Stay cable force rapid measurement method and system based on digital image correlation method

Abstract

The invention discloses a method and a system for rapidly measuring cable force of a stay cable based on a digital image correlation method. The method comprises the following steps: acquiring video data of the stay cable; a line segment measurement calibration method is adopted, stay cable texture points serve as feature points, and a feature point selection mechanism is established; establishing a pixel-physical space coordinate mapping model based on a redundant observation method; model parameters are calculated through a collaborative calibration method; performing displacement measurement based on a digital image correlation method; establishing a double-layer feature extraction mechanism to complete local and global feature extraction; weight distribution of the local features and the global features is dynamically adjusted through a weight adaptive fusion strategy; and finally, establishing a double-layer tracking and failure recovery mechanism to complete hierarchical collaboration of coarse-grained global tracking and fine-grained local tracking. And carrying out fast Fourier transform on the dynamic displacement measurement result, and calculating to obtain the cable force. And reliable technical support is provided for non-contact accurate calculation and measurement of the cable force.

Description

Technical Field

[0001] The present invention belongs to the technical field of digital image correlation methods, and more specifically, relates to a method and system for quickly measuring the cable tension of a stay cable based on a digital image correlation method. Background Art

[0002] Cable-stayed bridges are the core structural form of modern long-span bridges. Cable tension monitoring is a key component of bridge health monitoring systems. Cable tension changes directly reflect the stress state of the bridge structure, and accurate dynamic displacement measurement is a fundamental prerequisite for cable tension calculation. However, current cable displacement measurement technology faces multiple technical bottlenecks in engineering applications, urgently requiring systematic breakthroughs.

[0003] Traditional contact measurement methods such as strain gauges and servo displacement meters require physical bonding or mechanical connection to contact the cable surface, which not only interferes with the original vibration characteristics of the cable and causes distortion of the measurement data, but also faces problems such as difficult installation and maintenance and short sensor life in complex environments such as high altitude and high humidity during the bridge operation period, making it difficult to achieve long-term continuous monitoring.

[0004] Among non-contact measurement technologies, the digital image correlation (DIC) method has been widely used due to its advantages of non-contact and full-field measurement. However, its matching mechanism, which relies on local texture features, has inherent defects: in long-distance monitoring scenarios, the resolution of the surface texture of the inclined cable is reduced in the image, and texture missing areas are prone to appear; large vibrations of the cable caused by strong winds, vehicle loads, etc. will cause local feature points to exceed the DIC search range; environmental factors such as sudden changes in illumination and atmospheric turbulence will interfere with the image grayscale distribution, resulting in a decline in matching quality, and ultimately leading to tracking failure and interruption of displacement data.

[0005] Another core pain point of existing technologies is insufficient calibration accuracy. The traditional line segment calibration method lacks scientific screening and spatial distribution optimization in feature point selection, making it difficult for calibration accuracy to meet engineering requirements. Single feature tracking mode is even more difficult to balance measurement accuracy and environmental adaptability - when relying solely on DIC local features, although high positioning accuracy can be achieved, reliability drops sharply in texture-missing or large displacement scenarios; when relying solely on global features, although contour semantic robustness is achieved, spatial positioning accuracy cannot meet requirements, and the fusion mechanism of global and local features lacks dynamic adjustment logic, making it difficult to switch intelligently according to real-time scenarios.

[0006] When displacement measurement data is erroneous or interrupted due to tracking failure, spectral analysis will produce frequency offsets, and cable force calculations will result in deviations due to distorted input data, exceeding the allowable range of the specification, seriously affecting the accuracy of bridge safety assessments.

[0007] In summary, existing technologies face core problems in the displacement measurement of inclined cables, such as "insufficient long-distance calibration accuracy" and "failure of tracking in complex environments". It is urgent to achieve high-precision monitoring through technological innovation to provide reliable technical support for accurate calculation of cable forces and safety assessment of bridge structures. Summary of the Invention

[0008] This invention aims to provide an improved method for measuring the displacement of stay cables, addressing existing issues such as insufficient calibration accuracy during long-distance monitoring and tracking failure in complex environments. By establishing a precise pixel-to-physical space mapping through line segment measurement calibration, combined with a multi-scale feature fusion tracking method and a dual-layer tracking mechanism, this method achieves a synergistic combination of high-precision measurement of local features and robustness assurance of global features, improving the accuracy and environmental adaptability of displacement measurement and providing reliable data support for the accurate calculation of stay cable forces.

