Method for calculating wave load borne by pile based on machine vision

Abstract

The invention discloses a method for calculating the wave load borne by a pile based on machine vision, and the method comprises the steps: collecting and obtaining a water surface image sequence containing the pile, carrying out the image processing of the water surface image sequence, obtaining a flow field image, carrying out the post-processing of the flow field image, and obtaining an optimized flow field image, the flow field is comprehensively monitored through machine vision, and accurate recognition of the flow field where the pile is located is improved; calibration of a preset image scaling is carried out on the optimized flow field image, pixel coordinates in the optimized flow field image are converted into actual coordinates, pile periphery data under a corresponding flow field are obtained based on a particle image velocimetry (PIV) algorithm, and a velocity vector field of fluid particles can be obtained through the particle image velocimetry algorithm (PIV); the flow velocity distribution, vortex and other fluid mechanics parameters of the whole water area are further deduced, pile periphery data under the corresponding flow field can be obtained, then the wave load borne by the pile is obtained based on the Morison equation, and the defect that according to an existing wave load method, due to the fact that the working environment of underwater ocean engineering structures is complex and severe or the safety requirement is high, the wave load is poor is overcome. The problems of low wave load accuracy and high investment cost are solved, and the accuracy and efficiency of the wave load borne by the pile are greatly improved.

Description

Technical Field

[0001] The invention relates to the technical field of marine engineering, and in particular to a method for calculating wave loads on piles based on machine vision. Background Art

[0002] With the continuous development of marine engineering and port engineering, pile foundations are widely used in structures such as cross-sea bridges, ports, docks and offshore platforms due to their high bearing capacity and convenient construction. These underwater structures are usually affected by wave loads.

[0003] The relevant technologies for studying the stress conditions of underwater submerged marine engineering structures under the action of waves currently usually use field observation methods and physical model experimental methods. For the field observation method, due to the complex and harsh operating environment of underwater marine engineering structures, a large amount of manpower, material and financial resources are required. In addition, the performance of the on-site measuring instruments and the safety of operation are required to be high during the observation process, which makes this method have certain limitations. The physical model experimental method usually scales the underwater structure in proportion and conducts simulation research in a wave tank or experimental pool, but the cost of this method is relatively high.

[0004] Therefore, it is urgent to propose a calculation method for piles subjected to wave loads based on machine vision. Summary of the invention

[0005] The present invention provides a method for calculating wave loads on piles based on machine vision to overcome the above technical problems.

[0006] In order to achieve the above object, the technical solution of the present invention is:

[0007] A method for calculating wave loads on piles based on machine vision specifically comprises the following steps:

[0008] S1: collecting and acquiring a water surface image sequence including piles, and performing image processing on the water surface image sequence to acquire a flow field image;

[0009] S2: Post-processing the flow field image to obtain an optimized flow field image;

[0010] The post-processing includes image-based verification processing and speed-based verification processing;

[0011] The image-based verification process includes filtering bright areas of the image based on a high-pass filter, implementing low-contrast areas based on an image threshold segmentation method, and interpolation processing of missing data.

[0012] The velocity-based verification process includes setting a flow field velocity boundary value, and removing outliers in the flow field velocity based on a standard deviation filter and a local median filter;

[0013] S3: Calibrate the optimized flow field image with a preset image scaling ratio to convert the pixel coordinates in the optimized flow field image into actual coordinates, and obtain the pile circumference data under the corresponding flow field based on the particle image velocimetry algorithm PIV;

[0014] The pile peripheral data includes the position coordinates of any point around the pile column at different times and the horizontal velocity component and vertical velocity component of any point;

[0015] S4: Based on the Morison equation, the wave load on the pile is obtained according to the horizontal velocity component and the vertical velocity component.

[0016] Furthermore, the S1 specifically includes the following steps:

[0017] S11: setting camera calibration parameters, camera exposure and frequency to obtain a number of water surface sample images containing piles and sequentially encode them to obtain a water surface image sequence;

[0018] S12: The water surface image sequence is sequentially encoded and imported into the PIVlab toolbox for region selection to obtain a rectangular region of interest ROI;

[0019] S13: Divide the rectangular area ROI into a number of window image blocks of the same size, and perform adaptive clipping histogram equalization processing on each window image block to obtain an equalized image with enhanced local detail contrast, wherein the local details at least include texture and ripple features of the water surface;

[0020] S14: filtering the equalized image based on a high-pass filter to obtain a local detail enhancement feature map;

[0021] S15: removing the image background brightness deviation from the local detail enhancement feature map to obtain an optimized feature map;

[0022] S16: Perform a cross-correlation operation on the optimized feature map to obtain a flow field image.

