Foundation pit displacement monitoring method and system based on machine vision and monitoring equipment

Abstract

The invention relates to the technical field of foundation pit detection, and discloses a foundation pit displacement monitoring method and system based on machine vision and monitoring equipment. The method comprises the steps that a high-definition camera collects image data of a foundation pit area in real time, the collected image data are preprocessed, reference points in the image data are analyzed through an optical flow method, displacement of the reference points in a foundation pit is estimated, and three-dimensional position changes of the reference points are calculated through the stereoscopic vision technology and image matching. The system comprises an image acquisition module, an image preprocessing module, a displacement calculation module, a filtering optimization module, an anomaly detection and early warning module and a data storage module. The equipment comprises a high-definition monitoring camera, a data processing terminal, a storage unit, an alarm device and a communication module. According to the invention, through image acquisition of a multi-camera system and a stereoscopic vision technology, comprehensive monitoring of the displacement of the foundation pit is ensured, high-quality image data can be provided in various environments, and precise acquisition of three-dimensional depth information is realized in combination with stereoscopic vision.

Description

Technical Field

[0001] The present invention relates to the technical field of foundation pit monitoring, and specifically to a foundation pit displacement monitoring method, system and monitoring equipment based on machine vision. Background Art

[0002] A foundation pit is a common structure in construction, and is widely used in underground construction, tunnel excavation, mine exploitation and other fields. During the construction process of a foundation pit, displacement is an important indicator to measure the safety of the foundation pit. Foundation pit displacement includes horizontal displacement and vertical displacement, which are usually caused by factors such as foundation pit excavation, construction load, and groundwater change. Once the foundation pit displacement is too large, it may lead to the collapse of the foundation pit, damage to surrounding buildings, and even trigger major safety accidents. Timely and accurate monitoring of foundation pit displacement, especially the monitoring of tiny displacement, is the key to ensuring construction safety.

[0003] Existing foundation pit displacement monitoring methods are diverse, mainly including manual measurement, traditional sensor monitoring, and optical measurement. Manual measurement usually relies on total station or level to measure the foundation pit displacement regularly, which is common in small-scale construction or projects with limited budgets. Traditional sensors, such as displacement gauges and inclinometers, are also widely used in foundation pit displacement monitoring. Their working principle is to install sensors at key positions of the foundation pit to monitor displacement data in real time or regularly, and can provide relatively accurate displacement change information.

[0004] However, for existing foundation pit displacement monitoring methods, manual measurement requires frequent operations and is limited by personnel experience and working environment, which is prone to errors. Displacement gauges and inclinometers may be affected by temperature change, humidity change or equipment aging, resulting in a decrease in data reliability. And although optical measurement methods can obtain image data more comprehensively, due to their dependence on viewing angle limitations, the lack of depth information often leads to a decrease in the accuracy of monitoring results. Therefore, the present invention provides a foundation pit displacement monitoring method, system and monitoring equipment based on machine vision to solve the deficiencies existing in the prior art. Summary of the Invention

[0005] Aiming at the deficiencies of the prior art, the present invention provides a foundation pit displacement monitoring method, system and monitoring equipment based on machine vision, and solves the problems that for existing foundation pit displacement monitoring methods, manual measurement requires frequent operations and is limited by personnel experience and working environment, which is prone to errors, and displacement gauges and inclinometers may be affected by temperature change, humidity change or equipment aging, resulting in a decrease in data reliability.

[0006] To achieve the above purposes, the present invention is realized through the following technical solutions: A foundation pit displacement monitoring method based on machine vision includes the following steps: Deploy multiple high-definition cameras to collect image data of the foundation pit area in real time and transmit the image data to the central processing unit; preprocess the collected image data, and the preprocessing includes denoising and image correction to ensure the monitoring accuracy; Analyze the reference points in the image data based on the optical flow method, calculate the optical flow velocity components of each pixel point, and estimate the displacement of the reference points in the foundation pit; Obtain the image data of multiple high-definition cameras through stereo vision technology, perform three-dimensional reconstruction, calculate the three-dimensional positions of the reference points in the foundation pit and deduce their displacements; Use Kalman filtering and particle filtering to dynamically optimize the estimation of the displacement data of the reference points in the foundation pit, and trigger an early warning notification through anomaly detection.

[0007] Preferably, the step of preprocessing the collected image data includes denoising the image through Gaussian filtering and performing image registration to ensure the consistency between images.

[0008] Preferably, the steps of the optical flow method include calculating the optical flow velocity of each pixel point using the Lucas-Kanade algorithm, and the optical flow equation is: I x ·u + I y ·v + I t =0; where, I x , I y , I t are the gradients of the image in the x direction, y direction and time t respectively, and u and v are the velocity components of the pixel point in the x and y directions.