[0009] In view of the above defects or improvement needs of the prior art, as a first aspect of the present invention, the present invention provides a method for quickly measuring the tension of a stay cable based on a digital image correlation method, comprising: S1. Collect video data of the cable-stayed cable; S2. Using the line segment measurement calibration method, with the cable texture points as feature points, a feature point optimization selection mechanism is established; a pixel-to-physical space coordinate mapping conversion model is established based on the redundant observation method; and the model parameters are calculated using the collaborative calibration method; S3. Displacement measurement is performed based on the digital image correlation (DIC) method. A two-layer feature extraction mechanism is established to extract local and global features. The weight distribution of local and global features is dynamically adjusted through a weight-adaptive fusion strategy. Finally, a two-layer tracking and failure recovery mechanism is established to achieve hierarchical coordination of "coarse-grained global tracking" and "fine-grained local tracking." S4. Perform fast Fourier transform on the dynamic displacement measurement results to calculate the cable force.

[0010] Furthermore, the feature point optimization selection mechanism in S2 is specifically as follows: The dense intersection points of the threads on the surface of the cable are selected as feature points. Their high texture contrast ensures sub-pixel positioning accuracy and avoids tracking errors caused by missing features in smooth areas. The image gradient response is calculated using the Sobel operator to establish a candidate set of feature points with gradient values ​​greater than 128. This filters out points with low contrast and those susceptible to noise, thereby improving the reliability of feature points. A grid layout strategy is adopted to divide the image into multiple sub-regions. Feature points are evenly selected in each sub-region to avoid calibration parameter deviation caused by local aggregation and ensure balanced global calibration accuracy.

[0011] Furthermore, the conversion model of the pixel-physical space coordinate mapping in S2 is specifically: Establish a feature point set , defining pixel coordinates To physical coordinates The conversion relationship is: , , in, is the local calibration coefficient of each point, and The offset is determined by subsequent collaborative calibration method optimization to achieve accurate mapping of pixel displacement to physical displacement.

[0012] Furthermore, the collaborative calibration method in S2 iteratively optimizes the initial calibration parameters through RANSAC robust estimation and weighted fusion: The RANSAC robust estimation randomly selects at least 4 feature points, calculates the transformation matrix H, counts the number of inliers with a reprojection error < 2 pixels, and saves the optimal transformation parameters; The weighted fusion is assigned weights according to the stability of the feature points, and the calculation formula is: , in, For the The standard deviation of the tracking error of feature points is The corner point response value ensures that feature points with high stability have a greater weight in the calibration and optimizes the overall accuracy; is the global error benchmark; is a feature point; For the Adaptive weights of feature points.

[0013] Furthermore, the dual-layer feature extraction mechanism in S3 includes local feature layer extraction driven by DIC and global feature layer extraction driven by CNN; The local feature layer extraction dynamically sets the subset size (2M+1)×(2M+1) according to the texture density of the inclined cable. A small-size subset is used in the texture-dense area, that is, M is not greater than 8 to ensure positioning accuracy; a large-size subset is used in the smooth area, that is, M is not less than 16 to enhance grayscale correlation; texture richness screening is performed by calculating the texture richness index of each subset, as follows: , Where, is the grayscale standard deviation, is the grayscale mean, only retain A subset of is used as tracking feature points to avoid tracking failure due to lack of texture in smooth areas; is the texture richness index of the subset; is the texture richness threshold set; The global feature layer extracts the input cable ROI image, outputs the cable contour probability map, and extracts the cable global contour features through semantic segmentation to make up for the shortcomings of the DIC local features when they vibrate or are occluded in a large range. The boundary enhancement loss function is used to simultaneously optimize the contour pixel classification accuracy and boundary positioning accuracy to ensure the spatial positioning accuracy of the global features. , in, is the total loss, is the pixel classification loss, which measures the overlap between the predicted contour and the true contour. It is the boundary localization loss, which measures the distance between the predicted contour boundary and the true boundary.

[0014] Furthermore, the weight adaptive fusion strategy in S3 is as follows: , in, is the final feature weight after fusion; is the weight of the DIC local feature; is the weight of CNN global features; is the weight coefficient, which is calculated as follows: , For DIC matching quality, comprehensive considerations The correlation coefficient and sub-region displacement standard deviation are calculated as follows: , Where, Indicates the subset Correlation coefficient; Indicates the The standard deviation of the displacement of the subsets; represents the total number of DIC subsets involved in the evaluation; The quality of CNN contour extraction is evaluated by contour completeness and image signal-to-noise ratio.

[0015] Furthermore, the double-layer tracking process in S3 is specifically as follows: A lightweight semantic segmentation network is used to extract the complete outline of the cable and output a discrete set of centerline points to achieve robust estimation of large-scale displacement: , Where, is the discrete point set of the global center line of the cable; For the The coordinates of the discrete points on the center line; By differentiating the centerline point set between frames, the global displacement field is obtained: , Where, is the global displacement field of the cable; for The set of cable centerline points at time; for The set of cable centerline points at time; is the timeframe index; When texture is missing or illumination changes suddenly, the semantic robustness of CNN contours can maintain basic tracking and avoid the "cliff-like failure" of local feature matching; In global displacement Within the defined “small search window”, the IC-GN method is applied to the DIC subset to compensate for the “coarse-grained error” of global positioning; the sub-pixel displacement residual is solved by optimizing the grayscale correlation of pixels within the subset. , the final output fusion displacement: , Where, Represents the high-precision displacement of the final output; represents the rough displacement of the global layer; Represents the local layer refinement displacement.

