A cable force detection method, system and device based on computer vision

Abstract

The present invention discloses a method, system and device for detecting cable force based on computer vision, the method comprising: obtaining relevant information of the cable to be detected; collecting a tiny vibration video of the cable to be detected, and preprocessing the video frame by frame to obtain a target image; using a phase-based Euler motion amplification algorithm to obtain an amplified image sequence for the target image, and reconstructing the amplified video; using a sub-pixel template matching algorithm to determine the displacement time-history response data corresponding to the motion amplified video; obtaining the fundamental frequency corresponding to the cable to be detected based on the displacement time-history response data, and selecting a suitable cable force-frequency relationship to calculate the cable force according to the actual boundary conditions of the cable. It can be seen that the present application realizes the detection of cable force with very small amplitude of bridge cables, and does not require additional sensors, which reduces costs and has broad application prospects.

Description

Technical Field

[0001] The present invention relates to the technical field of bridge health monitoring, and in particular to a cable force detection method, system and device based on computer vision. Background Art

[0002] Cable-stayed bridges are widely used in long-span bridges due to their advantages such as strong spanning capacity, beautiful appearance and high cost performance. Cable-stayed bridges are hyperstatic structural systems with the main beam bearing bending, the bridge tower bearing pressure and the cable under tension. The cables of cable-stayed bridges are often easily damaged and relaxed due to corrosion, fatigue and other reasons. As important load-bearing components of cable-stayed bridges, cables often produce large displacements due to small stress and strain changes, which leads to relaxation and stress loss. Moreover, the damage of cables may bring catastrophic consequences to the overall structure. It is precisely because of these characteristics that the cable force detection of cable-stayed bridges is of great significance in both the structural construction and use stages. In addition, the safety and durability assessment of bridge structures requires the cable force to be used as an important measurement and evaluation indicator, which provides a basis for bridge condition assessment. Therefore, how to quickly and accurately measure and evaluate the cable force of bridges is very important. It is the premise for ensuring the safe operation of bridges and a strong guarantee for the real-time monitoring of bridge health.

[0003] At present, the most commonly used methods for cable tension detection are oil pressure gauge measurement method, pressure sensor measurement method, frequency method, magnetic flux method, resistor measurement method, sag method, elongation method, etc. The oil pressure gauge measurement method, pressure sensor measurement method and magnetic flux method are not suitable for cable tension detection during the operation stage. The frequency method usually uses contact sensors such as accelerometers to pick up the vibration signal of the cable. Its measurement results are reliable, but there are shortcomings such as difficulty in instrument installation, high cost, and introduction of additional mass. In recent years, with the continuous development of computer vision technology and image acquisition equipment, computer vision-based structural displacement monitoring methods have continued to emerge and have been verified in actual engineering applications. At present, the cable force identification method based on computer vision mainly uses the target tracking algorithm to track the artificial or natural markers on the surface of the structure to obtain the displacement time-history response data. However, the vibration amplitude of the cable under environmental excitation is small, and it is difficult to obtain high-precision cable displacement time-history data based on the general motion target tracking algorithm, which affects the cable force detection result; and when extracting the cable vibration displacement, the result obtained by using digital image-related template matching is pixel-level displacement, but higher measurement accuracy is required in practical applications, so further optimization search is required to reach the sub-pixel level. Therefore, it is necessary to study how to apply high-precision and rapidly developing photogrammetry technology and equipment to the traditional bridge engineering field. For this purpose, a combination of the phase-based Euler motion magnification algorithm and sub-pixel template matching is proposed to establish a set of non-contact, convenient, economical and efficient cable force detection methods, which is of great significance for ensuring the safe operation of bridges and achieving long-term stable use of bridges. Summary of the invention

[0004] Purpose of the invention: In view of the shortcomings of the background technology, the present invention proposes a cable force detection method, system and device based on computer vision to achieve fast and accurate detection of cable force.

