Phase-based method and system for detecting micro-vibration frequency of bridge cable

Abstract

The application discloses a bridge cable micro-vibration frequency detection method and system based on phase, and belongs to the technical field of image processing. The method is: collecting cable video under natural excitation; in the spatial domain, performing spatial pyramid decomposition on the collected video image frame to obtain image phase information, combining the cable edge features in the phase space, and using bilateral filtering and phase amplification to enhance the cable phase information; in the time domain, the video image frame is divided into several sub-regions, the time domain signal of each sub-region is extracted, different weight values are given according to the contribution rate of the cable signal, and finally the final cable domain vibration signal is synthesized through weighted summation to calculate the cable signal frequency. The application enhances the cable phase signal in the spatial domain and designs a new signal weighting mechanism in the time domain, enhances the cable phase information and reduces the noise phase information, superimposes the sub-region signals to synthesize a pure cable signal, and improves the accuracy of calculating the cable frequency.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and more specifically, to a phase-based method and system for detecting minute vibration frequencies of bridge cables. Background Technology

[0002] With rapid urbanization, bridges, as a crucial component of urban infrastructure, are vital to public safety due to their structural health. Bridge cables, as key supporting elements of the bridge structure, exhibit minute variations in their vibration frequencies, which can serve as sensitive indicators for assessing the structural health of a bridge. In the field of bridge monitoring, cable vibration frequency detection is one of the key aspects of ensuring the safe operation of bridges.

[0003] Currently, commonly used methods for detecting cable vibration frequencies include traditional sensor monitoring methods such as accelerometers, hydraulic gauges, magnetic flux sensors, and tension gauges, as well as microwave radar methods and video monitoring methods based on computer vision technology. However, cable vibrations under natural excitation are extremely small, making it difficult to extract the cable vibration signal.

[0004] Existing literature proposes methods such as deep learning, enhanced feature point detection, and cable straightness detection. Among related technologies, for example, Chinese patent document CN110411686B provides a method for monitoring and diagnosing the holographic health status of bridges using static and dynamic images, which obtains the changing patterns of the bridge's structural state based on machine deep learning of historical monitoring data; another example is Chinese patent document CN114528887A, which provides a bridge monitoring method based on micro-vibration amplification technology, monitoring the bridge condition by superimposing amplified vibration signal images. However, these methods cannot accurately measure the vibration frequency of cables under natural excitation.

[0005] As can be seen from the above, there is a serious problem with the existing technology: it cannot accurately extract the minute vibration signals of the cable under natural excitation, which leads to the failure of cable frequency detection. Summary of the Invention

[0006] 1. Technical problems to be solved

[0007] To address the problem in existing technologies that fail to accurately extract minute vibrations of cables under natural excitation, leading to the failure of cable frequency detection, this invention provides a phase-based method and system for detecting minute vibration frequencies of bridge cables. This method can filter cable signals and noise signals under natural excitation to extract weak vibration signals and frequencies of the cable structure.

[0008] 2. Technical Solution

[0009] The objective of this invention is achieved through the following technical solutions.

[0010] A phase-based method for detecting minute vibration frequencies in bridge cables includes the following steps:

[0011] Video capture: Capture vibration videos of the object being tested;

[0012] Video image region segmentation and preprocessing: The frame images of the acquired video are segmented and filtered;

[0013] Image phase information extraction: Spatial pyramid decomposition is performed on the segmented and processed image to obtain the image's phase information;

[0014] Image cable phase enhancement: Based on the spatial differences between cable phase and noise phase, a phase enhancement method combining bilateral filtering and phase amplification is designed to enhance the cable phase and reduce the noise phase.

[0015] Image temporal signal extraction: Based on the acquired enhanced phase information, the temporal information of each sub-region is calculated using a vibration analysis-based method to obtain the temporal vibration signal of the sub-region;

[0016] Sub-region signal weighted summation: Based on the contribution of each sub-region signal to the cable signal, a maximum contribution combination algorithm (MCC) is designed to assign different weight values ​​to each sub-region signal according to its contribution. Finally, all sub-region signals are weighted and summed to synthesize the time-domain vibration signal of the cable.

[0017] Signal frequency calculation: The frequency value of the object being detected is calculated based on the time-domain vibration signal of the synthesized object.

[0018] Furthermore, the specific steps for video image region division and preprocessing are as follows: the length and width of each frame of video image are divided into n and m equal parts, respectively, that is, into n parts and m parts, where m and n are natural numbers greater than 1.