[0016] Furthermore, the failure recovery process in S3 is specifically as follows: When local DIC tracking is at risk of failure due to poor matching quality or abnormal displacement residuals, the system activates the "global feature backup + expanded range rematch" mechanism to ensure continuous and reliable displacement measurement. The details are as follows: When any of the following warning signals exists, it indicates that local tracking is unreliable: Grayscale matching index of DIC subset ; Local displacement residual ; When any of the above conditions is met, local tracking is considered invalid and the following recovery process is triggered: Call the lightweight semantic segmentation network to re-extract the complete outline of the cable and generate a new centerline point set ;According to the semantic robustness of the contour, the tracking anchor points are quickly reconstructed to avoid direct data loss; The search window of the DIC subset is expanded from the conventional small range to the possible area predicted by the global layer, and the texture is re-matched in combination with the IC-GN method; by "expanding the search radius", the displacement mutation or texture missing areas are covered, thereby improving the success rate of DIC rematching.

[0017] As a second aspect of the present invention, a system for rapidly measuring the tension of a stay cable based on a digital image correlation method is provided, comprising: A video acquisition unit, used for acquiring video data of the stay cables; The calibration modeling unit is used to establish a feature point optimization selection mechanism using the line segment measurement calibration method and the cable texture points as feature points; establish a pixel-to-physical space coordinate mapping conversion model based on the redundant observation method; and calculate the model parameters through the collaborative calibration method; The displacement measurement unit is used to perform displacement measurement based on the digital image correlation method (DIC). A dual-layer feature extraction mechanism is established to extract local and global features. The weight distribution of local and global features is dynamically adjusted through a weight-adaptive fusion strategy. Finally, a dual-layer tracking and failure recovery mechanism is established to achieve hierarchical coordination of "coarse-grained global tracking" and "fine-grained local tracking." The cable force calculation unit is used to perform fast Fourier transform on the dynamic displacement measurement results to calculate the cable force of the inclined cable.

[0018] As a third aspect of the present invention, a computer-readable storage medium is provided, on which a computer program is stored, and the computer program is executed by a processor to execute any step of the method for quickly measuring the tension of a cable-stayed cable based on a digital image correlation method.

[0019] In general, the above technical solutions conceived by the present invention can achieve the following beneficial effects compared with the prior art: 1. The method for rapid measurement of cable tension in the present invention based on the digital image correlation method realizes high-precision pixel-to-physical space conversion by constructing a systematic calibration system. The line segment measurement calibration method is adopted, and the texture points of the cable are used as feature points. Through the feature point optimization selection mechanism, the dense intersection points of the thread lines are selected and combined with Sobel gradient screening and grid layout to ensure the reliability and balanced spatial distribution of the feature points. A pixel-to-physical space coordinate mapping conversion model is established based on redundant observations, and the model parameters are iteratively optimized through the collaborative calibration method of RANSAC robust estimation and weighted fusion. This calibration system effectively overcomes the problems of arbitrary feature point selection and low calibration accuracy in traditional methods, establishes an accurate correspondence between pixels and physical space, and provides an accurate basis for subsequent displacement measurement.

[0020] 2. The fast cable tension measurement method based on the digital image correlation method of the present invention improves the accuracy and stability of displacement measurement through an innovative dual-layer feature extraction and weight adaptive fusion strategy. A dual-layer feature extraction mechanism is established. The local feature layer is driven by DIC, dynamically adjusts the subset size according to the texture density, and filters the texture-rich areas; the global feature layer is driven by CNN to extract the cable contour features. At the same time, a weight adaptive fusion strategy is adopted to dynamically adjust the weights of the two according to the DIC matching quality and the CNN contour extraction quality. This strategy solves the limitations of single feature tracking in complex environments. In scenarios such as texture loss and lighting changes, the dominant feature can be flexibly switched to ensure the reliability and accuracy of displacement measurement under different conditions.

