A method and device for three-dimensional reconstruction of tidal bores based on binocular cameras

Abstract

The present invention discloses a method and device for three-dimensional reconstruction of tidal bores based on binocular cameras. The method steps include: S1: determining the position of a tidal bore monitoring high tower according to the three-dimensional reconstruction area of the tidal bore, and determining the camera positions and lens parameters of the binocular cameras; S2: building a target field, cyclically moving to photograph the target field to obtain calibration images, associating the pixel coordinates of the target points in the calibration images with the physical coordinates of the targets in the target field, and calculating the internal parameters of the binocular cameras; S3: deploying two binocular cameras, and calculating the external parameters of the two binocular cameras respectively by using the parallax method; S4: the two binocular cameras simultaneously and periodically collect images until the tidal bore ends, use the camera internal parameter calibration to correct the distortion of the collected images, densely reconstruct the three-dimensional point cloud, adjust and output the three-dimensional reconstruction model of the tidal bore after evaluating the accuracy of the three-dimensional point cloud area. The present invention can solve the problem of simultaneous shooting of tidal bore images and achieve high-precision three-dimensional reconstruction of the tidal bore morphology.

Description

Technical Field

[0001] The present invention relates to the technical field of tidal bore monitoring, and particularly to a method and device for three-dimensional reconstruction of tidal bores based on binocular cameras. Background Art

[0002] As a spectacular and complex hydrological phenomenon in nature, the monitoring of tidal bores is of great necessity. When a tidal bore occurs, the advance of the tide poses a serious threat to infrastructure such as dikes, bridges, and docks along the river banks, and also affects the lives and property safety of surrounding residents. Through continuous and accurate monitoring, the dynamic changes of tidal bores can be grasped in a timely manner, providing key data support for the tidal bore early warning system and effectively reducing the losses caused by natural disasters. In addition, tidal bore monitoring is also helpful for scientific research, deepening the understanding of tidal patterns, and providing a scientific basis for the comprehensive management and development of estuary areas.

[0003] Traditional tidal bore monitoring mainly relies on single-point water level monitoring. Since a tidal bore is a large-scale and non-uniform hydrodynamic event, single-point water level monitoring of the tidal bore water level cannot determine the shape of the tidal bore area, cannot meet the accurate assessment of the destructive power of the tidal bore, resulting in a significant discount in the early warning results. At the same time, it cannot provide more detailed tidal bore shape data for scientific research and cannot guarantee the reliability and accuracy of research results.

[0004] In recent years, with the development of three-dimensional monitoring technologies using various bands of radar, high-precision three-dimensional reconstruction has become popular. However, due to the strong absorption and scattering of radar waves by tidal bores, resulting in serious signal attenuation, it is difficult to obtain accurate tidal bore shape information. Therefore, various bands of radar cannot effectively monitor it, resulting in an unsatisfactory three-dimensional reconstruction effect of radar for tidal bores. Binocular vision three-dimensional imaging technology shows significant potential in tidal bore three-dimensional reconstruction monitoring and can effectively address the deficiencies of existing technologies.

[0005] However, for existing binocular camera products on the market, due to the limited baseline distance, the monitoring distance is restricted and the accuracy cannot be guaranteed, thus it is impossible to achieve long-distance three-dimensional reconstruction of tidal bores. How to monitor the shape of tidal bores with high precision remains a technical problem to be solved urgently. Summary of the Invention

[0006] To solve the problem of insufficient accuracy in tidal bore monitoring in the prior art, the present invention proposes a method and device for three-dimensional reconstruction of tidal bores based on binocular cameras.