[0019] Define a i b j Let be the image region located in the i-th row and j-th column, where i = 1…m, j = 1…n;

[0020] A two-dimensional Gaussian filter is applied to each image region to remove low-frequency noise caused by camera electronic imaging. The two-dimensional Gaussian function G(x,y) is expressed as:

[0021]

[0022] σ is the standard deviation of the function, (x,y) are the graph coordinates, and e is the natural constant.

[0023] Furthermore, the specific steps for extracting image phase information are as follows: spatial pyramid decomposition is performed on each image region to obtain the image's phase information; this is achieved by convolving the image with two-dimensional Gabor functions of different scales and orientations.

[0024]

[0025] in:

[0026]

[0027] x θ =xcosθ + ysinθ, y θ = -xsinθ + ycosθ

[0028] g(x,y;λ,θ,ψ,σ,γ) is a two-dimensional Gabor function, where λ is the wavelength, ψ is the phase shift of the complex exponential function, γ is the scaling factor controlling the shape of the Gabor function, θ is the direction angle of the Gabor function, and x... θ and y θ Let (x,y) be the coordinates after rotation, ξ and η be time-dependent variables and subsets of x and y respectively, I(x,y,t0) be the intensity value of image pixel (x,y) at time t0, and I′(x,y,t0) be the intensity value result after pyramid decomposition, which can be regarded as the phase value of pixel at time t0.

[0029] The phase information at this point includes the phase of the cable and the phase of the noise, that is:

[0030] I′(x,y,t0)=I′(x,y,t0) C +I′(x,y,t0) N

[0031] Where I′(x,y,t0) C And I′(x,y,t0) N These represent the cable phase and the noise phase, respectively. To enhance the cable phase while reducing the noise phase, we perform bilateral filtering on the mixed phase signal, i.e.:

[0032]

[0033] Where, Φ ω (x,t) represents the filtered phase, W p (x,t) represents the sum of weights for each phase value in the filter window, used for weight normalization. G u and G rThese are the phase similarity weights and spatial distance weights, respectively. After phase filtering, the phase information at the cable edge is selectively protected, while noise-related phase interference at the far edge is suppressed. That is, I′(x,y,t0). C It remains unchanged, while I′(x,y,t0) N Reduced. Φ ω (x,t) contains motion information. Motion change information can be obtained by calculating the phase difference between two frames, where the phase difference between the two frames is:

[0034] Φ ω (x,t) dif =Φ ω (x,t)-Φ ω (x,0)

[0035] Φ ω (x,0) is the phase value at t=0. Then we use the phase difference signal Φ ω (x,t) dif An amplification factor α is applied to enhance the phase difference signal, i.e.:

[0036] Φ ω (x,t) mag =α·Φ ω (x,t) dif

[0037] Φ ω (x,t) mag This is the enhanced cable phase.

[0038] Furthermore, the specific steps for extracting the temporal signal of the image sub-region are as follows: after obtaining the enhanced image phase information in the phase space, the temporal vibration signal of the image sub-region is extracted using a vibration analysis method based on video phase.

[0039] y(t)={y1(t),y2(t),y3(t)…y s (t)}

[0040] y s (t) represents the s-th (s∈1,2,…,a) pixel in the image. i b jThe traditional signal processing method involves averaging and weighting the time-domain vibration signals from all regions, assigning the same weight to all regions, and then summing the results to obtain the final target signal. However, since the vibration signals in each region contribute differently to the cable signal, the average weighting method cannot maximize noise removal and target signal enhancement. Therefore, this invention designs a Maximum Contribution Combination (MCC) algorithm, which assigns different weights to each sub-region signal according to its contribution to the cable signal, and finally sums the weighted signals from all sub-regions to synthesize the cable's time-domain vibration signal.

[0041] Furthermore, the time-domain signal of each sub-region can be represented as a combination of cable signal and noise signal, i.e.:

[0042] y s (t)=k s ·r(t)+n s (t)

[0043] k s Let n be the contribution weight of the cable signal r(t) in region s. s (t) represents the noise signal in region s. s (t) can be viewed as different channel signals with different intensities and different noise levels that receive the same desired signal t(t) from the same cable. Then, we use contribution-weighted summation to combine all these different channel signals.