[0021] 3. The method for rapid measurement of cable tension in inclined cables based on the digital image correlation method of the present invention ensures the continuity of displacement measurement through a double-layer tracking and failure recovery mechanism. In the double-layer tracking, a lightweight semantic segmentation network is first used to achieve global positioning, and the cable centerline displacement is obtained as a rough estimate. Then, based on the global displacement, the local DIC subset is refined by the IC-GN method to obtain high-precision displacement. When there is a risk of failure in local DIC tracking, such as when the ZNCC correlation coefficient is too low or the displacement residual is abnormal, the failure recovery mechanism is triggered, and the CNN global features are reinitialized and the search range is expanded to rematch. This mechanism effectively avoids the data interruption problem caused by local tracking failure, ensuring that the displacement measurement of the inclined cable can remain continuous and stable under complex working conditions such as strong winds and vibrations, providing complete and reliable data for cable tension calculation. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 This is a flow chart of a method for quickly measuring the tension of a stay cable based on a digital image correlation method according to an embodiment of the present invention; Figure 2 This is a vertical layout diagram of a stay cable according to an embodiment of the present invention; Figure 3 A schematic diagram of camera calibration according to an embodiment of the present invention; Figure 4 2 is a diagram of system units according to an embodiment of the present invention. DETAILED DESCRIPTION

[0023] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only intended to illustrate the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.

[0024] Example 1 Please refer to Figure 1 This embodiment 1 provides a method for quickly measuring the tension of a stay cable based on a digital image correlation method, comprising: S1. Collect video data of the cable-stayed cable; S2. Using the line segment measurement calibration method, with the cable texture points as feature points, a feature point optimization selection mechanism is established; a pixel-to-physical space coordinate mapping conversion model is established based on the redundant observation method; and the model parameters are calculated using the collaborative calibration method; S3. Displacement measurement is performed based on the digital image correlation (DIC) method. A two-layer feature extraction mechanism is established to extract local and global features. The weight distribution of local and global features is dynamically adjusted through a weight-adaptive fusion strategy. Finally, a two-layer tracking and failure recovery mechanism is established to achieve hierarchical coordination of "coarse-grained global tracking" and "fine-grained local tracking." S4. Perform fast Fourier transform on the dynamic displacement measurement results to calculate the cable force.

[0025] This embodiment 1 further explains the above steps.

[0026] (1) Video acquisition For non-contact cable force measurement of cable-stayed bridges, it is necessary to first collect video data of the specific cables; please refer to Figure 2 In a preferred embodiment, non-contact cable tension measurement was performed on cable #B-04 of the Yujiang Bridge. The image acquisition device used was an MV-CS050-10UM, the industrial lens was an HTFA12035-12MP, and a x2 teleconverter was used to increase the focal length. 160 seconds of video data were captured for cable #B-04, with an image resolution of 2448×2048 and a frame rate of 20Hz, totaling 3205 frames of image data.

[0027] (2) Calibration modeling Because the acquisition scene is over 150 meters away, a laser rangefinder cannot be used to determine the distance between the object and the camera. Furthermore, traditional targets cannot be attached to the surface of the measured cable. Therefore, calibration methods that rely on targets or distance measurement are abandoned, and a line segment measurement calibration method is adopted. This method uses the natural texture features of the cable surface (such as thread lines) to achieve calibration by mapping the pixel length of the line segment in the image to its actual physical length, eliminating the need for external targets and distance measurement equipment.

[0028] In a preferred embodiment, a feature point optimization selection mechanism is established to further screen the feature points; the dense intersection points of the thread lines on the surface of the inclined cable are selected as feature points, and their high texture contrast is used to ensure sub-pixel positioning accuracy, avoiding tracking errors caused by feature loss in smooth areas; the image gradient response is calculated using the Sobel operator, and a feature point candidate set with gradient values ​​> 128 is established, and points with low contrast and susceptible to noise interference are filtered out to improve the reliability of feature points; a grid layout strategy is adopted to divide the image into multiple sub-areas, and feature points are evenly selected in each sub-area to avoid calibration parameter deviation caused by local aggregation and ensure balanced global calibration accuracy.

[0029] At the same time, based on the theory of redundant observation, by increasing the number of characteristic point observations (e.g., selecting n characteristic points), the number of system equations is increased by n times. Least squares or weighted least squares methods are then used to fit and solve these multiple sets of observation data. This process is equivalent to "averaging" single-point noise, preventing measurement deviations at a single characteristic point caused by environmental interference (such as sudden changes in illumination or local vibration) from affecting the overall results. For example, if a characteristic point produces an outlier due to noise, the observation data from the other n-1 characteristic points can be used to offset this anomaly through redundant calculations, effectively smoothing out random errors.

[0030] During long-distance measurements, images are susceptible to atmospheric turbulence and slight camera shake, leading to occasional errors in feature point positioning. Redundant observations, by adding independent observations, make calibration results more robust to these occasional errors. For example, traditional single-point calibration may result in deviations due to a single jitter. However, redundant observations can minimize errors by cross-validating multiple sets of data.