[0023] Furthermore, the local detail enhancement feature map described in S15 is used to remove the image background brightness deviation, and its expression is:

[0024]

[0025] Where: I′(x,y) represents the pixel intensity after the image background brightness deviation is removed; I(x,y) represents the original pixel intensity of the local detail enhancement feature map; Represents the average intensity of the local detail enhancement feature map;

[0026] Furthermore, the cross-correlation operation is performed on the optimized feature map in S16 to obtain the flow field image, specifically:

[0027] Assume that the position of the particle in the current window image block at time t is (x, y); the position of the particle in the query window image block at time t+Δt is (x+ξ, y+η);

[0028] Wherein, the particle represents the flow field feature point that is expected to be tracked in the optimization feature map; ξ represents the displacement of the particle in the x-direction of the preset coordinate axis; η represents the displacement of the particle in the y-direction of the preset coordinate axis;

[0029] According to the constructed cross-correlation function, the position point of the particle in the query image block at time t+Δt is obtained as (x+ξ,y+η) based on the position point in the current image block as (x,y), so as to realize the flow path tracking of the desired flow field feature point, and then obtain the flow field image;

[0030] And the constructed cross-correlation function expression is

[0031]

[0032]

[0033] Where: M0 represents the length of the current window image block; N0 represents the width of the current window image block; I1 and I2 represent the pixel grayscale values ​​of the window image block at time t and time t+Δt respectively; i represents the horizontal offset of the window image block; j represents the vertical offset of the window image block; I1(x+i,y+j) represents the pixel grayscale value of the window image block at time t; represents the average pixel grayscale value of the sub-window image block at time t; I2(x+ξ+i,y+η+j) represents the pixel grayscale value of the window image block at time t+Δt; Represents the average pixel grayscale value of the window image block at time t+Δt.

[0034] Furthermore, in S3, the pixel coordinates in the optimized flow field image are converted into actual coordinates, and the expression is:

[0035] x real =x pixel ×Scale

[0036] y real =y pixel ×Scale

[0037] Where: x pixel represents the pixel horizontal coordinate of the pile in the optimized flow field image; pixel represents the pixel ordinate of the column in the optimized flow field image; x realrepresents the actual horizontal coordinate of the pile in the optimized flow field image; real It represents the actual vertical coordinate of the column in the optimized flow field image; Scale represents the actual distance corresponding to each pixel.

[0038] Furthermore, the S4 specifically includes the following steps:

[0039] S41: Obtaining the function of the pile structural performance parameters and the wave velocity along the water depth;

[0040] The structural performance parameters include at least the pile diameter, the pile inertia coefficient and the pile drag coefficient;

[0041] The function of the wave speed varying along the water depth includes any one of a constant function, a linear function or a quadratic function;

[0042] And the constant function: V(h)=Va;

[0043] Linear function: V(h) = (h / H)*Va;

[0044] Quadratic function: V(h)=(h 2 / H 2 )*Va

[0045] Where Va represents the wave velocity on the pile at the water surface, and includes the horizontal velocity component u and the vertical velocity component v; h represents the water depth from a certain point of the pile to the water surface; H represents the total water depth;

[0046] S42: Based on the data around the pile at different times and step S41, the wave load on the pile is obtained based on the Morison equation, and its expression is:

[0047]

[0048] Where: F D Indicates the drag force of water on the pile; F H Indicates the wave load on the pile; F I represents the inertial force of water on the pile column; ρ represents the density of water; V(h) represents the velocity of the water particle at the axis of the column; D represents the diameter of the pile column; C D Indicates the drag coefficient; C M represents the inertia force coefficient; h0 represents the depth from the water surface to a certain point on the pile; d(h) represents the integral of the water depth h.