[0044]

[0045] Where r′(t) is the estimated signal of the final target signal after weighted summation. A s Let be the signal weight value for region s.

[0046] Furthermore, assume that the signal k in each region s r(t) and noise n s (t) is independent, and k is independent of each region. s • r(t) is continuous, meaning there is no time delay between the cable signals r(t) extracted from different regions, and r(t) is independent of the noise component n. s (t) is irrelevant. This is because the main spectrum of the cable signal r(t) is concentrated within a narrow frequency range, while the noise signal n... s The spectrum of (t) is distributed within the [cHz, dHz] frequency band. Based on this spectrum distribution characteristic, we will assign weights A... s Defined as signal y within the main vibration frequency range s Power spectral density (PSD) of Y(t) i(f) The ratio of the power spectral density (PSD) to the noise within the frequency band [c Hz, d Hz]. That is:

[0047]

[0048] Where Peak is the main vibration frequency of the cable, [Peak-τ, Peak+τ] is a small frequency band centered on Peak, and τ ranges from (0,1). Since the cable vibration r(t) under natural excitation is usually very small, we exclude signals from regions with abnormally large signals. Large signal variations are mainly caused by the movement of objects in the background (such as swaying trees, walking pedestrians, etc.). Therefore, we remove all signals with amplitudes greater than the threshold a. th In the region where the last set of video image sequences is synthesized from the cable signals, the formula for the time-domain vibration signal r(t) becomes:

[0049]

[0050]

[0051] y max,s and y min,s y s The maximum and minimum values ​​in the (t) signal.

[0052] Furthermore, the specific steps for calculating the signal frequency are as follows: the frequency value of the detected object can be calculated by performing a Fourier transform on the synthesized time-domain vibration signal of the above-mentioned object.

[0053] The system based on the phase-based method for detecting minute vibration frequencies of bridge cables described above,

[0054] Includes a video acquisition module for capturing vibration videos of the object being tested;

[0055] The video image region segmentation and preprocessing module is used to segment and filter the frame images of the acquired video.

[0056] The image phase information extraction module is used to perform spatial pyramid decomposition on the segmented and processed image and obtain the image's phase information;

[0057] The image phase information enhancement module is used to enhance the cable phase signal while reducing noise signals;

[0058] The image time-domain signal extraction module is used to calculate the time-domain vibration signal based on the acquired image enhancement phase information using a vibration analysis-based method.

[0059] The regional weight allocation and weighted summation module is used to determine the contribution of each region to the cable signal and to calculate the final cable signal by weighting the signals of all regions with different weights.

[0060] The signal frequency calculation module is used to perform Fourier transform on the time-domain vibration signal of the synthesized detection object to calculate the frequency value of the detection object.

[0061] 3. Beneficial effects

[0062] Compared with the prior art, the advantages of this invention are:

[0063] The present invention relates to a phase-based method and system for detecting minute vibration frequencies of bridge cables. Based on the difference in edge characteristics between cable signals and noise signals in the spatial domain, a cable edge phase enhancement method combining bilateral filtering and phase amplification is designed. This method can effectively improve the identification of cable phase information and reduce the interference of noise signals. Simultaneously, in the time domain, a maximum contribution combination algorithm (MCC) is designed to assign different weight values ​​to each sub-region signal according to its contribution to the cable signal. Finally, all sub-region signals are weighted and summed to synthesize the time-domain vibration signal of the cable, which helps to improve the purity of the cable signal and thus improve the accuracy of cable vibration frequency calculation. Attached Figure Description

[0064] Figure 1 This is a flowchart of a frequency measurement method in one embodiment of the present invention;

[0065] Figure 2 This is a schematic diagram of cable phase enhancement (PE) in one embodiment of the present invention;

[0066] Figure 3 This is a schematic diagram of the Region Maximum Contribution Combination (MCC) algorithm in one embodiment of the present invention;

[0067] Figure 4 This is a measured scene diagram of a bridge cable in one embodiment of the present invention;

[0068] Figure 5 This is a schematic diagram comparing the measured experimental results of bridge cables in one embodiment of the present invention. Detailed Implementation

[0069] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments.

[0070] Example 1

[0071] Combination Figures 1 to 5 The phase-based method for detecting minute vibration frequencies of bridge cables of the present invention includes the following steps:

[0072] Video capture:

[0073] The camera is used to capture video of the vibration of the object being detected in a real-world scenario. Specifically, in this embodiment, the object being detected is a bridge cable.