[0031] Furthermore, a conversion model for pixel-physical space coordinate mapping is established based on the redundant observation method, as follows: Establish a feature point set , defining pixel coordinates To physical coordinates The conversion relationship is: , , in, is the local calibration coefficient of each point, and The offset is determined by subsequent collaborative calibration method optimization to achieve accurate mapping of pixel displacement to physical displacement.

[0032] Redundant observations provide a large amount of low-noise pixel coordinate data for the mathematical model through a multi-feature point and multi-equation observation system. After these data are processed by the least squares method or the weighted least squares method, the solution can be obtained more accurately ( , , ) and other parameters to avoid model fitting deviations caused by insufficient data. For example, when n = 100 feature points, the model can solve three unknown parameters through 100 sets of equations, forming an overdetermined system of equations, and use the least squares method to obtain the optimal solution.

[0033] The mathematical model converts pixel coordinate data obtained from redundant observations into coordinate values ​​with real physical meaning, making subsequent geometric constraints (such as segment length consistency verification) and distortion correction (such as radial distortion compensation) possible. For example, after the physical coordinates of feature points are calculated through the model, the mapping relationship between the physical distance and pixel distance of a known length segment on the cable surface can be verified to ensure that it conforms to the theoretical value, thereby further optimizing the calibration parameters.

[0034] Based on the established conversion model, the initial calibration parameters are iteratively optimized through a collaborative calibration method (RANSAC robust estimation + weighted fusion); In a preferred embodiment, the collaborative calibration method iteratively optimizes the initial calibration parameters through RANSAC robust estimation and weighted fusion: The RANSAC robust estimation randomly selects at least 4 feature points, calculates the transformation matrix H, counts the number of inliers with a reprojection error < 2 pixels, and saves the optimal transformation parameters; The weighted fusion is assigned weights according to the stability of the feature points, and the calculation formula is: , in, For the The standard deviation of the tracking error of feature points is The corner point response value ensures that feature points with high stability have a greater weight in the calibration and optimizes the overall accuracy; is the global error benchmark; is a feature point; For the Adaptive weights of feature points.

[0035] For example, we obtain the pixel coordinates of 100 feature points through redundant observation and substitute them into the mathematical model to obtain 100 sets of Initial value, after RANSAC excludes 20 outliers, the remaining 80 values ​​are calculated according to the weighted fusion formula .

[0036] At the same time, the output of the collaborative calibration method (optimal ) The applicability of the mathematical model can be reversely verified: If the parameters after weighted fusion have large deviations in multi-scale verification, the constraints of the mathematical model can be adjusted (such as adding nonlinear terms) to form a closed loop of "modeling-optimization-correction" to improve the overall calibration accuracy.

[0037] (3) Displacement measurement Please refer to Figure 3 , the displacement of the test point is measured based on the digital image correlation method (DIC); and the problem of insufficient accuracy in dynamic displacement measurement of inclined cables in long distances and complex environments is solved by integrating multi-scale feature fusion methods; The DIC method selects texture-rich areas on the cable surface (such as the intersection of thread lines) as a reference subset in the initial frame, and then tracks the displacement of the subset in subsequent frames through grayscale correlation analysis. The ZNCC (zero-mean normalized cross-correlation) method is used to calculate the pixel displacement of the subset between adjacent frames. Combined with the pixel-to-physical conversion ratio obtained by camera calibration (such as 0.852033 mm / px), the pixel displacement is converted into actual physical displacement.

[0038] A dual-layer feature extraction mechanism is established in the multi-scale feature fusion tracking method. The dual-layer feature extraction mechanism includes local feature layer extraction driven by DIC and global feature layer extraction driven by CNN. The local feature layer extraction dynamically sets the subset size (2M+1)×(2M+1) according to the texture density of the inclined cable. A small-size subset is used in the texture-dense area, that is, M is not greater than 8 to ensure positioning accuracy; a large-size subset is used in the smooth area, that is, M is not less than 16 to enhance grayscale correlation; texture richness screening is performed by calculating the texture richness index of each subset, as follows: , Where, is the grayscale standard deviation, is the grayscale mean, only retain A subset of is used as tracking feature points to avoid tracking failure due to lack of texture in smooth areas; is the texture richness index of the subset; is the texture richness threshold set; The global feature layer extracts the input cable ROI image, outputs the cable contour probability map, and extracts the cable global contour features through semantic segmentation to make up for the shortcomings of the DIC local features when they vibrate or are occluded in a large range. The boundary enhancement loss function is used to simultaneously optimize the contour pixel classification accuracy and boundary positioning accuracy to ensure the spatial positioning accuracy of the global features. , in, is the total loss, is the pixel classification loss, which measures the overlap between the predicted contour and the true contour. It is the boundary localization loss, which measures the distance between the predicted contour boundary and the true boundary.