[0049] Beneficial effect: The present invention provides a method for calculating the wave load on piles based on machine vision, which collects and obtains a water surface image sequence containing piles, performs image processing on the water surface image sequence to obtain a flow field image, and performs post-processing on the flow field image to obtain an optimized flow field image, and realizes comprehensive monitoring of the flow field through machine vision, thereby improving the accurate identification of the flow field where the piles are located; the optimized flow field image is calibrated with a preset image scaling ratio to convert the pixel coordinates in the optimized flow field image into actual coordinates, and obtains the pile circumference data under the corresponding flow field based on the particle image velocimetry algorithm PIV, and the velocity vector field of the fluid particles can be obtained through the particle image velocimetry algorithm PIV, and the flow velocity distribution, vorticity and other fluid mechanics parameters of the entire water area can be further derived, and the pile circumference data under the corresponding flow field can be obtained, and then the wave load on the piles can be obtained based on the Morison equation, which overcomes the problems of low wave load accuracy and high investment cost caused by the complex and harsh operating environment or high safety requirements of underwater marine engineering structures in the existing wave load method, and greatly improves the accuracy and efficiency of the wave load on the piles. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0051] Figure 1 It is a flow chart of the method for calculating the wave load on piles based on machine vision of the present invention;

[0052] Figure 2 This is a schematic diagram of the interface for sequentially encoding a water surface image sequence and importing it into the PIVlab toolbox in this embodiment;

[0053] Figure 3 This is a schematic diagram of the selection interface of the rectangular region of interest ROI in this embodiment;

[0054] Figure 4 It is a characteristic graph after the adaptive clipping histogram equalization processing and filtering processing in this embodiment;

[0055] Figure 5 This is a schematic diagram of the interface for setting the PIV algorithm parameters in this embodiment;

[0056] Figure 6 To perform image processing in this embodiment to obtain a flow field image;

[0057] Figure 7Schematic diagram of calibrating the optimized flow field image in this embodiment;

[0058] Figure 8 In this embodiment, the flow field image is post-processed to obtain an optimized flow field image;

[0059] Fig. 9 Schematic diagram of pile peripheral data in this embodiment;

[0060] Fig.10 Schematic diagram of the PIV algorithm tracking particles in the flow field in this embodiment;

[0061] Fig.11 Schematic diagram of the calculation interface of the pile column subjected to wave load based on machine vision in this embodiment;

[0062] Fig.12 is a schematic diagram of a flow field image at the first moment in this embodiment;

[0063] Fig.13 is a schematic diagram of a flow field image at the second moment in this embodiment;

[0064] Fig.14 is a schematic diagram of a flow field image at the third moment in this embodiment;

[0065] Fig.15 is a schematic diagram of a flow field image at the fourth moment in this embodiment;

[0066] Fig.16 is a histogram of horizontal velocity components of the flow field data in this embodiment;

[0067] Fig.17 is a histogram of the vertical velocity component of the flow field data in this embodiment;

[0068] Fig.18 is a histogram of the combined velocity components of the flow field data in this embodiment;

[0069] Fig.19 is a histogram of horizontal velocity components of the pile column data in this embodiment;

[0070] Fig. 20 is a histogram of the vertical velocity component of the pile column data in this embodiment;

[0071] Fig.21 is a histogram of the combined velocity components of the pile column data in this embodiment;

[0072] Fig. 22 It is a velocity variation diagram of the horizontal velocity component of the pile column data in this embodiment;

[0073] Fig.23 It is a velocity variation diagram of the vertical velocity component of the pile column data in this embodiment;

[0074] Fig.24 The speed variation diagram of the combined speed of the pile column data in this embodiment;

[0075] Fig.25 This is a diagram showing the variation of wave loads on the pile column in this embodiment. DETAILED DESCRIPTION

[0076] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0077] This embodiment provides a method for calculating the wave load on a pile based on machine vision, such as Figure 1 As shown, the specific steps include:

[0078] S1: collecting and acquiring a water surface image sequence including piles, and performing image processing on the water surface image sequence to acquire a flow field image;

[0079] The specific steps include:

[0080] S11: setting camera calibration parameters, camera exposure and frequency to obtain a number of water surface sample images containing piles;

[0081] S12: Figure 2 As shown, the water surface image sequence is imported into the PIVlab toolbox in order, and the rectangular region of interest ROI is manually selected;

[0082] Specifically, Figure 3 As shown, PIVlab uses the built-in imrect function of MATLAB or similar tools to realize the selection of ROI, that is, it can select the rectangular region of interest (ROI) by dragging and clicking the mouse. The imrect function allows the user to drag a rectangular box on the image to frame the region of interest. Once the rectangular region of interest ROI is selected, PIVlab will record the coordinates of the four vertices of the rectangular region for subsequent image processing and particle image velocimetry algorithm PIV. The main purpose of selecting the rectangular region of interest ROI is to set the part of interest in the entire water surface image to reduce the amount of calculation and focus on the key area of ​​the flow field;