[0074] The camera frame rate is adjusted according to the actual situation. To satisfy the Nyquist sampling theorem, the camera frame rate should be greater than twice the maximum vibration frequency of the detected object. In this embodiment, a camera frame rate of 50fps is used, and the video frame rate can be between 750 and 1000 frames. In this embodiment, it is saved as an AVI video. Generally speaking, the acquired video data can be in common video formats such as AVI, MOV, and MP4.

[0075] This step can obtain a video image sequence containing the detected object and the background.

[0076] Video image region segmentation and preprocessing:

[0077] like Figure 3 As shown in (b), each frame of the video image is divided into n and m equal parts in length and width, respectively, where m and n are natural numbers greater than 1. Each video region can be represented as a. i b j (i=1...m, j=1...n), that is, a i b j Let be the image region located in the i-th row and j-th column.

[0078] Because the partitioning is random, a sub-region of the image may contain the target object, or it may not, or it may only contain partial information about the target object. Partitioning the image into regions can improve the accuracy of frequency detection.

[0079] Then, a two-dimensional Gaussian filter is applied to each image region:

[0080]

[0081] This removes low-frequency noise caused by camera electronic imaging. Here, σ is the standard deviation of the function, (x,y) are the image coordinates, and e is the natural constant.

[0082] Image phase information extraction:

[0083] Spatial pyramid decomposition is performed on each image sub-region to obtain the image's phase information. This involves convolving the image with two-dimensional Gabor functions of different scales and orientations, i.e.:

[0084]

[0085] in:

[0086]

[0087] g(x,y;λ,θ,ψ,σ,γ) is a two-dimensional Gabor function, where λ is the wavelength, ψ is the phase shift of the complex exponential function, and γ is the scaling factor that controls the shape of the Gabor function.

[0088] in:

[0089] x θ =xcosθ + ysinθ, y θ = -xsinθ + ycosθ

[0090] θ is the direction angle of the Gabor function, and (x,y) are the image coordinates. In this embodiment, the direction angle θ = 0; x θ and y θ Let (x,y) be the rotated coordinates, ξ and η be time-dependent variables and subsets of x and y, respectively, and let I(x,y,t0) represent the intensity value of image pixel (x,y) at time t0; I′(x,y,t0) represent the intensity value result after pyramid decomposition. This step obtains the required amplitude and phase information in the image spatial domain by performing complex controllable pyramid decomposition on the image. The phase information after decomposition is as follows: Figure 2 As shown in (b).

[0091] Image phase information enhancement:

[0092] The phase information at this point includes the phase of the cable and the phase of the noise, that is:

[0093] I′(x,y,t0)=I′(x,y,t0) C +I′(x,y,t0) N

[0094] Where I′(x,y,t0) C And I′(x,y,t0) N These represent the cable phase and the noise phase, respectively. To enhance the cable phase while reducing the noise phase, we perform bilateral filtering on the mixed phase signal, i.e.:

[0095]

[0096] Where, Φ ω (x,t) is the filtered phase, W p (x,t) is the sum of weights for each phase value in the filter window, used for weight normalization. G u and G r These are the phase similarity weights and spatial distance weights, respectively. After phase filtering, the phase information at the cable edge is selectively protected, while noise-related phase interference at the far edge is suppressed. That is, I′(x,y,t0).C It remains unchanged, while I′(x,y,t0) N Reduced. The result after phase filtering is as follows: Figure 2 As shown in (c). Φ ω (x,t) contains motion information. Motion change information can be obtained by calculating the phase difference between two frames, where the phase difference between the two frames is:

[0097] Φ ω (x,t) dif =Φ ω (x,t)-Φ ω (x,0)

[0098] Φ ω (x,0) is the phase value at t=0. Then we use the phase difference signal Φ ω (x,t) did An amplification factor α is applied to enhance the phase difference signal, i.e.:

[0099] Φ ω (x,t) mag =α·Φ ω (x,t) dif

[0100] Φ ω (x,t) mag For the enhanced cable phase, such as Figure 2 As shown in (d).