[0039] In a preferred embodiment, the multi-scale feature fusion tracking method adopts a weighted adaptive fusion strategy, as follows: , in, is the final feature weight after fusion; is the weight of the DIC local feature; is the weight of CNN global features; is the weight coefficient, which is calculated as follows: , For DIC matching quality, comprehensive considerations The correlation coefficient and sub-region displacement standard deviation are calculated as follows: , Where, Indicates the subset Correlation coefficient; Indicates the The standard deviation of the displacement of the subsets; represents the total number of DIC subsets involved in the evaluation; The quality of CNN contour extraction is evaluated by contour completeness and image signal-to-noise ratio.

[0040] In a preferred embodiment, the multi-scale feature fusion tracking method further includes a dual-layer tracking and failure recovery mechanism, wherein the dual-layer tracking process is specifically as follows: A lightweight semantic segmentation network (such as a U-Net variant) is used to extract the complete outline of the cable and output a discrete set of centerline points to achieve robust estimation of large-scale displacements: , Where, is the discrete point set of the global center line of the cable; For the The coordinates of the discrete points on the center line; By differentiating the centerline point set between frames, the global displacement field is obtained: , Where, is the global displacement field of the cable; for The set of cable centerline points at time; for The set of cable centerline points at time; is the timeframe index; When texture is missing (such as oil stains on the cable surface) or illumination changes suddenly (such as strong light at the tunnel entrance), the semantic robustness of CNN contours can maintain basic tracking and avoid the "cliff-like failure" of local feature matching; In global displacement The IC-GN (Iterative Closest Point-Gauss-Newton) method is applied to the DIC subset (texture-rich area) within a defined "small search window" (e.g., ±5 pixels) to compensate for the "coarse-grained error" of global positioning; the sub-pixel displacement residual is solved by optimizing the grayscale correlation of pixels within the subset. , the final output fusion displacement: , Where, Represents the high-precision displacement of the final output; represents the rough displacement of the global layer; Represents the local layer refinement displacement.

[0041] In a preferred embodiment, the failure recovery process is specifically as follows: When local DIC tracking is at risk of failure due to poor matching quality or abnormal displacement residuals, the system activates the "global feature backup + expanded range rematch" mechanism to ensure continuous and reliable displacement measurement. The details are as follows: When any of the following warning signals exists, it indicates that local tracking is unreliable: Grayscale matching index of DIC subset , which means that interference such as texture blur and sudden illumination changes make the reliability of local feature matching fall below 50%; local displacement residual , that is, displacement mutation, feature mismatch, etc., which causes the residual to exceed 3 times the standard deviation of the noise and deviate from the reasonable range; When any of the above conditions is met, local tracking is considered invalid and the following recovery process is triggered: Call the lightweight semantic segmentation network to re-extract the complete outline of the cable and generate a new centerline point set Based on the semantic robustness of the contour (independent of texture details), the tracking anchors are quickly reconstructed to avoid direct data loss. The search window of the DIC subset is expanded from the conventional small range to the possible area predicted by the global layer, and the texture is re-matched in combination with the IC-GN method; by "expanding the search radius", the displacement mutation or texture missing areas are covered, thereby improving the success rate of DIC rematching.

[0042] The multi-scale feature fusion tracking method uses the collaborative mechanism of "local detail precision measurement + global structural constraints" to ensure sub-pixel accuracy in texture-rich areas, and to make up for the tracking shortcomings in complex environments through global features. It upgrades displacement measurement from a "micro mode" that relies solely on texture to a "macro-micro linkage mode" that integrates structural information, providing a reliable displacement data foundation for bridge health monitoring.

[0043] (4) Cable force calculation After completing video data acquisition, high-precision calibration modeling, and stable displacement measurement of the cable, dynamic displacement measurement results reflecting the cable's vibration were obtained. These results record the cable's displacement changes at different times and contain rich time-domain information.

[0044] The dynamic displacement measurement results were then subjected to a Fast Fourier Transform (FFT). Fast Fourier Transform is a highly efficient frequency-domain analysis method that converts time-domain displacement signals into frequency-domain signals, extracting the natural frequencies of the cable's vibration. Through FFT processing, the originally complex time-domain displacement curve is decomposed into a superposition of different frequency components, clearly revealing the frequency characteristics of the cable's vibration, which are closely related to its physical properties.

[0045] Finally, based on the extracted natural frequency and combined with the structural parameters of the cable (such as length and linear density), the cable force is calculated using relevant mechanical formulas as follows: , , Where, Indicates cable force; represents the linear density of the cable; Indicates the calculated length of the cable; Indicates the natural frequency order; represents the bending stiffness of the cable; Represents the vibration characteristic parameters of the inclined cable.

[0046] This calculation process is based on the principles of structural dynamics and establishes a mathematical relationship between frequency and cable force. By precisely substituting parameters and formulas, the vibration frequency information is converted into specific cable force values, thereby achieving a quantitative assessment of the stress state of the inclined cable and providing key data support for bridge structure health monitoring.