[0009] Preferably, the camera image data is obtained through stereo vision technology, and the stereo vision technology matches the corresponding points in the multi-camera images through the fundamental matrix F, satisfying the following relationship: where, x and y are the coordinates of the corresponding points in the two images respectively, represents the transpose, and F is the fundamental matrix, which is used to describe the geometric relationship between the two cameras.

[0010] Preferably, the three-dimensional reconstruction uses a stereo matching algorithm to calculate the depth information of the reference points in the foundation pit, and deduces the displacement through a three-dimensional reconstruction algorithm. The steps of the stereo matching algorithm: Shoot the same target from different angles through two cameras to obtain two images with different perspectives on the left and right; Match the corresponding points in the two images; Calculate the disparity of this point in the left and right images, and then combine the internal and external parameters of the camera and the principle of triangulation to deduce the depth information of the target point, and finally obtain the three-dimensional coordinates.

[0011] Preferably, the multiple high-definition cameras perform self-calibration using multi-frame image data, and optimize the internal and external parameters of the cameras through a non-linear optimization algorithm that minimizes the projection error. The objective function is: where P is the R, T, K projection matrix, and x i represents the pixel coordinates of the i-th point in the image, K is the internal parameter matrix, and X i is the three-dimensional coordinate of the i-th point, N is the total number of points, and R and T are the rotation matrix and translation vector, representing the transformation from the world coordinate system to the camera coordinate system.

[0012] Preferably, the Kalman filter performs a preliminary estimation of the displacement data through a state space model, and the steps of the particle filter further correct the displacement estimation, and improve the accuracy of the displacement estimation by weighting the average particles.

[0013] Preferably, the anomaly detection sets a displacement threshold, and according to the detected displacement range, automatically triggers an alarm notification and notifies the management personnel when it exceeds the preset range.

[0014] A foundation pit displacement monitoring system based on machine vision is also provided, including: An image acquisition module, configured to collect image data of the foundation pit area in real time through multiple high-definition cameras, and transmit the image data to the preprocessing module; An image preprocessing module, configured to perform denoising, image correction, and registration processing on the collected image data; A displacement calculation module, configured to analyze the movement of reference points in the image data based on the optical flow method, and combine stereo vision matching to calculate the three-dimensional position change of the reference points, and estimate the actual displacement of the foundation pit; A filtering and optimization module, configured to perform a preliminary estimation of the displacement data of the reference point positions using the Kalman filter, and combine the particle filter to optimize the displacement measurement results, and improve the monitoring accuracy and robustness; An anomaly detection and warning module, configured to monitor the foundation pit displacement data based on a set displacement threshold, and trigger a warning to notify the management personnel if the detected displacement exceeds the preset threshold; A data storage module, configured to store the image data, displacement calculation results, and historical monitoring data for subsequent analysis and backtracking.

[0015] A foundation pit displacement monitoring device based on machine vision is also provided, including the following modules: A high-definition monitoring camera is used to collect image data of the foundation pit area in real time and perform automatic focusing and exposure adjustment to adapt to different lighting environments; A data processing terminal is used to receive the image data collected by the high-definition monitoring camera and perform image preprocessing, displacement calculation, filtering optimization, and anomaly detection operations; A storage unit is used to store the image data, calculated displacement information, and historical monitoring records; An alarm device is used to automatically trigger an alarm and send a warning message to the management personnel when it detects that the displacement of the foundation pit exceeds the set threshold; A communication module is used to transmit the monitoring results of the processing terminal to a remote server and a monitoring platform.

[0016] The present invention provides a foundation pit displacement monitoring method, system, and monitoring device based on machine vision. It has the following beneficial effects: 1. The present invention adopts an image acquisition and stereo vision technology based on a multi-camera system to ensure comprehensive monitoring of the foundation pit displacement. Through the deployment of high-definition cameras, high-quality image data can be provided in various environments, and precise three-dimensional depth information can be obtained by combining stereo vision. Compared with the traditional single-camera method, the present invention avoids the monitoring dead corners caused by the perspective limitation and solves the problems of insufficient accuracy and coverage of single-perspective monitoring in the prior art.

[0017] 2. The present invention adopts a displacement calculation method that combines the optical flow method and stereo vision. Through efficient image processing algorithms, the displacement of the reference points in the foundation pit is accurately estimated. Compared with the prior art that only relies on simple displacement detection or only uses a single technology, the present invention can effectively solve the problem of insufficient accuracy of the optical flow method in complex environments and ensure high accuracy and high real-time performance of displacement calculation.

[0018] 3. The present invention combines the Kalman filter and particle filter algorithms for dynamic optimization estimation, improving the accuracy and robustness of displacement data. By smoothing the displacement data with the Kalman filter and further optimizing the estimation results with the particle filter, compared with the simple averaging or single application of the Kalman filter in traditional methods, the present invention can effectively cope with the challenges brought by noise and non-linear systems and solve the problem of large fluctuations in displacement data in complex environments.