[0005] Technical solution: In order to achieve the purpose of the present invention, the present invention proposes a cable force detection method, system and device based on computer vision, and the specific solution is as follows:

[0006] First aspect: This application discloses a cable force detection method based on computer vision, which specifically includes the following steps:

[0007] Step 1: Obtain relevant information of the cable to be tested;

[0008] Step 2: Collect the micro-vibration video of the cable to be tested, and pre-process the video frame by frame to obtain the target image;

[0009] Step 3: using the phase-based Euler motion magnification algorithm to obtain a magnified image sequence for the target image, and reconstructing the magnified video;

[0010] Step 4: Determine the displacement time-course response data corresponding to the motion-amplified video using a sub-pixel template matching algorithm;

[0011] Step 5: based on the displacement time-history response data, the fundamental frequency corresponding to the cable to be tested is obtained, and according to the actual boundary conditions of the cable, a suitable cable force-frequency relationship is selected to calculate the cable force;

[0012] Furthermore, in step 1, the relevant information includes the model, material, diameter, calculated length, line density, inclination angle, boundary conditions, usage time, number or position of the cable.

[0013] Further, the specific process of step 2 is as follows:

[0014] Step 2.1: Select 1 / 4 of the cable as the detection area, and observe whether there are natural markers at this point. If not, set a target point at this point; fix the high-precision industrial camera in a suitable position, and check the lighting conditions on site to determine whether fill light is needed. If fill light is needed, turn on the fill light with its own power supply for lighting; adjust the camera position to ensure that the cable target can be fully captured by the camera and will not exceed the video range when vibrating; adjust the lens focal length and aperture to ensure that the cable image is clearly visible in the camera field of view; set the camera sampling frequency to collect the cable's tiny vibration video;

[0015] Step 2.2: Decompose the cable micro-vibration video captured by the camera into continuous frame images, and set the same region of interest for each frame image; the region of interest includes the target point and the cable movement area to be detected;

[0016] Step 2.3: Preprocess the disassembled continuous frame images frame by frame, including cropping, rotating and scaling the target image sequence, and removing large noise in the target image to highlight the cable structure of interest and reduce the influence of other environmental factors.

[0017] Further, the specific method of step 3 is as follows:

[0018] Step 3.1, spatial domain filtering: Use the complex controllable pyramid to perform spatial domain filtering on the input target image to obtain high-pass residuals, low-pass residuals, and image sequences of different scales and directions, namely, local amplitude spectrum and local phase spectrum;

[0019] Step 3.2, time domain filtering: Calculate the phase difference based on the obtained local phase spectrum, manually set the frequency range and filter, and then perform time domain bandpass filtering to extract the phase difference signal in the frequency band of interest;

[0020] Step 3.3, amplifying the motion signal: manually setting the amplification factor, linearly amplifying the phase difference of the motion signal of interest extracted by time domain filtering, and obtaining the amplified motion signal;

[0021] Step 3.4, video reconstruction and synthesis: Reconstruct the digital image data after motion amplification using the image sequence synthesized by the amplified motion signal, the high-pass residual and the low-pass residual obtained by spatial domain filtering, and finally synthesize the amplified video.

[0022] Furthermore, the step 4 uses a sub-pixel template matching algorithm to determine the displacement time-course response data corresponding to the motion-amplified video, including:

[0023] The amplified video is processed by frame division to obtain an amplified video image. The zero-mean normalized cross-correlation theory in statistics is used to calculate the sub-pixel displacement between the first frame amplified video image and the other frames amplified video images through a sub-pixel template matching algorithm. The processing results of each frame are then connected in series based on time to obtain the vibration displacement time-history response data of the target object cable.

[0024] Furthermore, in step 5, the fundamental frequency corresponding to the cable to be tested is obtained based on the displacement time-history response data, and the cable force is calculated by selecting a suitable cable force-frequency relationship according to the actual boundary conditions of the cable, including:

[0025] The obtained displacement time-history response data is used to obtain the fundamental frequency corresponding to the to-be-detected cable in the micro-vibration video by fast Fourier transform;

[0026] The practical formulas for calculating cable tension using the vibration method are all derived from simplified theoretical models and have different scopes of application. It is necessary to select the appropriate cable tension-frequency relationship to calculate the cable tension based on the actual situation of the cable.