[0101] Image temporal signal extraction:

[0102] After obtaining the enhanced image phase information in the phase space, the temporal vibration signals of each sub-region of the image are extracted using a vibration analysis method based on video phase:

[0103] y(t)={y1(t),y2(t),y 3(t) …y s (t)}

[0104] t s (t) represents the s-th (s∈1,2,…,a) pixel in the image. i b jThe traditional signal processing method involves averaging and weighting the time-domain vibration signals from all regions, assigning the same weight to all regions, and then summing the results to obtain the final target signal. However, since the contribution of the vibration signals from each region to the cable signal varies, the average weighting method cannot maximize noise removal and target signal enhancement. Therefore, this invention designs a Maximum Contribution Combination (MCC) algorithm, which assigns different weights to each sub-region signal according to its contribution to the cable signal, and finally sums the weighted signals from all sub-regions to synthesize the cable's time-domain vibration signal, as shown below. Figure 3 As shown;

[0105] Regional weight allocation and weighted summation module:

[0106] Furthermore, the time-domain signal of each sub-region can be represented as a combination of cable signal and noise signal, i.e.:

[0107] y s (t)=k s ·r(t)+n s (t)

[0108] k s Let n be the contribution weight of the cable signal r(t) in region s. s (t) represents the noise signal in region s. s (t) can be viewed as different channel signals with different intensities and different noise levels receiving the same desired signal r(t). Then, we use contribution-weighted summation to combine all these different channel signals, such as... Figure 3 As shown.

[0109]

[0110] Furthermore, assume that the signal k in each region s r(t) and noise n s (t) is independent, and k is independent of each region. s • r(t) is continuous, meaning there is no time delay between the cable signals r(t) extracted from different regions, and r(t) is independent of the noise component n. s (t) is irrelevant. This is because the main spectrum of the cable signal r(t) is concentrated within a narrow frequency range, while the noise signal n... s The spectrum of (t) is distributed within the [cHz, dHz] frequency band. Based on this spectrum distribution characteristic, we will assign weights A... s Defined as signal y within the main vibration frequency range s Power spectral density (PSD) of Y(t) i(f) The ratio of the power spectral density (PSD) to the noise within the frequency band [cHz, dHz]. That is:

[0111]

[0112] Where Peak is the main vibration frequency of the cable, [Peak-τ, Peak+τ] is a small frequency band centered on Peak, and the value of τ ranges from (0,1).

[0113] Peak can be estimated as follows: we take all A in the above formula. s The value is set to 1, and the synthesized cable signal r′(t) is then regarded as an approximate estimate of the real cable signal r(t). The peak value of the dominant frequency is obtained through FFT and denoted as Peak. The value of τ is selected in the range of 0 < τ < 1, and [c Hz, d Hz] is a wide frequency band that includes the dominant frequency of the cable and the noise signal. This invention selects [0.2Hz, 5.5Hz]. Since the cable vibration r(t) under natural excitation is usually very small, we exclude signals in areas with abnormally large signals. Large signal changes are mainly caused by the movement of objects in the background (such as swaying trees, walking pedestrians, etc.). Therefore, we remove all amplitudes greater than the threshold a. th In the region where the last set of video image sequences is synthesized from the cable signals, the formula for the time-domain vibration signal r(t) becomes:

[0114]

[0115]

[0116] y max,s and y min,s y s (t) The maximum and minimum values ​​of the signal. th The selection principle is based on the time-domain signal y s The point that is abnormally large in (t).

[0117] Signal frequency calculation:

[0118] The frequency value of the object can be calculated by performing a Fourier transform on the time-domain vibration signal r′(t) of the object obtained above.

[0119] Combination Figures 1 to 5 The phase-based method for detecting minute vibration frequencies of bridge cables of the present invention includes the following steps:

[0120] Step 1: Video Acquisition Module

[0121] Use a video capture device to record a video of the object to be measured, for approximately 15 seconds.

[0122] Step 2: Video Image Region Segmentation and Preprocessing Module

[0123] The video image sequence is divided into several small rectangular regions, each containing, not containing, or partially containing the target object. Gaussian filtering is then applied to each region of the video image sequence to smooth the image and remove noise.

[0124] Step 3: Image Phase Information Extraction Module

[0125] Spatial pyramids are used to spatially decompose each video image sequence in each region. By performing spatial bandpass filtering and downsampling on each frame, amplitude and phase spectra at different scales and in different directions are obtained. Then, amplitude weighting is applied to the phase spectrum to further filter out noise interference.

[0126] Step 4: Cable Phase Enhancement Module

[0127] Based on the difference in edge characteristics between cable signals and noise signals in the spatial domain, a cable edge phase enhancement method combining bilateral filtering and phase amplification is designed, which can effectively improve the identification of cable phase information and reduce the interference of noise signals.