[0019] 4. Through the anomaly detection module of the present invention, an alarm can be immediately triggered when the displacement of the foundation pit exceeds the set threshold, and a warning message can be sent to the management personnel in a timely manner. This real-time response ability avoids the delay processing problem that often exists in traditional monitoring methods, can discover potential risks in the first time, and provides more timely protection for the safety of the foundation pit. Description of the Drawings

[0020] Figure 1 It is a flow chart of the method steps of the present invention; Figure 2 is the system architecture diagram of the present invention; Figure 3 is the schematic diagram of the device of the present invention. Specific embodiments

[0021] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts shall fall within the protection scope of the present invention.

[0022] Embodiment 1: Please refer to the attached Figure 1 , the embodiment of the present invention provides a foundation pit displacement monitoring method based on machine vision, including the following steps: S1. Deploy multiple high-definition cameras to collect image data of the foundation pit area in real time and transmit the image data to the central processing unit; S2. Preprocess the collected image data, and the preprocessing includes denoising and image correction to ensure the monitoring accuracy; S3. Analyze the reference points in the image data based on the optical flow method, calculate the optical flow velocity components of each pixel point, and estimate the displacement of the reference points in the foundation pit; S4. Obtain the image data of multiple high-definition cameras through stereovision technology, perform three-dimensional reconstruction, calculate the three-dimensional positions of the reference points in the foundation pit and deduce their displacements; S5. Use Kalman filtering and particle filtering to dynamically optimize the estimation of the displacement data of the reference points in the foundation pit, and trigger an early warning notification through anomaly detection.

[0023] For step S1, in this embodiment, multiple high-definition cameras are deployed in the foundation pit area, and these cameras monitor the foundation pit area at specific angles and field of view angles. The arrangement method of the cameras can be flexibly adjusted according to the actual scale and environmental requirements of the foundation pit. To ensure that all areas where displacement may occur around the foundation pit can be monitored, the cameras usually need to be distributed in multiple directions of the foundation pit. Generally, the cameras will be deployed around the foundation pit and in special areas where displacement changes may occur, such as the pit wall, bottom, etc.

[0024] As an option, the specific positions and angles of the camera arrangement can be calculated to ensure that the field of view of each camera does not overlap too much with other cameras, so as to achieve a non-blind spot coverage of the monitoring area. The field of view angle of each camera is usually selected according to the specific requirements of the construction site, and a field of view angle of 60 to 120 degrees is generally selected to ensure the extensiveness of the monitoring range.

[0025] In a possible implementation, each camera is configured with autofocus and auto-exposure functions to cope with changes in ambient light. Considering that the foundation pit environment may have significant light variations (such as differences between day and night, cloudy and sunny days), the autofocus and auto-exposure functions can ensure that the camera can clearly capture images whether the light is strong or weak. Specifically, when the light changes significantly, the camera will automatically adjust its aperture and shutter speed to ensure image clarity in different environments.

[0026] In addition, in some embodiments, the camera can also be configured with an image sensor with night vision function to ensure that sufficiently clear images can still be obtained at night or in low-light environments. The night vision function is usually achieved by using infrared illumination and an infrared camera, and images of the foundation pit area can be obtained without external light sources.

[0027] The frequency of image acquisition can usually be set according to on-site monitoring requirements. In a high-risk operation environment, the image acquisition frequency can be set to once per second to promptly capture displacement changes even in rapidly changing situations. In a normal environment, the acquisition frequency is generally once per minute, balancing the data volume and processing efficiency.

[0028] Regarding the image data transmission method, the image data captured by the camera is transmitted in real time to the central processing unit via a wireless network or a wired network. The central processing unit receives the data transmitted by all cameras and performs subsequent image processing and displacement analysis. The delay of image transmission is very low to ensure the effectiveness of real-time monitoring. To ensure the stability and security of image transmission, the image data can be compressed before transmission, thereby reducing the bandwidth requirements for data transmission.

[0029] Specifically, the images captured by each camera are marked according to timestamps, facilitating subsequent image analysis. During the image transmission process, the image data stream transmission method is adopted to ensure the integrity of the image data without packet loss. After the image data is transmitted to the central processing unit, it enters the next preprocessing stage for denoising and correction.

[0030] In certain embodiments, the system can be equipped with multiple high-definition camera modules, each module containing one or more high-definition cameras, capable of focusing on different foundation pit areas for key monitoring. This modular design can flexibly adjust the camera layout according to the construction progress, ensuring that the monitoring system has good adaptability and scalability.