[0083] S13: Divide the rectangular area ROI into a number of window image blocks of the same size, and perform adaptive limit histogram equalization (CLAHE) processing on each window image block to obtain an equalized image with enhanced local detail contrast, wherein the local details at least include texture and ripple features of the water surface;

[0084] Specifically, CLAHE, or adaptive clipping histogram equalization, divides the image into several small windows (window image blocks), and independently performs histogram equalization in each window to better enhance local detail features, such as the texture and ripple features of the water surface, and interpolates the equalization results between each window to ensure the overall smoothness of the image.

[0085] S14: filtering the equalized image based on a high-pass filter to obtain a local detail enhancement feature map;

[0086] Specifically, Figure 4 As shown in FIG. 1 , the high-pass filter is used to remove low-frequency components (such as uneven background illumination, etc.) in the water surface image and enhance high-frequency details (such as particles) in the image to obtain a particle-enhanced image; the "particles" here refer to the detail features that appear due to filtering in image processing, and the role of the high-pass filter is to enhance these details so that the fine features that may be covered by the background (such as fine lines on the water surface) become more obvious and clear;

[0087] S15: removing the image background brightness deviation from the local detail enhancement feature map to obtain an optimized feature map;

[0088] Specifically, by subtracting the mean intensity of the image to remove the overall deviation in background brightness, the contrast of the image is improved, making the particles more visible and facilitating subsequent PIV analysis.

[0089] The local detail enhancement feature map is used to remove the image background brightness deviation, and its expression is:

[0090]

[0091] Where: I′(x,y) represents the pixel intensity after the image background brightness deviation is removed; I(x,y) represents the original pixel intensity of the local detail enhancement feature map; Represents the average intensity of the local detail enhancement feature map;

[0092] S16: Perform cross-correlation operation on the optimized feature map to obtain a flow field image, specifically:

[0093] like Figures 12 to 15As shown, the position point of the particle in the current window image block at time t is set to (x, y); the position point of the particle in the query window image block at time t+Δt is set to (x+ξ, y+η);

[0094] Wherein, the particle represents the flow field feature point that is expected to be tracked in the optimization feature map; ξ represents the displacement of the particle in the x-direction of the preset coordinate axis; η represents the displacement of the particle in the y-direction of the preset coordinate axis;

[0095] According to the constructed cross-correlation function, the position point of the particle in the query image block at time t+Δt is obtained as (x+ξ,y+η) based on the position point in the current image block as (x,y), so as to realize the flow path tracking of the desired flow field feature point, and then obtain the flow field image;

[0096] And the constructed cross-correlation function expression is

[0097]

[0098] Where: M0 represents the length of the current window image block; N0 represents the width of the current window image block; I1 and I2 represent the pixel grayscale values ​​of the window image block at time t and time t+Δt respectively; i represents the horizontal offset of the window image block; j represents the vertical offset of the window image block; I1(x+i,y+j) represents the pixel grayscale value of the window image block at time t; represents the average pixel grayscale value of the sub-window image block at time t; I2(x+ξ+i,y+η+j) represents the pixel grayscale value of the window image block at time t+Δt; Represents the average pixel grayscale value of the window image block at time t+Δt;

[0099] S2: Post-processing the flow field image to obtain an optimized flow field image;

[0100] like Figure 6 and Figure 8 As shown, the post-processing includes image-based verification processing and speed-based verification processing;

[0101] The image-based verification process includes filtering bright areas of the image based on a high-pass filter, implementing low-contrast areas based on an image threshold segmentation method, and interpolation processing of missing data.