[0027] Second aspect: This application discloses a cable force detection system based on computer vision, comprising:

[0028] A cable related information acquisition module, used to acquire relevant information of the cable to be detected;

[0029] An image acquisition module, used to acquire a micro-vibration video of the cable to be detected, and acquire a target image based on the micro-vibration video;

[0030] An image preprocessing module is used to preprocess the target image, including cropping, rotating and scaling the target image, and removing large noise in the target image to highlight the cable structure of interest and reduce the influence of other environmental factors;

[0031] The motion magnified video acquisition module is used to use a complex adjustable pyramid to perform spatial domain filtering on the input target image to obtain image sequences of different directions and scales, manually set the frequency range to perform time domain filtering on the image sequence to extract the phase difference information of the motion signal of interest, amplify the phase difference by a preset amplification factor to obtain the amplified image sequence, and then reconstruct the amplified video by the pyramid inverse process;

[0032] A displacement time-history response data determination module, used to determine the displacement time-history response data corresponding to the motion-amplified video using a sub-pixel template matching algorithm;

[0033] The cable force calculation module is used to obtain the fundamental frequency corresponding to the cable to be tested based on the displacement time-history response data, and select a suitable cable force-frequency relationship to calculate the cable force according to the actual boundary conditions of the cable.

[0034] The third aspect: The present application discloses a cable tension detection device based on computer vision, which is characterized by comprising: a memory and a processor, wherein the memory is used to store a computer program executable by the processor, and the processor is used to implement the steps of a cable tension detection method based on computer vision described in the present invention when executing the computer program.

[0035] Fourth aspect: The present application discloses a computer-readable storage medium, characterized in that a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the steps of a cable force detection method based on computer vision described in the present invention are implemented.

[0036] Beneficial effects: Compared with the prior art, the present invention has the following beneficial effects:

[0037] The present invention discloses a cable force detection method, system and device based on computer vision, which overcomes the shortcomings of the traditional cable force test frequency method, such as the difficulty in sensor installation, high equipment cost and low test efficiency. The method obtains the relevant information of the cable to be detected; collects the tiny vibration video of the cable to be detected, and pre-processes the video frame by frame to obtain the target image; uses the phase-based Euler motion amplification algorithm to obtain the amplified image sequence for the target image, and reconstructs the amplified video; uses the sub-pixel template matching algorithm to determine the displacement time-history response data corresponding to the motion amplified video; obtains the fundamental frequency corresponding to the cable to be detected based on the displacement time-history response data, and selects the appropriate cable force-frequency relationship to calculate the cable force according to the actual boundary conditions of the cable, providing a new technical approach for cable-stayed bridge cable force testing. The method realizes the detection of cable force with very small amplitude of bridge cables, and does not require additional sensors, which reduces the cost. The method has the advantages of low cost, convenience, easy operation, high scalability and accurate cable force detection, and has broad application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 A flow chart of a cable force detection method based on computer vision provided in this application;

[0039] Figure 2 A schematic diagram of cable force detection based on computer vision provided in this application.

[0040] Figure 3 A schematic diagram of the structure of a cable force detection system based on computer vision provided in this application. DETAILED DESCRIPTION

[0041] The present invention is further described in detail below through specific embodiments in conjunction with the accompanying drawings.

[0042] The present invention discloses a cable force detection method based on computer vision, referring to Figure 1 , specifically including the following steps:

[0043] S1: Obtain relevant information of the cable to be tested;

[0044] S2: Collect the micro-vibration video of the cable to be tested, and pre-process the video frame by frame to obtain the target image;

[0045] S3: using a phase-based Euler motion magnification algorithm to obtain a magnified image sequence for the target image, and reconstructing a magnified video;

[0046] S4: Determine the displacement time-course response data corresponding to the motion-amplified video using a sub-pixel template matching algorithm;

[0047] S5: obtaining the fundamental frequency corresponding to the cable to be tested based on the displacement time-history response data, and selecting a suitable cable force-frequency relationship to calculate the cable force according to the actual boundary conditions of the cable.

[0048] Furthermore, in step S1, the relevant information includes the cable model, material, diameter, calculated length, line density, inclination angle, boundary conditions, use time, number or position, etc. Obtaining the relevant information of the cable is the basic information for calculating the cable force later.