[0031] For step S2, the collected images may be interfered by environmental changes, device factors or other external factors, resulting in noise, distortion or other factors affecting data quality in the images. Therefore, these image data must be preprocessed to ensure the accuracy and stability of subsequent steps. The key objectives of preprocessing are to remove unnecessary noise, correct image distortion, and ensure the consistency of the image in space and time.

[0032] In this embodiment, the preprocessing of image data mainly includes two major steps: denoising and image correction. The purpose of denoising is to eliminate the noise caused by environmental factors (such as light changes, weather changes, device vibrations, etc.), thereby improving the image quality. The goal of image correction is to eliminate the geometric distortion caused by the camera itself or the installation position, and ensure the accuracy of the image data.

[0033] In terms of denoising, Gaussian filtering is widely used in the present invention. Gaussian filtering can effectively smooth the image and remove the noise in the image by performing weighted averaging on each pixel point in the image and its surrounding pixels. Specifically, for a certain pixel point I(x, y) in the given image, its value I ′ (x, y) after Gaussian filtering is calculated by the formula: where I(x, y) is the pixel value of the original image at the point (x, y), I ′ (x, y) represents the pixel value of the image after Gaussian filtering at (x, y), I(x + i, y + j) is the pixel value of the original image at the position (x + i, y + j), (i, j) is the offset of the current filtering window, σ is the standard deviation of Gaussian filtering, which controls the range and intensity of filtering, is the normalization factor, is the weight part of the Gaussian function, n is the window size of the filter, which determines the number of surrounding neighboring pixels. Through Gaussian filtering, the high-frequency noise in the image will be effectively removed, making the image smoother and clearer, facilitating subsequent displacement estimation and analysis.

[0034] As an option, under specific environmental conditions, median filtering can also be used for denoising. Different from Gaussian filtering, median filtering replaces the value of each pixel by taking the median of the pixel neighborhood, and is particularly effective in removing salt-and-pepper noise. In the case where the image contains strong noise, using median filtering can improve the image quality.

[0035] In terms of image correction, in this embodiment, an image distortion correction method based on camera internal parameters and external parameters is adopted. Since the images captured by the camera may be affected by lens distortion, the straight parts of the images may appear curved. To solve this problem, image correction first needs to obtain the internal parameters (focal length, principal point position, etc.) and external parameters (camera position and attitude) of the camera. Through these parameters, the distortion in the image can be corrected.

[0036] Specifically, the image correction in the present invention is completed by using a perspective transformation or a distortion correction algorithm. Given a pixel point (x, y) in the image, its corrected coordinates (x ′ , y ′ ) are calculated through the following equation: where H is the image correction matrix, usually obtained through camera calibration, (x, y) are the coordinates in the original image, and (x ′ , y ′ ) are the coordinates after correction.

[0037] Through the above transformation, the distorted parts in the image will be corrected, and the geometric structure of the real scene will be restored. Specifically, in the implementation process, the calibration methods of binocular cameras or multi-camera systems may also be adopted to further optimize the geometric correction of the image, especially in the case of poor image quality or strong distortion.

[0038] In a possible implementation, the camera is calibrated regularly by using a calibration board to ensure the accuracy of the camera during long-term use. The internal and external parameters of the camera can be obtained during the calibration process, so as to accurately correct the distortion in the captured image.

[0039] Through the above two processes, the quality of the image is significantly improved, the noise is effectively removed, the geometric distortion is corrected, ensuring the accuracy of the image data, and providing a stable basis for displacement analysis and 3D reconstruction in the subsequent steps.

[0040] For step S3, the displacement of the reference points in the foundation pit is further estimated by calculating the optical flow velocity components of each pixel point. The optical flow method estimates the movement of an object or a reference point in three-dimensional space by analyzing the pixel movement of the object or the reference point in consecutive frame images, and then obtains the displacement of the foundation pit.

[0041] In this embodiment, the optical flow method calculates the displacement by analyzing the movement of the reference point in the image. Specifically, the optical flow method calculates the velocity component of the object through the movement of pixels based on the spatial and temporal changes of the image. For two consecutive frames of images I1 (x, y) and I2 (x, y), the velocity component of the point is calculated by analyzing the change of the pixel position (x, y) in the two frames. This calculation is usually implemented based on the optical flow equation, which is expressed as: I x ·u+I y ·v+I t =0; Among them, I x and I y are the spatial gradients of the image I(x,y) in the x and y directions, indicating the changes in pixel values ​​in the horizontal and vertical directions, respectively. t is the temporal gradient of the image, indicating the pixel change between two consecutive frames of images. u and v are the optical flow velocity components of the pixel in the x and y directions, respectively, indicating the movement speed of the pixel in the image plane.

[0042] As an option, in some embodiments, the optical flow velocity components u and v may be estimated by the classic Lucas-Kanade method, which smoothes the local area in the image, calculates the velocity of the pixels in each local window, and further infers the displacement of the reference point based on the velocity.