[0102] Specifically, filtering the bright areas of the image based on a high-pass filter is usually achieved through image threshold segmentation (this is a commonly used technique that sets a grayscale threshold to segment or mask areas in the image with brightness higher than the threshold) or local filtering technology (using the average value or standard deviation of the local area to identify and process overly bright areas, which can be achieved through filters such as high-pass filters to suppress the influence of these areas). The purpose is to remove areas that may cause interference due to overexposure or excessive brightness;

[0103] Realize low contrast areas based on image threshold segmentation method: "Filtering low contrast areas" is usually achieved through image threshold segmentation technology (by setting a grayscale threshold, segmenting or masking the areas in the image with brightness lower than the threshold), the purpose is to reduce the difficulty of particle identification caused by low contrast in the image, thereby improving the accuracy of velocity field calculation and the overall quality of data. By filtering these areas, it can be ensured that only clear and information-rich image areas are used for subsequent PIV velocity analysis;

[0104] Interpolation of missing data: usually achieved through existing interpolation technology, by using the spatial distribution of known data points to predict and fill in the missing data, the purpose is to supplement the data missing due to insufficient particle density or image quality problems, thereby ensuring the continuity and integrity of the flow field calculation. This operation helps to generate more complete and accurate velocity field data and improve the reliability of the analysis;

[0105] The velocity-based verification process includes setting a flow field velocity boundary value, and removing outliers in the flow field velocity based on a standard deviation filter and a local median filter;

[0106] Specifically, the removal of outliers in the flow field velocity based on the standard deviation filter and the local median filter is to ensure the accuracy and reliability of the flow field data. By screening and cleaning the velocity data, outliers and noise can be removed, data errors can be reduced, and the credibility of the results can be improved. This process helps to identify and correct untrue or inconsistent data points, ensuring that the results of flow field analysis and calculations are more accurate and consistent;

[0107] S3: Figure 7 As shown in FIG. 1 , the optimized flow field image is calibrated with a preset image scaling ratio to convert the pixel coordinates in the optimized flow field image into actual coordinates, and the pile circumference data under the corresponding flow field is obtained based on the particle image velocimetry algorithm PIV, as shown in FIG. Figures 16 to 24 As shown,

[0108] The pixel coordinates include pile pixel coordinates and flow field pixel coordinates;

[0109] like Figure 5 and Fig.10As shown, the particle image velocimetry algorithm PIV is a transient flow field measurement technology. Its basic principle is to spread some appropriate tracer particles in the flow field for tracking, use pulsed laser light to illuminate the flow field area where the pile is located, and use the existing imaging recording system to capture the particle image exposed twice or multiple times to form a moving particle image. The particle image is randomly divided into several small sub-areas, and the average displacement of the particle image in each small area is obtained, thereby determining the two-dimensional fluid velocity of the entire area on the flow field area, so as to determine the velocity field of the flow field area, and then obtain the pile circumference data under the corresponding flow field. Among them, the specific implementation process and working principle of the particle image velocimetry algorithm PIV are existing well-known technical means, which will not be described in detail here;

[0110] like Fig. 9 As shown, the pile peripheral data includes the position coordinates of any point around the pile column at different times and the horizontal velocity component and vertical velocity component of any point;

[0111] Specifically, the calibration method is carried out through the image scaling ratio, that is, the calibration tool is the pixel length of the known object and the scale ratio of the image. The purpose of calibration is to convert the pixel coordinates into actual coordinates, so as to obtain the pile perimeter data and flow field data, so that the data in the image can correspond to the actual distance; that is, the image scaling ratio is calculated by the pixel length of the known object, and this ratio is used to convert other pixel coordinates into actual distances;

[0112] The pixel coordinates in the optimized flow field image are converted into actual coordinates, and the expression is:

[0113] x real =x pixel ×Scale

[0114] y real =y pixel ×Scale

[0115] Where: x pixel represents the pixel horizontal coordinate of the pile in the optimized flow field image; pixel represents the pixel ordinate of the column in the optimized flow field image; x real represents the actual horizontal coordinate of the pile in the optimized flow field image; real Indicates the actual vertical coordinate of the pile in the optimized flow field image; Scale indicates the actual distance corresponding to each pixel (unit: mm / pixel, m / pixel, etc.)

[0116] S4: Based on the Morison equation, the wave load on the pile is obtained according to the horizontal velocity component and the vertical velocity component.