[0049] Further, the specific process of step S2 is as follows:

[0050] S2.1: Select 1 / 4 of the cable as the detection area, and observe whether there are natural landmarks at this point. If not, set a target point at this point; fix the high-precision industrial camera in a suitable position, and check the lighting conditions on site to determine whether fill light is needed. If fill light is needed, turn on the fill light with its own power supply for lighting; adjust the camera position to ensure that the cable target can be fully captured by the camera and will not exceed the video range when vibrating; adjust the lens focal length and aperture to ensure that the cable image is clearly visible in the camera field of view; set the camera sampling frequency to collect the cable's tiny vibration video;

[0051] S2.2: Decompose the cable micro-vibration video captured by the camera into continuous frame images, and set the same region of interest for each frame image; the region of interest includes the target point and the cable movement area to be detected;

[0052] S2.3: Preprocess the disassembled continuous frame images frame by frame, including compressing the digital image, cropping, rotating and scaling the target video image sequence, and removing large noise in the target image to highlight the cable structure of concern and reduce the influence of other environmental factors.

[0053] Specifically, compressing digital images means that in order to reduce the amount of computation required for computer processing of images, the three primary color information is often converted into grayscale information according to certain rules, that is, grayscale conversion is performed on the original image data to obtain grayscale image data, which can significantly reduce the amount of computation. The calculation formula for converting the three primary color image into a grayscale image according to different weight coefficients is:

[0054] H(x,y)=0.299R(x,y)+0.587G(x,y)+0.114B(x,y)

[0055] Further, the specific method of step S3 is as follows:

[0056] S3.1, spatial domain filtering: Use a complex controllable pyramid to filter the input target image in the spatial domain, and obtain high-pass residuals, low-pass residuals, and image sequences of different scales and directions, namely local amplitude spectrum and local phase spectrum; the complex controllable pyramid is a multi-directional, multi-scale, self-conversion multi-resolution image decomposition algorithm with advantages such as directional controllability and translation invariance. Its basis function is similar to a sine function composed of a Gaussian function envelope. It is worth noting that the high-pass residuals and low-pass residuals output by this step are not subjected to time domain filtering and motion signal amplification processing, but are directly used for pyramid reconstruction of motion-amplified image data.

[0057] S3.2, time domain filtering: Calculate the phase difference based on the obtained local phase spectrum, manually set the frequency range and select the filter, and then perform time domain bandpass filtering to extract the phase difference signal in the frequency band of interest;

[0058] S3.3, amplifying the motion signal: manually setting the amplification factor α, linearly amplifying the phase difference of the motion signal of interest extracted by time domain filtering, and obtaining the amplified motion signal;

[0059] S3.4, video reconstruction and synthesis: the image sequence synthesized by the amplified motion signal, the high-pass residual and low-pass residual obtained by spatial domain filtering are reconstructed to obtain the digital image data after motion amplification, and finally the amplified video is synthesized.

[0060] Specifically, motion magnification is to amplify the small motions in the video sequence or image sequence so that the amplitude of these small motions is increased so as to extract valuable information from these small motions. The phase-based Euler motion magnification algorithm is proposed based on the Fourier shift theorem. In a two-dimensional image, the phase corresponds to the movement of the object. The image is Fourier transformed and the phase contained is processed to achieve object motion magnification. This method directly operates on the phase information contained in the image. In terms of noise processing, it only translates the noise without amplifying it, reduces the appearance of motion artifacts, and supports a larger magnification factor.

[0061] In order to explain the principle of the algorithm intuitively, we take a one-dimensional image as an example. Assume that f(x) is the brightness function of the one-dimensional image, x is the pixel coordinate of the image, and after time t, assume that the object is translated by δ(t) on the image, then the image brightness at time t is f(x+δ(t)). Perform Fourier transform on f(x) and f(x+δ(t)) respectively, and we get

[0062]

[0063]

[0064] Where ω is the harmonic frequency, A ω is the harmonic amplitude.

[0065] For a certain harmonic frequency ω, the phase difference between the harmonic components of f(x) and f(x+δ(t)) is

[0066]

[0067] Obviously, this phase difference is directly related to the signal δ(t) and contains motion information.

[0068] If the phase is magnified by α times, the corresponding harmonic component is adjusted to, and the reconstructed image brightness function is

[0069]

[0070] By comparing the image brightness function f(x) and f(x+δ(t)), the amplified motion signal (1+αδ(t)) can be obtained. It is worth noting that in order to amplify the motion signal within a certain frequency range, δ(t) is actually filtered in the time domain.