[0128] Step 4: Image Time Domain Signal Extraction Module

[0129] Using a video phase-based vibration analysis method, the temporal vibration signal y of each region's image sequence is extracted. s (t), where s represents the s-th (s∈1,2,…,a) pixel in the image. i b j ) areas;

[0130] Step 5: Regional Weight Allocation and Weighted Summation Module

[0131] A maximum contribution combination (MCC) algorithm is designed to assign different weight values ​​to the signals of each sub-region according to their contribution to the cable signal. Finally, the signals of all sub-regions are weighted and summed to synthesize the final time-domain vibration signal of the cable. This method helps to improve the purity of the cable signal and thus improves the accuracy of the cable vibration frequency calculation.

[0132] Step Six: Signal Frequency Calculation Module

[0133] After obtaining the synthesized global vibration signal of the target object in the time domain, a Fourier transform is performed on the time domain signal to obtain the vibration frequency of the target object.

[0134] The actual measurement scenario of the bridge cables in this embodiment is as follows: Figure 4 As shown, the experimental results of the bridge cables are as follows: Figure 5 As shown. Selected as... Figure 4(a) shows 10 cables as the test objects, and their vibration frequencies under natural and intense conditions are measured, such as Figure 4 (b) and Figure 4 (c) Schematic diagrams show the acquisition of cable video using a camera and the acquisition of cable vibration signals using an accelerometer, respectively. Note that the acquisition is performed simultaneously. The Accelerometer method is used as a truth reference; the PVEDIR method (reference "Camera-Based Micro-Vibration Measurement for Lightweight Structure Using an Improved Phase-Based Motion Extraction") combines vibration detection and vibration direction based on video phase; the PVMDCSM method (reference "Modalanalysis and tension estimation of stay cables using noncontact vision-based motion magnification method") combines motion magnification and centroid detection; the PVMLT method (reference "Cable vibration measurement based on broad-band phase-based motion magnification and linetracking algorithm") combines motion magnification and line segment tracking; and EnhancedPVE is the method of this invention. The frequency measurement results of 10 cables under natural excitation are as follows: Figure 5 As shown.

[0135] This embodiment employs the phase-based bridge cable micro-vibration frequency detection method and system of the present invention. It eliminates the need for sensors on the object being tested, requiring only a general video acquisition device. Furthermore, it effectively extracts weak vibration signals and frequencies of the cable structure under natural excitation, thereby enabling accurate calculation of the cable force. The flowchart of the frequency measurement method is as follows: Figure 1 As shown, compared with the traditional video-based method for measuring cable force, this method eliminates the need for manual marking of special points on the object being tested, pre-identification of the cable, and is unaffected by changes in lighting conditions. This improves the accuracy and convenience for technicians using non-contact video methods to measure cable force and health status.

[0136] The invention and its embodiments have been described above illustratively. This description is not restrictive, and the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. The accompanying drawings are only one embodiment of the invention, and the actual structure is not limited thereto. No reference numerals in the claims should limit the scope of the claims. Therefore, if a person skilled in the art is inspired by this description and designs a similar structure and embodiment without departing from the spirit of the invention, such design should fall within the scope of protection of this patent. Furthermore, the word "comprising" does not exclude other elements or steps, and the word "a" preceding an element does not exclude the inclusion of "a plurality" of that element. Multiple elements stated in the product claims may also be implemented by a single element through software or hardware. The terms "first," "second," etc., are used to indicate names and do not indicate any specific order.