[0047] Example 2 Please refer to Figure 4 This embodiment 2 provides a system for quickly measuring the cable force of a stay cable based on a digital image correlation method, comprising: A video acquisition unit, used for acquiring video data of the stay cables; The calibration modeling unit is used to establish a feature point optimization selection mechanism using the line segment measurement calibration method and the cable texture points as feature points; establish a pixel-to-physical space coordinate mapping conversion model based on the redundant observation method; and calculate the model parameters through the collaborative calibration method; The displacement measurement unit is used to perform displacement measurement based on the digital image correlation method (DIC). A dual-layer feature extraction mechanism is established to extract local and global features. The weight distribution of local and global features is dynamically adjusted through a weight-adaptive fusion strategy. Finally, a dual-layer tracking and failure recovery mechanism is established to achieve hierarchical coordination of "coarse-grained global tracking" and "fine-grained local tracking." The cable force calculation unit is used to perform fast Fourier transform on the dynamic displacement measurement results to calculate the cable force of the inclined cable.

[0048] Example 3 This embodiment 3 also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it can implement any step of a method for quickly measuring the tension of a cable-stayed cable based on a digital image correlation method.

[0049] The computer-readable storage medium may include: a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, etc., which can store program codes.

[0050] For an introduction to the computer-readable storage medium provided in this application, please refer to the above method embodiment, and this application will not go into details here.

[0051] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for rapid measurement of cable tension based on digital image correlation method, characterized in that: include: S1. Collect video data of the cable-stayed cable; S2. Using the line segment measurement calibration method, with the cable texture points as feature points, a feature point optimization selection mechanism is established; a pixel-to-physical space coordinate mapping conversion model is established based on the redundant observation method; and the model parameters are calculated using the collaborative calibration method; S3. Displacement measurement is performed based on the digital image correlation (DIC) method. A two-layer feature extraction mechanism is established to extract local and global features. The weight distribution of local and global features is dynamically adjusted through a weight-adaptive fusion strategy. Finally, a two-layer tracking and failure recovery mechanism is established to achieve hierarchical coordination of "coarse-grained global tracking" and "fine-grained local tracking." S4. Perform fast Fourier transform on the dynamic displacement measurement results to calculate the cable force.

2. The method for rapid measurement of cable tension based on digital image correlation method according to claim 1 is characterized in that: The feature point optimization selection mechanism in S2 is specifically as follows: The dense intersection points of the threads on the surface of the cable are selected as feature points. Their high texture contrast ensures sub-pixel positioning accuracy and avoids tracking errors caused by missing features in smooth areas. The image gradient response is calculated using the Sobel operator to establish a candidate set of feature points with gradient values ​​greater than 128. This filters out points with low contrast and those susceptible to noise, thereby improving the reliability of feature points. A grid layout strategy is adopted to divide the image into multiple sub-regions. Feature points are evenly selected in each sub-region to avoid calibration parameter deviation caused by local aggregation and ensure balanced global calibration accuracy.

3. The method for rapid measurement of cable tension based on digital image correlation method according to claim 1 is characterized in that: The conversion model of the pixel-physical space coordinate mapping in S2 is specifically: Establish a feature point set , defining pixel coordinates To physical coordinates The conversion relationship is: ， ， in, is the local calibration coefficient of each point, and The offset is determined by subsequent collaborative calibration method optimization to achieve accurate mapping of pixel displacement to physical displacement.

4. The method for rapid measurement of cable tension based on digital image correlation method according to claim 1 is characterized in that: The collaborative calibration method in S2 iteratively optimizes the initial calibration parameters through RANSAC robust estimation and weighted fusion: The RANSAC robust estimation randomly selects at least 4 feature points, calculates the transformation matrix H, counts the number of inliers with a reprojection error < 2 pixels, and saves the optimal transformation parameters; The weighted fusion is assigned weights according to the stability of the feature points, and the calculation formula is: ， in, For the The standard deviation of the tracking error of feature points is The corner point response value ensures that feature points with high stability have a greater weight in the calibration and optimizes the overall accuracy; is the global error benchmark; is a feature point; For the Adaptive weights of feature points.