[0007] The specific technical solution is as follows: A method for three-dimensional reconstruction of tidal bores based on binocular cameras, the steps include:

[0008] S1: Determine the position of the tidal bore monitoring high tower according to the tidal bore three-dimensional reconstruction area, and determine the camera position and lens parameters of the binocular camera according to the tidal bore three-dimensional reconstruction area and the position of the tidal bore monitoring high tower;

[0009] S2: Construct a target range, obtain the physical coordinates of the target, cyclically move and shoot the target range according to the camera position parameters and lens parameters to obtain calibration images, write the pixel coordinates of the target points in the calibration images into the reference coordinate system of the first moving shot, associate the pixel coordinates of the target points in the calibration images with the physical coordinates of the targets in the target range, and calculate the internal parameters of the binocular camera;

[0010] S3: Deploy two binocular cameras, measure three co-visible calibration points within the image range of the two binocular cameras through a total station, and calculate the external parameters of the two binocular cameras respectively using the parallax method;

[0011] S4: The two binocular cameras simultaneously and periodically collect images until the end of the tidal bore, use the camera internal parameter calibration to correct the distortion of the collected images, densely reconstruct the three-dimensional point cloud according to the camera internal parameters and external parameters, and adjust and output the three-dimensional reconstruction model of the tidal bore after evaluating the accuracy of the three-dimensional point cloud area.

[0012] Furthermore, the acquisition of the camera internal parameters specifically includes:

[0013] Cyclically move and shoot according to the camera position parameters and lens parameters, and splice all the captured images to obtain calibration images;

[0014] Obtain the physical position of the binocular camera during each moving shot and calculate the rotation parameters, and use the transformation relationship to uniformly write the pixel coordinates of the target points in all the captured images into the reference coordinate system of the first moving shot;

[0015] Associate the pixel coordinates of the target points in the calibration images with the physical coordinates of the targets in the target range, and obtain the distortion coefficient and internal parameter matrix of the binocular camera through the calculation of the distortion coefficient.

[0016] Furthermore, the calculation of the rotation parameters includes: measuring the coordinates C1 of the binocular camera and the coordinates P1 of the target point corresponding to the image center point through a total station;

[0017] Match the image center point and find the coordinates P1 of the corresponding target point in the world coordinate system;

[0018] Define the x-axis direction of the initial coordinate system with the vector from the camera position of the binocular camera to the target point:

[0019] ;

[0020] ;

[0021] Among them, represents the vector from the camera position to the target point, represents the unitized vector, pointing to the x-axis vector of the reference coordinate system;

[0022] Select the vertical direction as the y-axis direction in the plane perpendicular to and denote it as . Calculate the z-axis direction through cross product: to obtain the rotation matrix parameters for the first shot: .

[0023] Furthermore, the calculation of the distortion coefficient includes:

[0024] Extract the position information of the target in the image through an image processing algorithm, and match it with the coordinates of the target point in the world coordinate system to obtain the homography matrix of the mapping relationship between the calibration image and the target field plane;

[0025] Derive the distortion coefficient and internal parameter matrix of the binocular camera in reverse through the coordinates of the calibration points and their projection coordinates in the calibration image.

[0026] Furthermore, the deployment of the binocular camera includes: Fix two binocular cameras at both ends of the camera bracket respectively, and install the camera bracket on the tidal bore monitoring high tower.

[0027] Furthermore, the image acquisition specifically includes:

[0028] The industrial control computer system of the tidal bore monitoring high tower synchronizes the time of the two binocular cameras using the precise time synchronization protocol to keep them always consistent;

[0029] The industrial controller continuously issues the acquisition image instructions at the same moment to the two binocular cameras according to the periodic acquisition requirements, and the binocular cameras take pictures according to the instruction moment after receiving the instructions.

[0030] Furthermore, the dense reconstruction of the three-dimensional point cloud includes:

[0031] Use the PatchMatch algorithm to estimate the disparity of each pixel in the images of two perspectives for the internal and external parameters of the camera, and replace the disparity of the current pixel with the disparity with the minimum matching cost among the neighboring pixels;

[0032] Measure the similarity of two Patches, use triangulation to convert the disparity of the pixel points into depth information, and obtain the dense three-dimensional point cloud.