[0043] Specifically, in the embodiment of the present invention, after the optical flow method analyzes and calculates the optical flow velocity component, the displacement of the reference point in the three-dimensional space can be further estimated according to the pixel displacement in two adjacent frames of the image. By calculating the movement of each reference point in the image, the displacement information of the two-dimensional image can be mapped to the three-dimensional space, thereby obtaining the actual displacement data in the foundation pit.

[0044] In one possible implementation, if there are multiple high-definition cameras and the image data comes from different angles, the optical flow method can combine the image data from multiple perspectives for comprehensive analysis. The different perspectives provided by multiple high-definition cameras combined with the depth information calculated by stereo vision technology can improve the accuracy and robustness of the optical flow method, especially in complex or occluded situations, further improving the accuracy of displacement calculation.

[0045] For step S4, stereo vision technology is used in combination with image data obtained by multiple high-definition cameras to calculate the three-dimensional position of the reference point in the foundation pit through three-dimensional reconstruction, thereby inferring the displacement of the reference point in the foundation pit.

[0046] In this embodiment, the stereo vision technology captures images of the foundation pit area from different angles through multiple high-definition cameras, and performs three-dimensional reconstruction by using the geometric relationships between the images. Specifically, first, the corresponding points in the images obtained by the multiple high-definition cameras are matched, geometric constraints are imposed on these corresponding points through the fundamental matrix F, and the depth information of each reference point is calculated through stereo matching. Through the three-dimensional reconstruction algorithm, the position of each reference point in the three-dimensional space is calculated, and thus the displacement of this point is further deduced.

[0047] In this embodiment, the stereo vision technology captures images of the foundation pit area from different angles through multiple high-definition cameras, and performs three-dimensional reconstruction by using the geometric relationships between the images. Specifically, first, the corresponding points in the images obtained by the multiple high-definition cameras are matched, geometric constraints are imposed on these corresponding points through the fundamental matrix F, and the depth information of each reference point is calculated through stereo matching. Through the three-dimensional reconstruction algorithm, the position of each reference point in the three-dimensional space is calculated, and thus the displacement of this point is further deduced.

[0048] As an option, the core of stereo vision is to estimate the depth information by calculating the disparity between corresponding points in multiple images. In this process, based on the calculation of the fundamental matrix, by matching the corresponding points in two images, the geometric relationship between the two cameras is described. Specifically, the fundamental matrix F satisfies the following constraint relationship: where x and y are the coordinates of the corresponding points in the two images, represents the transpose, F is the fundamental matrix, which is used to describe the geometric relationship between the two cameras.

[0049] Specifically, the result calculated through the fundamental matrix F enables the corresponding points found in the two images to be accurately aligned, and their three-dimensional positions are obtained by using the geometric relationship. Next, through the principle of triangulation and the calculation of depth information, combined with the internal and external parameters of the camera, the exact position of each reference point in the three-dimensional space can be calculated.

[0050] In a possible implementation, the internal and external parameters of the camera (such as focal length, principal point position, rotation matrix, and translation vector) can be obtained through the calibration process. Through the accurate calibration of the internal and external parameters, the error caused by the position change between the cameras can be eliminated, and the accuracy of three-dimensional reconstruction can be improved. Further, the three-dimensional coordinates deduced by the stereo vision technology can accurately reflect the actual spatial position of the reference points in the foundation pit.

[0051] As an option, in the usage scenario of multiple high-definition cameras, the wider the viewing angle, the higher the calculated three-dimensional position and displacement accuracy. Specifically, through the comprehensive calculation of multiple perspectives, not only can the positioning accuracy of a single reference point be improved, but also the blind area problem caused by partial occlusion or viewing angle issues can be overcome.

[0052] For step S5, Kalman filtering and particle filtering techniques are introduced to optimize the estimation of displacement data. Through the combination of these two filtering techniques, the system can effectively smooth the data, eliminate noise, and provide high-precision estimation results for dynamic changes. In addition, when the displacement data exceeds the preset threshold, an early warning notification is triggered through the anomaly detection module to alert the management personnel to pay attention to potential risks.

[0053] In this embodiment, Kalman filtering is used to perform preliminary optimization estimation on the displacement data of the reference point. Kalman filtering is a recursive algorithm suitable for state estimation in linear dynamic systems. It obtains the optimal estimation result through two steps of prediction and update, combining the prior information of the system and real-time measurement data. In the present invention, Kalman filtering can effectively smooth the displacement data, thereby reducing the noise impact caused by environmental interference or data transmission delay. The core formula of Kalman filtering is as follows: x k = Ax k-1 + Bu k + w k ; where, x k is the state vector at time k, representing the displacement of the reference point, x k-1 is the state vector at time k - 1, A is the state transition matrix, representing the transformation of the state from the previous moment to the current moment, B is the control matrix, representing the influence of external control input on the state, u k is the control input, representing environmental changes or system changes, w k is the process noise, assumed to be Gaussian noise with zero mean.