5. The method for rapid measurement of stay cable tension based on digital image correlation method according to claim 1 is characterized in that: The dual-layer feature extraction mechanism in S3 includes local feature layer extraction driven by DIC and global feature layer extraction driven by CNN; The local feature layer extraction dynamically sets the subset size (2M+1)×(2M+1) according to the texture density of the inclined cable. A small-size subset is used in the texture-dense area, that is, M is not greater than 8 to ensure positioning accuracy; a large-size subset is used in the smooth area, that is, M is not less than 16 to enhance grayscale correlation; texture richness screening is performed by calculating the texture richness index of each subset, as follows: ， Where, is the grayscale standard deviation, is the grayscale mean, only retain A subset of is used as tracking feature points to avoid tracking failure due to lack of texture in smooth areas; is the texture richness index of the subset; is the texture richness threshold set; The global feature layer extracts the input cable ROI image, outputs the cable contour probability map, and extracts the cable global contour features through semantic segmentation to make up for the shortcomings of the DIC local features when they vibrate or are occluded in a large range. The boundary enhancement loss function is used to simultaneously optimize the contour pixel classification accuracy and boundary positioning accuracy to ensure the spatial positioning accuracy of the global features. ， in, is the total loss, is the pixel classification loss, which measures the overlap between the predicted contour and the true contour. It is the boundary localization loss, which measures the distance between the predicted contour boundary and the true boundary.

6. The method for rapid measurement of cable tension based on digital image correlation method according to claim 1 is characterized in that: The weight adaptive fusion strategy in S3 is as follows: ， in, is the final feature weight after fusion; is the weight of the DIC local feature; is the weight of CNN global features; is the weight coefficient, which is calculated as follows: ， For DIC matching quality, comprehensive considerations The correlation coefficient and sub-region displacement standard deviation are calculated as follows: ， Where, Indicates the subset Correlation coefficient; Indicates the The standard deviation of the displacement of the subsets; represents the total number of DIC subsets involved in the evaluation; The quality of CNN contour extraction is evaluated by contour completeness and image signal-to-noise ratio.

7. The method for rapid measurement of stay cable tension based on digital image correlation method according to claim 1, characterized in that: The double-layer tracking process in S3 is as follows: A lightweight semantic segmentation network is used to extract the complete outline of the cable and output a discrete set of centerline points to achieve robust estimation of large-scale displacement: ， Where, is the discrete point set of the global center line of the cable; For the The coordinates of the discrete points on the center line; By differentiating the centerline point set between frames, the global displacement field is obtained: ， Where, is the global displacement field of the cable; for The set of cable centerline points at time; for The set of cable centerline points at time; is the timeframe index; When texture is missing or illumination changes suddenly, the semantic robustness of CNN contours can maintain basic tracking and avoid the "cliff-like failure" of local feature matching; In global displacement The IC-GN method is applied to the DIC subset within the defined "small search window" to compensate for the "coarse-grained error" of global positioning; the sub-pixel displacement residual is solved by optimizing the grayscale correlation of pixels within the subset. , the final output fusion displacement: ， Where, Represents the high-precision displacement of the final output; represents the rough displacement of the global layer; Represents the local layer refinement displacement.

8. The method for rapid measurement of stay cable tension based on digital image correlation method according to claim 1 is characterized in that: The specific process of failure recovery in S3 is as follows: When local DIC tracking is at risk of failure due to poor matching quality or abnormal displacement residuals, the system activates the "global feature backup + expanded range rematch" mechanism to ensure continuous and reliable displacement measurement. The details are as follows: When any of the following warning signals exists, it indicates that local tracking is unreliable: Grayscale matching index of DIC subset ; Local displacement residual ; When any of the above conditions is met, local tracking is considered invalid and the following recovery process is triggered: Call the lightweight semantic segmentation network to re-extract the complete outline of the cable and generate a new centerline point set ;According to the semantic robustness of the contour, the tracking anchor points are quickly reconstructed to avoid direct data loss; The search window of the DIC subset is expanded from the conventional small range to the possible area predicted by the global layer, and the texture is re-matched in combination with the IC-GN method. By "expanding the search radius", the areas with displacement mutations or texture missing are covered, thereby improving the success rate of DIC rematching.

9. A fast cable force measurement system based on digital image correlation method, characterized in that: include: A video acquisition unit, used for acquiring video data of the stay cables; The calibration modeling unit is used to establish a feature point optimization selection mechanism using the line segment measurement calibration method and the cable texture points as feature points; establish a pixel-to-physical space coordinate mapping conversion model based on the redundant observation method; and calculate the model parameters through the collaborative calibration method; The displacement measurement unit is used to perform displacement measurement based on the digital image correlation method (DIC). A dual-layer feature extraction mechanism is established to extract local and global features. The weight distribution of local and global features is dynamically adjusted through a weight-adaptive fusion strategy. Finally, a dual-layer tracking and failure recovery mechanism is established to achieve hierarchical coordination of "coarse-grained global tracking" and "fine-grained local tracking." The cable force calculation unit is used to perform fast Fourier transform on the dynamic displacement measurement results to calculate the cable force of the inclined cable.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: The computer program is executed by a processor to implement a method for quickly measuring the tension of a stay cable based on a digital image correlation method as described in any one of claims 1 to 8.

Images