[0071] Furthermore, the step 4 uses a sub-pixel template matching algorithm to determine the displacement time-course response data corresponding to the motion-amplified video, including:

[0072] The amplified video is processed by frame division to obtain the amplified video image, and the zero-mean normalized cross-correlation theory in statistics is used to calculate the sub-pixel displacement between the amplified video image of the first frame and the amplified video image of other frames through the sub-pixel template matching algorithm, and then the processing results of each frame are connected in series with time as the baseline, so as to obtain the vibration displacement time-history response data of the target object cable. And taking a single inclined cable as an example, its vibration in space mainly occurs in three directions: the chord direction in the vertical plane, the vertical chord direction in the vertical plane, and the out-of-plane direction. In general, the vibration amplitude of the cable along the chord direction is much smaller than the vibration amplitude in the other two directions. Considering that the aerodynamic damping of the bridge inclined cable in the vertical chord direction is usually about half of the lateral direction, it is believed that the main vibration direction of the bridge cable is the vertical chord direction in the vertical plane. Therefore, the present invention mainly studies the use of computer vision methods to test the vibration characteristics of the inclined cable in the vertical chord direction in the vertical plane.

[0073] Specifically, the displacement calculation process of sub-pixel template matching is as follows: Assume that there are two images f(x, y) and h(x, y) of the same size (M×N), where h(x, y) has a relative translation with the reference image f(x, y). The cross-correlation relationship between f(x, y) and h(x, y) after Fourier transform can be defined as:

[0074]

[0075] Where: M and N are the image sizes; (x0, y0) is the amount of coordinate shift; “*” indicates complex conjugate; F(u, v) and H * (u,v) denotes the discrete Fourier transform of f(x,y) and h(x,y), respectively.

[0076] The expression of F(u,v) is:

[0077]

[0078] By R hf The peak value of the structure vibration is extracted from the pixel-level displacement. Then, in R hf The sub-pixel displacement of the structural vibration is extracted by performing discrete Fourier transform cross-correlation based on time-sensitive matrix multiplication in the area near the initial peak.

[0079] In particular, it is worth noting that only frequency information is needed to estimate the cable force. In other words, there is no need to determine a scaling factor to convert pixel coordinate vibration displacement to physical coordinate vibration displacement, which will make the vision-based measurement process more efficient and practical.

[0080] Furthermore, in step 5, the fundamental frequency corresponding to the cable to be tested is obtained based on the displacement time-history response data, and the cable force is calculated by selecting a suitable cable force-frequency relationship according to the actual boundary conditions of the cable, including:

[0081] The obtained displacement time-history response data is used to obtain the fundamental frequency corresponding to the cable to be tested in the tiny vibration video using fast Fourier transform; the first-order vibration frequency of the cable, namely the fundamental frequency, is an important dynamic characteristic of the cable; at the same time, engineers who test cable tension on site are accustomed to using the fundamental frequency to calculate the cable tension, so the present invention calculates the cable tension by obtaining an accurate fundamental frequency.

[0082] The practical formulas for calculating cable tension using the vibration method are derived from simplified theoretical models and have different scopes of application. It is necessary to select the appropriate cable tension-frequency relationship to calculate the cable tension according to the actual situation of the cable. At present, the cable tension calculation models based on vibration are mainly divided into four categories: tension string model theory, simply supported beam model theory, fixed beam theory, and complex boundary model theory.

[0083] Tension string model theory: The tension string theory simplifies the inclined cable into a tension string with no stiffness and no sag, and ignores the bending stiffness.

[0084] F=4ml 2 f 2

[0085] Simply supported beam model theory: The simply supported beam model simplifies the inclined cable into a horizontal beam with one end hinged and the other end simply supported.

[0086]

[0087] Theory of the fixed beam model: The fixed beam model uses fixed constraints at both ends, which is different from the simply supported beam model in terms of boundary conditions. Considering the influence of bending stiffness, the calculation formula is as follows:

[0088]

[0089]

[0090] F=4ml 2 f 2 (210≤ξ)

[0091] Considering the influence of sag, the calculation formula is as follows:

[0092] F=4ml 2 f 2 (λ 2 ≤0.17,4π 2 ≤λ 2 )

[0093]

[0094] Complex boundary model theory: In actual application, the boundary conditions of the cable-stayed cable are relatively complex, which may be between simply supported and fixedly supported, and there is an elastic boundary.

