A cable force real-time measurement method based on EDLines straight line detection

Abstract

The application discloses a cable force real-time measurement method based on EDLines straight line detection, and a camera synchronously collects calculation graphs of multiple cables in the same field of view; first, straight line feature detection is performed on the calculation region ROI of each cable; then, the distance from the straight line feature to the original point of the ROI is calculated; the pixel displacement of the straight line feature in the normal direction thereof can be obtained by repeatedly performing the operation on the continuously collected images; in addition to extracting the cable displacement, the relative displacement between the cables also needs to be calculated; the displacement is subjected to FFT transformation; and finally, the frequency method is adopted to calculate the cable force. The application overcomes the limitations of low efficiency and complicated arrangement of the traditional accelerometer measurement method, and can realize full-field multi-point synchronous measurement; and the application overcomes the influence of complex background and camera disturbance, and improves the accuracy of cable force monitoring.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of structure monitoring, and particularly relates to a camera-based real-time cable force measurement method. BACKGROUND

[0002] In bridge engineering, the measurement of cable force has become an important part of engineering acceptance. As the main force transmission component of the bridge, the stress state of the cable directly reflects the health state of the whole bridge, and is the core index of bridge monitoring. Various methods have been developed by engineering technicians for bridge cable force testing, such as the magnetic flux method, the pressure method, the fundamental frequency method, etc. Among them, the fundamental frequency method is widely used in bridge cable force detection and monitoring due to its low cost and convenience. Most conventional bridge cable force testing methods belong to point detection, such as detection by using acceleration sensors, pressure gauges, strain testing elements, etc., which can only obtain data of limited points on the structure. It is obviously not comprehensive to use less data as the basis for the detection, evaluation, operation and maintenance of the bridge health state. Most of the conventional detection methods need to install sensors, which belong to contact measurement, and sometimes need to interrupt traffic. For dynamic load testing, the load effect also needs to be considered sometimes. For large-scale structures with complex vibration states, multiple sensors need to be installed at different parts of the structure, which is difficult to operate, and it is difficult to achieve synchronous data acquisition. The test period is long, and a large amount of manpower, material resources and financial resources are needed.

[0003] In recent years, the non-contact cable force measurement method with the aid of camera images has been widely applied due to its advantages of full-field multi-target synchronous measurement. However, there are still some problems to be further solved.

[0004] Firstly, the selection of target tracking and sub-pixel displacement extraction algorithm, the ideal algorithm should be robust and fast. The most prominent feature of the cable is that the feature texture is almost nothing. In order to increase the contrast of the target and improve the positioning accuracy, some circular targets are set on the cable, and the centroid detection is carried out. Obviously, it is not convenient for engineering application. The optical flow method is sensitive to the change of light for long-term monitoring. The correlation method is unstable for tracking the cable with smooth and monotonous surface. When the feature point tracking algorithm is applied to the cable surface with monotonous features, there are few stable feature points, which leads to abnormal data in the extracted vibration response. The most prominent and stable feature information of the cable is its edge information, which is more stable and non-contact target. The edge detection method is simple and efficient, which is helpful for the realization of fast tracking of multiple targets. However, the method has a significant disadvantage, that is, in the displacement extraction, there is no real similarity matching, but it is assumed that the pixel point in a certain position is the same point in structure. Moreover, the matching method based on pixel gray is sensitive to the gray change of image, which is difficult to adapt to the problem of shielding. However, the straight line feature will not be affected in the case of shielding. It is helpful for the synchronous measurement of cable group, so the EDLines algorithm is used for edge feature detection of cable in this patent.