[0117] like Fig.25 As shown, the specific steps include:

[0118] S41: Fig.11 As shown, the structural performance parameters of the pile are obtained as a function of the wave velocity changing along the water depth;

[0119] The structural performance parameters include at least the pile diameter, the pile inertia coefficient and the pile drag coefficient;

[0120] The function of the wave speed varying along the water depth includes any one of a constant function, a linear function or a quadratic function;

[0121] And the constant function: V(h)=Va;

[0122] Linear function: V(h) = (h / H)*Va;

[0123] Quadratic function: V(h)=(h 2 / H 2 )*Va

[0124] Where Va represents the wave velocity on the pile at the water surface, and includes the horizontal velocity component u and the vertical velocity component v; h represents the water depth from a certain point of the pile to the water surface; H represents the total water depth;

[0125] S42: Based on the data around the pile at different times and step S41, the wave load on the pile is obtained based on the Morison equation, and its expression is:

[0126]

[0127] Where: F D Indicates the drag force of water on the pile; F H Indicates the wave load on the pile; F I represents the inertial force of water on the pile column; ρ represents the density of water; V(h) represents the velocity of the water particle at the axis of the column; D represents the diameter of the pile column; C D Indicates the drag coefficient; C M represents the inertia coefficient; h0 represents the depth from the water surface to a certain point on the pile; d(h) represents the integral of the water depth h;

[0128] That is, according to the expression of the wave load on the pile column, the "horizontal velocity component" is substituted into the water particle velocity of the Morison formula to calculate the "wave load in the x direction" at different times; the "vertical velocity component" is substituted into the water particle velocity of the Morison formula to calculate the "wave load in the y direction" at different times; and the square root values ​​of the "wave load in the x direction" and the "wave load in the y direction" at different times are calculated, which is the total "wave load" on the pile column at different times;

[0129] S43: obtaining a corresponding image of the wave load on the pile based on the existing drawing technology, that is, drawing an image of the wave load on the pile over time according to the wave load on the pile at different times.

[0130] In this embodiment, a sequence of water surface images containing piles is collected and acquired, and image processing is performed to acquire flow field images; the flow field images are post-processed to acquire optimized flow field images; the optimized flow field images are calibrated with a preset image scaling ratio to convert pixel coordinates in the optimized flow field images into actual coordinates, and the data around the piles under the corresponding flow field is acquired based on the particle image velocimetry algorithm PIV. This process is machine vision processing, wherein machine vision is a method that utilizes computer vision technology and image processing, which acquires image or video data through a camera or sensor, and analyzes and processes it to realize the technology of detecting, identifying, measuring and judging objects. The application of machine vision in the perception of water surface flow fields is to capture images or videos of the water surface through cameras or sensors, and use image processing and computer vision algorithms to analyze the texture, ripples, flow paths and other features on the water surface, so as to perceive and measure the dynamic characteristics of the water flow, such as speed, direction, turbulence, etc. By extracting and analyzing these visual features, machine vision can achieve comprehensive monitoring of the flow field. Based on the particle image velocimetry (PIV) algorithm, it can calculate the velocity vector field of fluid particles, and further derive the velocity distribution, vorticity and other fluid mechanics parameters of the entire water area. Then, the wave load on the pile can be calculated through the Morrison empirical formula, which solves the problem of low accuracy and high investment cost of wave loads caused by the complex and harsh operating environment of underwater marine engineering structures or high safety requirements of existing wave load methods, and greatly improves the accuracy and efficiency of wave loads on piles.

[0131] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for calculating wave loads on piles based on machine vision, characterized in that: The specific steps include: S1: collecting and acquiring a water surface image sequence including piles, and performing image processing on the water surface image sequence to acquire a flow field image; S2: Post-processing the flow field image to obtain an optimized flow field image; The post-processing includes image-based verification processing and speed-based verification processing; The image-based verification process includes filtering bright areas of the image based on a high-pass filter, implementing low-contrast areas based on an image threshold segmentation method, and interpolation processing of missing data. The velocity-based verification process includes setting a flow field velocity boundary value, and removing outliers in the flow field velocity based on a standard deviation filter and a local median filter; S3: Calibrate the optimized flow field image with a preset image scaling ratio to convert the pixel coordinates in the optimized flow field image into actual coordinates, and obtain the pile circumference data under the corresponding flow field based on the particle image velocimetry algorithm PIV; The pile peripheral data includes the position coordinates of any point around the pile column at different times and the horizontal velocity component and vertical velocity component of any point; S4: Based on the Morison equation, the wave load on the pile is obtained according to the horizontal velocity component and the vertical velocity component.

