exemplary embodiments, and other arrangements can be designed without departing from the spirit and scope of this application as defined by the appended claims. It should be understood that different dependent claims and features described herein can be combined in ways different from those described in the original claims. It is also understood that features described in conjunction with individual embodiments can be used in other described embodiments.

Claims

1. A method for measuring minute seismic displacements of building structures, characterized in that, include: Feature points on the building structure surface and ground are selected from the pre-earthquake images captured by the monitoring camera, and the region of interest is cropped based on the feature points. The monitoring camera is fixed on the building structure. The location of the region of interest in each frame of epicenter image captured by the monitoring camera is tracked, and the center point of the region of interest in the epicenter image is used as the feature point of the epicenter image; The feature point coordinates of the epicenter image are transformed from a two-dimensional pixel coordinate system to a three-dimensional world coordinate system based on the camera projection matrix, and then a displacement time history curve is generated based on the time sequence of each frame of the epicenter image.

2. The method for measuring minute seismic displacements of building structures according to claim 1, characterized in that, The step of selecting feature points on the building structure surface and ground in the pre-earthquake images captured by the monitoring camera, and cropping the region of interest based on the feature points, includes: In the pre-earthquake images captured by the monitoring camera, a pre-selected region of interest is selected for the building structure surface and the ground according to actual needs; The Shi-Tomasi corner detection algorithm was used to identify feature points in pre-selected regions of interest on the building structure surface and the ground, respectively. Centered on the selected feature points, regions of interest are generated for the building structure surface and the ground according to actual needs.

3. The method for measuring minute seismic displacements of building structures according to claim 2, characterized in that, The Shi-Tomasi corner detection algorithm is used to identify feature points in pre-selected regions of interest on the building structure surface and the ground, including: The feature value of each pixel in the pre-selected region of interest of the building structure surface and the ground is calculated respectively. The pixel with the largest feature value is taken as the feature point to obtain the feature points of the building structure surface and the ground in the pre-earthquake image.

4. The method for measuring minute seismic displacements of building structures according to claim 1, characterized in that, The tracking of the location of the region of interest in each frame of epicenter image captured by the monitoring camera includes: Subpixel template matching technology was used to track the regions of interest on the surface of building structures and the ground in each frame of the epicenter image.

5. The method for measuring minute seismic displacements of building structures according to claim 4, characterized in that, The method of using subpixel template matching to track the regions of interest on the surface of building structures and the ground in each frame of the epicenter image includes: Each frame of the epicenter image is converted to grayscale. A normalized cross-correlation algorithm is used for template matching to obtain the integer pixel coordinates of the region of interest; Subpixel-level coordinates of the region of interest are calculated using quadratic polynomial interpolation.

6. The method for measuring minute seismic displacements of building structures according to claim 1, characterized in that, The method for obtaining the camera projection matrix includes: The monitoring camera was calibrated using the direct linear transformation calibration method to obtain the camera projection matrix.

7. The method for measuring minute seismic displacements of building structures according to claim 6, characterized in that, The step of transforming the feature point coordinates of the epicenter image from a two-dimensional pixel coordinate system to a three-dimensional world coordinate system based on the camera projection matrix includes: The following formula is used to transform the feature points of the ground in the epicenter image from the two-dimensional pixel coordinate system to the three-dimensional world coordinate system: ； The following formula is used to transform the feature points of building structures in the epicenter image from a two-dimensional pixel coordinate system to a three-dimensional world coordinate system: ； In the formula, and The epicenters are respectively in a three-dimensional world coordinate system. Homogeneous coordinates of feature points on the ground and building structure surfaces at any given time. and The epicenters are respectively in a two-dimensional world coordinate system. Homogeneous coordinates of feature points on the ground and building structure surfaces at any given time. As a scale factor, For camera projection matrix, and These represent the number of rows and columns of the camera projection matrix, respectively.

8. The method for measuring minute seismic displacements of building structures according to claim 7, characterized in that, The step of generating displacement time history curves based on the time sequence of each frame of epicenter image includes: Calculate the vibration response of the ground and building structure surfaces: ， ， In the formula, and Pre-earthquake conditions in three-dimensional world coordinate system The coordinates of feature points on the ground and building structure surfaces at any given time. For the ground vibration response, This refers to the vibration response of the building structure relative to the ground. Displacement time history curves are generated by arranging the ground and building surface vibration responses corresponding to each frame of the epicenter image in chronological order.

9. A device for measuring minute displacements of building structures during earthquakes, characterized in that, The building structure seismic micro-displacement measurement device includes a processor and a memory, wherein the memory stores at least one instruction, which is loaded and executed by the processor to implement the building structure seismic micro-displacement measurement method as described in any one of claims 1 to 8.

10. A computer storage medium, characterized in that, The computer storage medium stores at least one instruction, which is loaded and executed by a processor to implement the method for measuring micro-displacement of building structures in seismic events as described in any one of claims 1 to 8.

Images