[0005] Finally, the key is the accuracy of cable frequency identification. For cable monitoring, the camera is usually set on the stable area of the bridge shore, ignoring the camera vibration. However, this method cannot completely meet the requirements of long-span bridges. If the camera is set on the bridge deck, the camera vibration will interfere with the subsequent cable frequency identification. Moreover, even if the camera is set on the stable ground, the system will also be affected by wind vibration. It is a simple and easy choice to use a stable base point to eliminate camera disturbance, but the effect of this method is affected by the distance difference between the base point and the target point. For this phenomenon, the author has discussed it in previous studies. At the same time, it is not always easy to find an ideal stable reference point in the background. Therefore, some scholars consider installing a sensor on the measuring device that can track the camera vibration. However, this method has certain requirements for the signal acquisition frequency of the two sensors. It has some limitations when applied to this study. Because the acquisition frequency of the camera is usually set within 10 frames per second, which basically meets the requirements of cable frequency identification of bridges. The acquisition frequency of the accelerometer is usually higher. Most importantly, the interference frequency of the camera system is usually low, and the accelerometer is not sensitive to low-frequency signals. This patent takes advantage of the synchronous monitoring of multiple targets in images, and further extracts the frequency of the relative image displacement between two cables, which is simple and effective to identify the interference frequency. It not only does not increase the additional calculation amount, but also does not need a higher acquisition frequency, which is more convenient for real-time implementation. SUMMARY

[0006] Purpose of the invention: The technical problem to be solved by the present invention is to provide a method for real-time measurement of cable force under complex imaging backgrounds and camera disturbances. The method focuses on extracting cable features and displacements under complex backgrounds during image processing, using the relative displacement method to eliminate interference frequencies, and then calculating cable force based on the frequency method.

[0007] Technical solution: A real-time cable force measurement method based on EDLines straight line detection, comprising the following steps:

[0008] (1) The camera simultaneously acquires a calculation diagram of multiple cables in the same field of view;

[0009] (2) Select the ROI of each cable in the initial calculation graph and use the optimized EDLines line feature detection operator to perform line feature detection;

[0010] (3) Select the cables and calculate the displacement of each cable and the relative displacement between different cables;

[0011] (4) Perform FFT transformation on the displacement to extract the cable frequency and remove interference frequencies. The peak frequencies in the FFT transformation results of each cable displacement include the fundamental frequency of the cable and the noise introduced by the camera disturbance. The frequency peaks that disappear in the FFT results of the relative displacement are noise, while those that remain are the fundamental frequency of the cable. In steps (3) and (4), not only the displacement and spectrum analysis of the cable itself are considered, but also the relative displacement between the two cables and the corresponding spectrum analysis are calculated. The purpose is to identify the noise caused by the camera disturbance in the cable vibration analysis, so as to accurately obtain the fundamental frequency of the cable.

[0012] (5) Calculate the cable force.

[0013] Furthermore, the optimized EDLines line feature detection operator described in step (2) adds a slope filtering strategy to the original EDLines algorithm, aiming to eliminate the influence of complex backgrounds. The specific steps are as follows:

[0014] Step 1: In the acquired computational graph, select the region of interest (ROI) and the coordinates (x1, y1) and (x2, y2) of two pixels on one side edge of the cable. The greater the distance between these two points, the better. Calculate the tilt angle of the line determined by these two pixels.

[0015]

[0016] As a subsequent screening threshold;

[0017] Step 2: Apply Gaussian filtering to the grayscale image corresponding to the ROI to suppress noise, and then calculate the gradient magnitude and gradient direction pixel by pixel, as shown in the following formula:

[0018]

[0019] Each feature point can get three information ((x, y), m, θ), where ((position), scale, direction); based on the angle threshold θ obtained in Step 1, the direction consistency screening is performed Th ;

[0020] Step 3: Extract the pixel with the maximum gradient amplitude, denoted as an anchor point, and sequentially link the extracted anchor points to form multiple edge lines; adopt the least square method to divide the multiple edge lines into multiple straight line segments, and calculate the shortest length n of the line segment by using the following formula:

[0021]

[0022] In the formula, N represents the resolution of the image, and p is a parameter value. p = 0.125.

[0023] Further, the direction consistency screening is to make the angle between the straight line direction determined in Step 1 and the pixel gradient direction within a certain angle range, and the determination of the range refers to the standard when calculating the gradient angle, that is, 22.5°, so the pixel points whose gradient directions do not satisfy [θ Th -0.196, θ Th +0.196] are not included in the calculation object for subsequent straight line detection, which is one of the optimization strategies proposed in the present research. It helps to reduce the interference of other non-target objects and greatly reduces the operation amount for subsequent calculation;

[0024] Further, in Step (3), the displacement of each cable is the displacement of the cable edge feature in the normal direction thereof, and the calculation formula is as follows:

[0025]

[0026] Wherein, d L and d R are the distances from the left and right edges of the cable to the origin of the ROI image coordinate system, and the final pixel displacement of the cable between adjacent time points i-1 and i is d i -d i-1 .