[0095]

[0096] Where F is the mass per unit length of the cable m; l is the length of the cable; f is the first-order vibration frequency of the cable, i.e., the fundamental frequency; EI represents the bending stiffness of the cable; EA is the axial stiffness of the cable; ξ is the dimensionless quantity of the bending stiffness, λ 2 is the dimensionless quantity of sag, λ 2 =[mgl / F]·(EA1 / FL e );L e =l[(1+(mglcosθ / F 2 / 8))];

[0097] The method obtains relevant information of the cable to be tested; collects a tiny vibration video of the cable to be tested, and preprocesses the video frame by frame to obtain a target image; uses a phase-based Euler motion amplification algorithm to obtain an amplified image sequence for the target image, and reconstructs the amplified video; uses a sub-pixel template matching algorithm to determine the displacement time-history response data corresponding to the motion amplified video; obtains the fundamental frequency corresponding to the cable to be tested based on the displacement time-history response data, and selects a suitable cable force-frequency relationship to calculate the cable force according to the actual boundary conditions of the cable, thereby providing a new technical approach for cable force testing of cable-stayed bridges.

[0098] The cable force detection system based on computer vision described in the present invention refers to Figure 2 ,include:

[0099] A cable related information acquisition module, used to acquire relevant information of the cable to be detected;

[0100] An image acquisition module, used to acquire a micro-vibration video of the cable to be detected, and acquire a target image based on the micro-vibration video;

[0101] An image preprocessing module is used to preprocess the target image, including cropping, rotating and scaling the target image, and removing large noise in the target image to highlight the cable structure of interest and reduce the influence of other environmental factors;

[0102] The motion magnified video acquisition module is used to use a complex adjustable pyramid to perform spatial domain filtering on the input target image to obtain image sequences of different directions and scales, manually set the frequency range to perform time domain filtering on the image sequence to extract the phase difference information of the motion signal of interest, amplify the phase difference by a preset amplification factor to obtain the amplified image sequence, and then reconstruct the amplified video by the pyramid inverse process;

[0103] A displacement time-history response data determination module, used to determine the displacement time-history response data corresponding to the motion-amplified video using a sub-pixel template matching algorithm;

[0104] The cable force calculation module is used to obtain the fundamental frequency corresponding to the cable to be tested based on the displacement time-history response data, and select a suitable cable force-frequency relationship to calculate the cable force according to the actual boundary conditions of the cable.

[0105] It should be noted that for more specific working processes of the above modules, reference may be made to the corresponding contents disclosed in the aforementioned embodiments, which will not be repeated here.

[0106] The present application discloses a cable tension detection device based on computer vision, which is characterized by comprising: a memory and a processor, wherein the memory is used to store a computer program executable by the processor, and the processor is used to implement the steps of a cable tension detection method based on computer vision described in the present invention when executing the computer program.

[0107] The present application discloses a computer-readable storage medium, characterized in that a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the steps of a cable force detection method based on computer vision described in the present invention are implemented.

[0108] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention belong to the scope of the present invention and its equivalent technology, the present invention is also intended to include these changes and variations.

Claims

1. A cable force detection method based on computer vision, characterized in that: The specific steps include: Step 1: Obtain relevant information of the cable to be tested; Step 2: Collect the micro-vibration video of the cable to be tested, and pre-process the video frame by frame to obtain the target image; Step 3: Using the phase-based Euler motion magnification algorithm to obtain the magnified image sequence, and reconstruct the magnified video; the specific steps of the phase-based Euler motion magnification algorithm are as follows: Step 3.1, spatial domain filtering: Use the complex controllable pyramid to perform spatial domain filtering on the input target image to obtain high-pass residuals, low-pass residuals, and image sequences of different scales and directions, namely, local amplitude spectrum and local phase spectrum; Step 3.2, time domain filtering: Calculate the phase difference based on the obtained local phase spectrum, manually set the frequency range and filter, and then perform time domain bandpass filtering to extract the phase difference signal in the frequency band of interest; Step 3.3, amplifying the motion signal: manually setting the amplification factor, linearly amplifying the phase difference of the motion signal of interest extracted by time domain filtering, and obtaining the amplified motion signal; Step 3.4, video reconstruction and synthesis: reconstruct the image sequence after motion amplification by using the image sequence synthesized by the amplified motion signal, the high-pass residual and the low-pass residual obtained by spatial domain filtering, and finally synthesize the amplified video; Step 4: using a sub-pixel template matching algorithm to determine the displacement time-history response data corresponding to the motion amplified video; including performing frame processing on the amplified video to obtain an amplified image sequence, using the zero-mean normalized cross-correlation theory in statistics, and calculating the sub-pixel displacement between the target image after the first frame is amplified and the target images after the other frames are amplified by the sub-pixel template matching algorithm, and then connecting the processing results of each frame of the picture in series based on time, so as to obtain the vibration displacement time-history response data of the target object cable; Step 5: Based on the displacement time-history response data, the fundamental frequency corresponding to the cable to be tested is obtained, and according to the actual boundary conditions of the cable, a suitable cable force-frequency relationship is selected to calculate the cable force.