[0054] In Kalman filtering, first, the displacement of the reference point is estimated through the prediction step, and then updated according to the newly acquired displacement data, so as to obtain the optimal estimated value.

[0055] As an option, to further improve the robustness and accuracy of the system, this embodiment combines the particle filtering method. Particle filtering is a Bayesian filtering method suitable for nonlinear systems. It estimates the system state by sampling possible states and combining the method of weighted average. In the foundation pit displacement monitoring, particle filtering can handle the non-Gaussian noise in the system and obtain a more accurate displacement estimation through the weighted average of multiple particles. The basic formula of particle filtering is: Among them, is the state estimate at time k, is the weight of the i-th particle at time k, is the state of the i-th particle at time k, and N is the total number of particles.

[0056] Specifically, particle filter samples the state space at each moment, calculates the state estimates of multiple particles, and performs weighted averaging according to the weights of the particles to obtain the optimal displacement estimation. Particle filter can handle situations with large uncertainties and complex noises in the system. Therefore, in some complex environments, particle filter can provide higher accuracy compared to Kalman filter.

[0057] In a possible implementation, Kalman filter and particle filter can be used in combination. Kalman filter is used to provide a preliminary displacement estimation, while particle filter further corrects the displacement estimation based on Kalman filter. Through this combination method, the system can balance the computational efficiency of linear systems and the accuracy of nonlinear systems, improving the overall monitoring effect.

[0058] Example 2: Please refer to the Figure 2 , the foundation pit displacement monitoring system based on machine vision includes: An image acquisition module, which is used to collect image data of the foundation pit area in real time through multiple high-definition cameras and transmit the image data to the preprocessing module; An image preprocessing module, which is used to perform denoising, image correction and registration processing on the collected image data; A displacement calculation module, which is used to analyze the movement of reference points in the image data based on the optical flow method, and combine stereo vision matching to calculate the three-dimensional position change of the reference points to estimate the actual displacement of the foundation pit; A filtering and optimization module, which is used to use Kalman filter to perform a preliminary estimation on the reference point displacement data, and combine particle filter to optimize the displacement measurement result to improve the monitoring accuracy and robustness; An anomaly detection and warning module, which is used to monitor the foundation pit displacement data based on the set displacement threshold. If the detected displacement exceeds the preset threshold, it will trigger a warning to notify the management personnel; A data storage module, which is used to store the image data, displacement calculation results and historical monitoring data for subsequent analysis and backtracking.

[0059] Specifically, in this embodiment, the image acquisition module is used to collect image data of the foundation pit area in real time through multiple high-definition cameras, and transmit the image data to the preprocessing module. The module performs real-time monitoring through a high-resolution camera and can ensure the clarity and stability of image acquisition under different lighting conditions. The real-time transmission of image data is usually carried out through a wireless network or a wired network to ensure that the image information can be transmitted to the processing system quickly and without delay.

[0060] The image preprocessing module is used to perform denoising, image correction and registration on the collected image data. This module uses efficient image denoising techniques, such as Gaussian filtering, mean filtering or median filtering, to eliminate noise caused by environmental factors or equipment vibration. For image correction, geometric correction and distortion correction techniques are used to correct image errors caused by camera perspective or lens distortion. The image registration step ensures that images collected from different cameras can be processed in the same coordinate system to achieve seamless multi-perspective fusion.

[0061] The displacement calculation module is used to analyze the movement of reference points in image data based on the optical flow method, and calculate the three-dimensional position change of the reference points in combination with stereo vision matching to estimate the actual displacement of the foundation pit. The optical flow method calculates the velocity component of the reference point based on the movement of pixels between images, and further infers the displacement. Combined with stereo vision technology, the images captured by multiple high-definition cameras are matched using the basic matrix to calculate the three-dimensional coordinate changes of the reference points and infer the tiny displacement in the foundation pit.

[0062] The filter optimization module is used to make a preliminary estimate of the reference point displacement data using Kalman filtering, and optimize the displacement measurement results in combination with particle filtering to improve monitoring accuracy and robustness. Kalman filtering smoothes the displacement of the reference point through a recursive process to reduce deviations caused by noise or measurement errors; particle filtering can handle nonlinear problems in the system and further optimize the estimation results, thereby improving the accuracy of displacement estimation and the stability of the system.

[0063] The anomaly detection and early warning module is used to monitor the foundation pit displacement data based on the set displacement threshold. If the displacement is detected to exceed the preset threshold, an early warning will be triggered to notify the management personnel. By real-time detection of the dynamic changes in the foundation pit displacement, the system can issue an alarm in time when abnormal displacement occurs in the foundation pit. The management personnel can take prompt measures through the received early warning information to prevent further risk expansion.