[0027] Further, in Step (3), the selected cables are as close as possible, and the selected calculation region ROI is also as close as possible. The purpose is to make the relative displacement better weaken the cable displacement measurement error caused by camera disturbance.

[0028] Further, Step (5) uses the following frequency method formula to calculate the cable force:

[0029]

[0030] wherein, f n represents the nth order frequency, W represents the mass per meter of the cable, that is, the linear density, L represents the cable length, E represents the elastic modulus of the cable material, and I represents the moment of inertia.

[0031] Advantages:

[0032] First, the present application does not require the installation of sensors on the measured object, has no load effect, belongs to non-contact measurement, and can avoid the problem of not being able to obtain measurement point data when the measured object cannot be contacted due to special circumstances.

[0033] Second, multi-point and line testing can be realized without on-site sensor wiring work, the operation is simple and the cost is low, synchronous measurement of multiple cables can be realized, the measurement efficiency is high, time and labor are saved.

[0034] Third, the present application does not require the camera to be erected at a stable observation site, and can also be disturbed by wind load and other environments.

[0035] Fourth, the present application does not require the imaging background of the cable to be monotonous, and allows the existence of complex buildings or other trees. BRIEF DESCRIPTION OF DRAWINGS

[0036] Figure 1 is a schematic diagram of the test system of the embodiment of the present application;

[0037] Figure 2 is a schematic diagram of the cable displacement extraction principle of the embodiment of the present application;

[0038] Figure 3 is a schematic diagram of the principle of identifying disturbance frequency and real cable frequency by the relative displacement method of the embodiment of the present application;

[0039] Figure 4 is a schematic diagram of the camera erected on the bridge deck to collect images of the embodiment of the present application;

[0040] Figure 5 is a calculation diagram of the embodiment of the present application;

[0041] Figure 6 is a pixel displacement extraction result diagram of the embodiment of the present application;

[0042] Figure 7 is a frequency analysis result of the embodiment of the present application;

[0043] Figure 8 is a cable force measurement effect comparison diagram of the embodiment of the present application. DETAILED DESCRIPTION

[0044] The technical solutions of the present application will be further described below with reference to the accompanying drawings.

[0045] (1) As Figure 1 As shown, after the camera is set up, the image is adjusted so that multiple cables are in the same field of view. Then, the image is continuously acquired and calculated. The acquisition frequency is at least twice the first-order fundamental frequency of the cables. The acquisition is carried out continuously for about 2 minutes.

[0046] (2) Figure 2 As shown, the computational region of interest (ROI) is determined in the computational graph, and linear features are extracted based on optimized EDLines. L and d R This is the distance from the left and right edges of the cable to the origin of the ROI image coordinate system, and the average of the two distances is taken as the final distance value. The final pixel displacement of the cable between adjacent times i-1 and i is d. i -d i-1 ;

[0047] (3) Calculate the relative displacement between the cables. Taking two cables, A and B, in the same field of view as an example, after calculating the pixel displacement of cable A and cable B, their relative displacement is obtained by subtracting them;

[0048] (4) Perform a Fourier transform on the displacement to extract the cable frequencies and remove interfering frequencies. For example... Figure 3 As shown, the peak values ​​that appear in the spectral analysis results of cables A or B but disappear in the relative displacement spectral results are the frequencies of the disturbance; the peak values ​​that always exist are the frequency values ​​of the corresponding cables.

[0049] (5) Cable force calculation. The cable force is calculated using the frequency method. The frequency selected here is unique to the image signal frequency.

[0050]

[0051] Among them, f n Let W represent the nth frequency, W represent the mass per meter of the cable (i.e., linear density), L represent the cable length, E represent the elastic modulus of the cable material, and I represent the moment of inertia.