2. The cable force detection method based on computer vision according to claim 1, characterized in that: In step 1, the relevant information includes the model, material, diameter, calculated length, line density, inclination angle, boundary conditions, usage time, number or position of the cable.

3. The cable force detection method based on computer vision according to claim 1, characterized in that: The specific process of step 2 is as follows: Step 2.1: Select 1 / 4 of the cable as the detection area, and observe whether there are natural markers at this point. If not, set a target point at this point; fix the high-precision industrial camera in a suitable position, and check the lighting conditions on site to determine whether fill light is needed. If fill light is needed, turn on the fill light with its own power supply for lighting; adjust the camera position to ensure that the cable target can be fully captured by the camera and will not exceed the video range when vibrating; adjust the lens focal length and aperture to ensure that the cable image is clearly visible in the camera field of view; set the camera sampling frequency to collect the cable's tiny vibration video; Step 2.2: Decompose the cable micro-vibration video captured by the camera into continuous frame images, and set the same region of interest for each frame image; the region of interest includes the target point and the cable movement area to be detected; Step 2.3: Preprocess the disassembled continuous frame images frame by frame, including cropping, rotating and scaling the target image sequence, and removing noise from the target image to highlight the cable structure of interest and reduce the influence of other environmental factors.

4. The cable force detection method based on computer vision according to claim 1, characterized in that: In step 5, the fundamental frequency corresponding to the cable to be tested is obtained based on the displacement time-history response data, and the cable force is calculated by selecting a suitable cable force-frequency relationship according to the actual boundary conditions of the cable, including: The obtained displacement time-history response data is used to obtain the fundamental frequency corresponding to the to-be-detected cable in the micro-vibration video by fast Fourier transform; The practical formulas for calculating cable tension using the vibration method are all derived from simplified theoretical models and have different scopes of application. It is necessary to select the appropriate cable tension-frequency relationship to calculate the cable tension based on the actual situation of the cable.

5. A cable force detection system based on computer vision, characterized in that: include: A cable related information acquisition module, used to acquire relevant information of the cable to be detected; An image acquisition module, used to acquire a micro-vibration video of the cable to be detected, and acquire a target image based on the micro-vibration video; An image preprocessing module, used for preprocessing the target image, including cropping, rotating and scaling the target image, and removing noise from the target image to highlight the cable structure of interest and reduce the influence of other environmental factors; The motion magnified video acquisition module is used to use a complex adjustable pyramid to perform spatial domain filtering on the input target image to obtain image sequences of different directions and scales, manually set the frequency range to perform time domain filtering on the image sequence to extract the phase difference information of the motion signal of interest, amplify the phase difference by a preset amplification factor to obtain the amplified image sequence, and then reconstruct the amplified video by the pyramid inverse process; A displacement time-history response data determination module, used to determine the displacement time-history response data corresponding to the motion-amplified video using a sub-pixel template matching algorithm; The cable force calculation module is used to obtain the fundamental frequency corresponding to the cable to be detected based on the displacement time-history response data, and select a suitable cable force-frequency relationship to calculate the cable force according to the actual boundary conditions of the cable.

6. A cable force detection device based on computer vision, characterized in that: include: A memory and a processor, wherein the memory is used to store a computer program executable by the processor, and the processor is used to implement the steps of a cable force detection method based on computer vision as described in any one of claims 1 to 4 when executing the computer program.

7. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by the processor, the steps of the cable force detection method based on computer vision described in any one of claims 1 to 4 are implemented.

Images