Claims

1. A phase-based method for detecting minute vibration frequencies of bridge cables, comprising the following steps: Video capture: Capture vibration videos of the object being tested; Video image region segmentation and preprocessing: The frame images of the acquired video are segmented and filtered; Image phase information extraction: Spatial pyramid decomposition is performed on the segmented and processed image to obtain the image's phase information; Image cable phase enhancement: Based on the spatial differences between cable phase and noise phase, a phase enhancement method combining bilateral filtering and phase amplification is designed to enhance the cable phase and reduce the noise phase. Image temporal signal extraction: Based on the acquired enhanced phase information, the temporal information of each sub-region is calculated using a vibration analysis-based method to obtain the temporal vibration signal of the sub-region; Sub-region signal weighted summation: Based on the contribution of each sub-region signal to the cable signal, a maximum contribution combination algorithm is designed to assign different weight values ​​to each sub-region signal according to its contribution. Finally, all sub-region signals are weighted and summed to synthesize the time-domain vibration signal of the cable. Signal frequency calculation: The frequency value of the object being detected is calculated based on the time-domain vibration signal of the synthesized object. The image cable phase enhancement includes: The mixed phase signals are subjected to bilateral filtering, i.e.: In the formula, The filtered phase, This is the sum of weights for each phase value in the filter window, used for weight normalization; and These are the phase similarity weight and the spatial distance weight, respectively; For pixels in The phase value at any given time, including the phase of the cable. Phase of noise ,Right now: It contains motion information, and motion change information is obtained by calculating the phase difference between two frames, where the phase difference between the two frames is: for Phase value when =0; In phase difference signal A magnification factor is assigned above. Enhance the phase difference signal to obtain the enhanced cable phase. : ； The weighted summation of the sub-region signals includes: After obtaining the enhanced image phase information in the phase space, the temporal vibration signal of the image sub-region is extracted using a vibration analysis method based on video phase: Indicates the first in the image ( Time-domain vibration signal of the region; The time-domain signal of each sub-region is represented as a combination of cable signal and noise signal, i.e.: For cable signal In the region Contribution weight in For the region Noise signals in; To receive the same expected signal from the cable The signals come from different channels with different intensities and different noise levels; the method to combine all these different channel signals by contribution weighted summation is as follows: In the formula, The final estimated target signal after weighted summation. For the region The signal weight value; Weight Defined as a signal within the main vibration frequency range PSD With frequency band [ c Hz, d The ratio of the internal noise PSD to the noise level within Hz, i.e.: In the formula, The dominant vibration frequency of the cable, [ , ] is a The small frequency band centered on The value range is (0,1).

2. The method for detecting minute vibration frequencies of bridge cables based on phase according to claim 1, characterized in that, The specific steps for video image region segmentation and preprocessing are as follows: The length and width of each frame of video image are respectively... n and m Divide into equal parts, that is, divide into separate parts. n Shares and m One, of which m、n All are natural numbers greater than 1; definition For the position located at the i Line number j The image region of the column, where ; Two-dimensional Gaussian filtering is applied to each image region to remove low-frequency noise caused by camera electronic imaging. The two-dimensional Gaussian function... Represented as: It is the standard deviation of the function. These are image coordinates. e It is a natural constant.

3. The method for detecting minute vibration frequencies of bridge cables based on phase according to claim 1, characterized in that, The specific steps for extracting image phase information are as follows: Perform spatial pyramid decomposition on each image region to obtain the image's phase information; this involves convolving the image with two-dimensional Gabor functions of different scales and orientations. in: For two-dimensional Gabor functions, It's the wavelength. It is the phase shift of a complex exponential function. It is a scaling factor that controls the shape of the Gabor function. It is the direction angle of the Gabor function. and for Rotated coordinates and They are time-related variables and are respectively x , y a subset of Image pixels In time The intensity value, This is the intensity value result after pyramid decomposition, representing the pixel's intensity. The phase value at any given moment.

4. The method for detecting minute vibration frequencies of bridge cables based on phase according to claim 1, characterized in that, Remove all amplitudes greater than the threshold The region, the time-domain vibration signal synthesized from the cable signals of the last set of video image sequences. The formula becomes: In the formula, and They are respectively The maximum and minimum values ​​in the signal.

5. A system for detecting the frequency of minute vibrations in bridge cables based on the phase-based method according to any one of claims 1-4, characterized in that, include: The video acquisition module is used to capture vibration videos of the object being tested. The video image region segmentation and preprocessing module is used to segment and filter the frame images of the acquired video. The image phase information extraction module is used to perform spatial pyramid decomposition on the segmented and processed image and obtain the image's phase information; The image phase information enhancement module is used to enhance the cable phase signal while reducing noise signals; The image time-domain signal extraction module is used to calculate the time-domain vibration signal based on the acquired image enhancement phase information using a vibration analysis-based method. The regional weight allocation and weighted summation module is used to determine the contribution of each region to the cable signal and to calculate the final cable signal by weighting the signals of all regions with different weights. The signal frequency calculation module is used to perform Fourier transform on the time-domain vibration signal of the synthesized detection object to calculate the frequency value of the detection object.

Images