2. The method for calculating wave loads on piles based on machine vision according to claim 1, characterized in that: The S1 specifically includes the following steps: S11: setting camera calibration parameters, camera exposure and frequency to obtain a number of water surface sample images containing piles and sequentially encode them to obtain a water surface image sequence; S12: The water surface image sequence is sequentially encoded and imported into the PIVlab toolbox for region selection to obtain a rectangular region of interest ROI; S13: Divide the rectangular area ROI into a number of window image blocks of the same size, and perform adaptive clipping histogram equalization processing on each window image block to obtain an equalized image with enhanced local detail contrast, wherein the local details at least include texture and ripple features of the water surface; S14: filtering the equalized image based on a high-pass filter to obtain a local detail enhancement feature map; S15: removing the image background brightness deviation from the local detail enhancement feature map to obtain an optimized feature map; S16: Perform a cross-correlation operation on the optimized feature map to obtain a flow field image.

3. The method for calculating wave loads on piles based on machine vision according to claim 2, characterized in that: The local detail enhancement feature map described in S15 is used to remove the image background brightness deviation, and its expression is: Where: I′(x,y) represents the pixel intensity after the image background brightness deviation is removed; I(x,y) represents the original pixel intensity of the local detail enhancement feature map; Represents the average intensity of the local detail enhanced feature map.

4. The method for calculating wave loads on piles based on machine vision according to claim 2, characterized in that: The cross-correlation operation is performed on the optimized feature map in S16 to obtain the flow field image, specifically: Assume that the position of the particle in the current window image block at time t is (x, y); the position of the particle in the query window image block at time t+Δt is (x+ξ, y+η); Wherein, the particle represents the flow field feature point that is expected to be tracked in the optimization feature map; ξ represents the displacement of the particle in the x-direction of the preset coordinate axis; η represents the displacement of the particle in the y-direction of the preset coordinate axis; According to the constructed cross-correlation function, the position point of the particle in the query image block at time t+Δt is obtained as (x+ξ,y+η) based on the position point in the current image block as (x,y), so as to realize the flow path tracking of the desired flow field feature point, and then obtain the flow field image; And the constructed cross-correlation function expression is Where: M0 represents the length of the current window image block; N0 represents the width of the current window image block; I1 and I2 represent the pixel grayscale values ​​of the window image block at time t and time t+Δt respectively; i represents the horizontal offset of the window image block; j represents the vertical offset of the window image block; I1(x+i,y+j) represents the pixel grayscale value of the window image block at time t; represents the average pixel grayscale value of the sub-window image block at time t; I2(x+ξ+i,y+η+j) represents the pixel grayscale value of the window image block at time t+Δt; Represents the average pixel grayscale value of the window image block at time t+Δt.

5. The method for calculating wave loads on piles based on machine vision according to claim 4, characterized in that: In S3, the pixel coordinates in the optimized flow field image are converted into actual coordinates, and its expression is x real =x pixel ×Scale and real =and pixel ×Scale Where: x pixel represents the pixel horizontal coordinate of the pile in the optimized flow field image; pixel represents the pixel ordinate of the column in the optimized flow field image; x real represents the actual horizontal coordinate of the pile in the optimized flow field image; real It represents the actual vertical coordinate of the column in the optimized flow field image; Scale represents the actual distance corresponding to each pixel.

6. The method for calculating wave loads on piles based on machine vision according to claim 5, characterized in that: The S4 specifically comprises the following steps: S41: Obtaining the function of the pile structural performance parameters and the wave velocity along the water depth; The structural performance parameters include at least the pile diameter, the pile inertia coefficient and the pile drag coefficient; The function of the wave speed varying along the water depth includes any one of a constant function, a linear function or a quadratic function; And the constant function: V(h)=Va; Linear function: V(h) = (h / H)*Va; Quadratic function: V(h)=(h 2 / H 2 )*Va Where Va represents the wave velocity on the pile at the water surface, and includes the horizontal velocity component u and the vertical velocity component v; h represents the water depth from a certain point of the pile to the water surface; H represents the total water depth; S42: Based on the data around the pile at different times and step S41, the wave load on the pile is obtained based on the Morison equation, and its expression is: Where: F D Indicates the drag force of water on the pile; F H Indicates the wave load on the pile; F I represents the inertial force of water on the pile column; ρ represents the density of water; V represents the abbreviated form of V(h), which is the velocity of the water particle at the axis of the column; D represents the diameter of the pile column; C D Indicates the drag coefficient; C M represents the inertia force coefficient; h0 represents the depth from the water surface to a certain point on the pile; d(h) represents the integral of the water depth h.

Images