[0052] The following example, using a real-world bridge application, illustrates the real-time steps of the invented cable force measurement method. Figure 4 As shown, the camera was mounted on the bank below the bridge and was subjected to significant wind-induced vibration loads. The collected calculation diagram is shown below. Figure 5 As shown, multiple cables numbered Z16 to Z30 were simultaneously acquired. The acquisition frequency was 10 frames / s. Taking cables Z16 and Z18 as examples, the extracted pixel displacement results are as follows. Figure 6 As shown, the results of the spectral analysis are as follows: Figure 7The cable force is calculated by frequency method and compared with the design value and the value measured by acceleration in 2018, and the results are shown in Figure 8 The results indirectly prove the reliability of the image measurement method.

[0053] The application of the optimized EDLines between the detectors to the situation is the first time. The relative displacement between the cables is used for spectral analysis and the interference frequency caused by camera disturbance is also removed for the first time. The method has the advantages of fast, convenient, reliable, low cost and real-time measurement, which is convenient for wide engineering application.

[0054] The above is a further detailed description of the application in combination with specific preferred embodiments, and the specific implementation of the application cannot be limited to these descriptions. For those skilled in the art to which the application belongs, without departing from the concept of the application, a number of equivalent substitutions or obvious modifications can be made, and the performance or use is the same, which should be regarded as belonging to the protection scope of the application.

Claims

1. A method for real-time measurement of cable force based on EDLines straight line detection, characterized in that, Includes the following steps: (1) The camera simultaneously acquires a calculation diagram of multiple cables in the same field of view; (2) Select the ROI of each cable in the initial calculation graph and use the optimized EDLines line feature detection operator to perform line feature detection; (3) Select the cables and calculate the displacement of each cable and the relative displacement between different cables; (4) Perform FFT transformation on the displacement to extract the cable frequency and remove interference frequencies; the peak frequencies in the FFT transformation results of each cable displacement include the fundamental frequency of the cable and the noise introduced by the camera disturbance. The frequency peaks that disappear in the FFT results of the relative displacement belong to the noise, while the frequencies that remain belong to the fundamental frequency of the cable. (5) Calculate the cable tension; The optimized EDLines line feature detection operator described in step (2) adds a slope filtering strategy to the original EDLines algorithm. The specific steps are as follows: Step 1: In the acquired computational graph, select the region of interest (ROI) and the coordinates (x1, y1) and (x2, y2) of two pixels on one side edge of the cable. Calculate the tilt angle of the line defined by these two pixels. As a subsequent screening threshold; The angle selection threshold θ obtained in Step 1 Th Perform direction consistency screening; The direction consistency filtering ensures that the angle between the straight line direction determined in Step 1 and the pixel gradient direction is within a certain range, and filters out gradient directions that do not satisfy [θ]. Th -0.196, θ Th Pixels with a value of +0.196 are not included in the subsequent line detection calculation.

2. The method for real-time cable force measurement based on EDLines line detection according to claim 1, characterized in that, The slope filtering strategy further includes the following steps: Step 2: Apply Gaussian filtering to the grayscale image corresponding to the ROI to suppress noise, and then calculate the gradient magnitude and gradient direction pixel by pixel, as shown in the following formula: Each feature point can obtain three pieces of information: *((x,y), m, θ), where *((position), scale, orientation); Step 3: Extract the pixel with the largest gradient magnitude and denote it as the anchor point. Connect the extracted anchor points sequentially to form multiple edge lines. Use the least squares method to divide the multiple edge lines into multiple straight line segments. Calculate the shortest line segment length n using the following formula: In the formula: N represents the resolution of the image, and P is the parameter value.

3. The method for real-time cable force measurement based on EDLines line detection according to claim 1, characterized in that, In step (3), the extracted displacements of each cable are the displacements of the cable edge features in their normal direction, and the calculation formula is as follows: Where, d L and d R The distances from the left and right edges of the cable to the origin of the ROI image coordinate system are given by d. The final pixel displacement of the cable between adjacent times i-1 and i is d. i -d i-1 .

4. The method for real-time cable force measurement based on EDLines line detection according to claim 1, characterized in that, In step (3), the selected cables are as close as possible, and the selected calculation area ROI is also as close as possible.

5. The method for real-time cable force measurement based on EDLines line detection according to claim 2, characterized in that, Step (5) Calculate the cable force using the following frequency method formula: Among them, f n Let W represent the nth frequency, W represent the mass per meter of the cable (i.e., linear density), L represent the cable length, E represent the elastic modulus of the cable material, and I represent the moment of inertia.

Images