[0064] A data storage module for storing the image data, displacement calculation results, and historical monitoring data for subsequent analysis and backtracking. This module can efficiently store all monitoring data and provide powerful data management and query functions. By storing historical data, the system can perform long-term trend analysis to help evaluate the stability of the foundation pit and provide data support for future engineering improvements.

[0065] Embodiment Three: Please refer to the appendix Figure 3 , a foundation pit displacement monitoring device based on machine vision, including: A high-definition monitoring camera for real-time acquisition of image data in the foundation pit area and automatic focusing and exposure adjustment to adapt to different lighting environments; A data processing terminal for receiving the image data collected by the high-definition monitoring camera and performing image preprocessing, displacement calculation, filtering optimization, and anomaly detection operations; A storage unit for storing the image data, calculated displacement information, and historical monitoring records; An alarm device for automatically triggering an alarm and sending a warning message to the management personnel when the foundation pit displacement is detected to exceed the set threshold; A communication module for transmitting the monitoring results of the processing terminal to a remote server and a monitoring platform.

[0066] A high-definition monitoring camera for real-time acquisition of image data in the foundation pit area and automatic focusing and exposure adjustment to adapt to different lighting environments. This high-definition monitoring camera can provide high-resolution images to ensure that the images of the foundation pit area can be clearly captured under any lighting conditions. The automatic focusing and exposure functions of the camera can adjust its focal length and exposure time in real time according to lighting changes, distance changes, etc., to ensure stable image quality, thereby improving the accuracy of subsequent displacement analysis and calculation. Specifically, the automatic exposure control system can dynamically adjust the image brightness in an environment with large lighting changes to ensure that image details are not lost. The automatic focusing function can adjust the focus in real time to ensure that even when the foundation pit is far away or the image is blurred, the target area can still be clearly captured.

[0067] A data processing terminal for receiving the image data collected by the high-definition monitoring camera and performing image preprocessing, displacement calculation, filtering optimization, and anomaly detection operations. The data processing terminal is the core part of the device, responsible for real-time processing of the collected raw image data. First, it performs preprocessing steps such as denoising, image correction, and registration to ensure data quality. It uses the optical flow method to analyze the movement of reference points in the image, combines stereo vision technology to calculate three-dimensional displacement, uses Kalman filtering and particle filtering for dynamic optimization, and finally performs anomaly detection to timely detect abnormal changes in the foundation pit displacement.

[0068] A storage unit for storing the image data, the calculated displacement information, and the historical monitoring records. The storage unit includes a high-capacity memory that can accommodate a large amount of image data and displacement information. All real-time image data, calculated displacement data, and historical monitoring records will be saved in this module, providing reliable data support for subsequent data analysis, backtracking, and further research. The storage unit supports efficient access to data and can quickly recover historical data in case of a failure or system restart to avoid information loss.

[0069] An alarm device for automatically triggering an alarm and sending a warning message to the management personnel when it detects that the foundation pit displacement exceeds the set threshold. The alarm device plays a crucial role in abnormal displacement monitoring. When the foundation pit displacement exceeds the preset threshold, the system will immediately trigger an alarm and notify the management personnel through a sound or visual warning. In addition, the system will send the warning message to the relevant personnel via text message, email, or other instant messaging methods to ensure that necessary safety measures are taken in a timely manner. The design of the alarm device ensures that in any emergency situation, it can warn the management personnel in the first place to prevent greater potential safety hazards.

[0070] A communication module for transmitting the monitoring results of the processing terminal to a remote server and a monitoring platform. The communication module ensures that the system can perform real-time data interaction with the remote server and the monitoring platform. Through wired or wireless communication technologies, the monitoring device can upload the monitoring results, displacement data, and abnormal alarms to the remote platform in a timely manner, facilitating project management personnel to remotely monitor the foundation pit displacement. This remote monitoring function can greatly improve work efficiency and response capabilities, enabling management personnel to track the dynamic changes of the foundation pit in real time at any location to ensure construction safety.

[0071] Although the embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for monitoring the displacement of a foundation pit based on machine vision, characterized in that, It includes the following steps: Deploy multiple high-definition cameras to collect image data of the foundation pit area in real time and transmit the image data to the central processing unit; Preprocess the collected image data, and the preprocessing includes denoising and image correction to ensure the monitoring accuracy; Analyze the reference points in the image data based on the optical flow method, calculate the optical flow velocity components of each pixel point, and estimate the displacement of the reference points in the foundation pit; Obtain the image data of multiple high-definition cameras through stereo vision technology, perform three-dimensional reconstruction, calculate the three-dimensional positions of the reference points in the foundation pit and deduce their displacements; Use Kalman filtering and particle filtering to dynamically optimize the estimation of the displacement data of the reference points in the foundation pit, and trigger an early warning notification through anomaly detection.

2. The method for monitoring the displacement of the foundation pit based on machine vision according to claim 1, wherein, The step of preprocessing the collected image data includes denoising the image through Gaussian filtering and performing image registration to ensure the consistency between images.

3. The method for monitoring the displacement of the foundation pit based on machine vision according to claim 1, wherein, The steps of the optical flow method include calculating the optical flow velocity of each pixel point using the Lucas-Kanade algorithm, and the optical flow equation is: I x ·u + I y ·v + I t = 0; where I x , I y , I t are the gradients of the image in the x-direction, y-direction, and time t, respectively, and u and v are the velocity components of the pixel points in the x- and y-directions.

4. The method for monitoring the displacement of the foundation pit based on machine vision according to claim 1, characterized in that, The camera image data is obtained through stereo vision technology, and the stereo vision technology matches the corresponding points in the multi-camera images through the fundamental matrix F, satisfying the following relationship: where x and y are the coordinates of corresponding points in the two images respectively, denotes transpose, and F is the fundamental matrix used to describe the geometric relationship between the two cameras.

5. The method for monitoring the displacement of the foundation pit based on machine vision according to claim 1, wherein The three-dimensional reconstruction uses a stereo matching algorithm to calculate the depth information of the reference points in the foundation pit, and deduces the displacement amount through a three-dimensional reconstruction algorithm. The steps of the stereo matching algorithm: Shoot the same target from different angles by two cameras to obtain two images with different perspectives on the left and right; Match the corresponding points in the two images; Calculate the disparity of the point in the left and right images, and then combine the internal and external parameters of the camera and the principle of triangulation to deduce the depth information of the target point and finally obtain the three-dimensional coordinates.

6. The method for monitoring the displacement of the foundation pit based on machine vision according to claim 1, wherein The multiple high-definition cameras perform self-calibration using multi-frame image data, and optimize the internal and external parameters of the cameras through a non-linear optimization algorithm that minimizes the projection error. The objective function is: where P is the projection matrix of R, T, K, and x i represents the pixel coordinates of the i-th point in the image, K is the intrinsic matrix, X i is the 3D coordinate of the i-th point, N is the total number of points, and R and T are the rotation matrix and translation vector, representing the transformation from the world coordinate system to the camera coordinate system.

7. The method for monitoring the displacement of the foundation pit based on machine vision according to claim 1, wherein, The Kalman filtering performs a preliminary estimation of the displacement data through a state space model, and the steps of the particle filtering further correct the displacement estimation, and improve the accuracy of the displacement estimation by weighted average particles.

8. The method for monitoring the displacement of a foundation pit based on machine vision according to claim 1, characterized in that, The anomaly detection triggers an alarm notification automatically according to the detected displacement range beyond the preset range by setting a displacement threshold, and notifies the management personnel.

9. A foundation pit displacement monitoring system based on machine vision, which is applied to the foundation pit displacement monitoring method based on machine vision according to any one of claims 1-8, characterized in that, It includes: An image acquisition module for collecting image data of the foundation pit area in real time through multiple high-definition cameras and transmitting the image data to the preprocessing module; An image preprocessing module for performing denoising, image correction and registration processing on the collected image data; A displacement calculation module for analyzing the movement of the reference points in the image data based on the optical flow method and combining stereo vision matching to calculate the three-dimensional position change of the reference points and estimate the actual displacement of the foundation pit; A filtering and optimization module for performing a preliminary estimation of the displacement data of the reference points using Kalman filtering and combining particle filtering to optimize the displacement measurement result and improve the monitoring accuracy and robustness; An anomaly detection and early warning module for monitoring the displacement data of the foundation pit based on the set displacement threshold, and triggering an early warning to notify the management personnel if the detected displacement exceeds the preset threshold; A data storage module for storing the image data, displacement calculation results, and historical monitoring data for subsequent analysis and backtracking.

10. A foundation pit displacement monitoring device based on machine vision, which is applied to the foundation pit displacement monitoring method based on machine vision according to any one of claims 1-8, characterized in that, It includes: A high-definition monitoring camera for collecting image data of the foundation pit area in real time and automatically focusing and adjusting exposure to adapt to different lighting environments; A data processing terminal for receiving the image data collected by the high-definition monitoring camera and performing image preprocessing, displacement calculation, filtering optimization, and anomaly detection operations; A storage unit for storing the image data, calculated displacement information, and historical monitoring records; An alarm device for automatically triggering an alarm and sending a warning message to the management personnel when the foundation pit displacement is detected to exceed the set threshold; A communication module for transmitting the monitoring results of the processing terminal to a remote server and a monitoring platform.

Images