[0033] Furthermore, the accuracy evaluation includes:

[0034] Conduct a ship verification in the tidal bore three-dimensional reconstruction area. Place RTK equipment and a target on the ship, set multiple control point positions in the tidal bore three-dimensional reconstruction area, sail the ship to each control point and stay, record the time interval between the first stay point and the last stay point and continuously record the RTK values, and the two binocular cameras simultaneously take multiple groups of images at equal time intervals;

[0035] Compare the target positions in multiple groups of images, calculate the average pixel deviation of the target positions in the images, intercept all RTK values within the time interval, and calculate the RTK average value;

[0036] Determine whether the absolute values of the pixel average deviation and the differences between all RTK values and the RTK average value are all less than the threshold. If all the judgment conditions are met, perform 3D reconstruction on the captured images;

[0037] Obtain the RTK average value of the control points that meet the judgment conditions among all control points and the 3D reconstruction coordinates of the binocular cameras, and compare the RTK average value and the 3D reconstruction coordinates to obtain the mean absolute error, mean height error, and mean horizontal error.

[0038] An apparatus for 3D reconstruction of tidal bores based on binocular cameras, applicable to a method for 3D reconstruction of tidal bores based on binocular cameras, includes: two binocular cameras and a tidal bore monitoring high tower. The tidal bore monitoring high tower is installed on the seawall. A camera bracket is provided on the tidal bore monitoring high tower, and the two binocular cameras are respectively installed at both ends of the camera bracket. The tidal bore 3D reconstruction area is smaller than the overlapping area of the shooting ranges of the two binocular cameras.

[0039] Furthermore, a distribution box is provided at the bottom of the tidal bore monitoring high tower. A power supply, an industrial control computer, and a network router are provided in the distribution box. The industrial control computer and the network router are electrically connected to the power supply. An image acquisition card is provided on the industrial control computer. The industrial control computer is communicatively connected to the two binocular cameras, and the power supply is electrically connected to the two binocular cameras.

[0040] The above technical solutions have the following advantages or technical effects:

[0041] 1. The present invention uses binocular cameras to perform 3D reconstruction on the tidal bore morphology, realizing the improvement of tidal bore monitoring from single points to a surface, and making up for the gap in tidal bore morphology monitoring.

[0042] 2. The present invention improves the image quality through camera internal parameter calibration and distortion correction. Through the self-developed integrated camera bracket design, the baseline distance between the two binocular cameras is increased. At the same time, through field calibration point measurement, the accuracy of camera external parameter calculation is ensured. The combination of multiple methods improves the accuracy of tidal bore 3D reconstruction.

[0043] 3. The present invention uses an accurate time synchronization protocol to achieve microsecond-level time synchronization, realizing the simultaneous acquisition of tidal bore images by the two cameras, and solving the problem of the flowing change at the tidal bore moment.

[0044] 4. The present invention uses 3D point cloud dense reconstruction to improve the point cloud density of the sparse point cloud of 3D reconstruction of some images, thereby improving the usability of the tidal bore 3D reconstruction point cloud.

[0045] 5. The present invention adopts ship-opening accuracy verification to ensure the accuracy of the point cloud and ensure that the reconstructed model meets the precise monitoring standard. Description of the Drawings

[0046] Figure 1 is the flowchart of the method of the present invention;

[0047] Figure 2 is the image captured by the left-view binocular camera of the present invention;

[0048] Figure 3 is the image captured by the right-view binocular camera of the present invention;

[0049] Figure 4 is the distortion correction diagram of the image captured from the left view of the present invention;

[0050] Figure 5 is the distortion correction diagram of the image captured from the right view of the present invention;

[0051] Figure 6 is the distribution diagram of the tidal bore point cloud from the top view of the present invention;

[0052] Figure 7 is the schematic diagram of the position of the device vertical pole of the present invention;

[0053] Figure 8 is the schematic diagram of the device of the present invention.

[0054] In the drawings, the labels are: 1 - binocular camera, 2 - camera bracket, 3 - tidal bore monitoring high tower, 4 - seawall, 5 - river surface, 6 - tidal bore three-dimensional reconstruction area, 7 - camera network cable, 8 - camera power cable, 9 - distribution box, 10 - industrial control computer, 11 - image acquisition card, 12 - power supply, 13 - network router. Detailed Embodiment

[0055] To make the technical solution of the present invention clearer, the following further describes the present invention in detail with reference to the drawings and specific embodiments.

[0056] Embodiment 1

[0057] As Figure 1 shown, a method for three-dimensional reconstruction of a tidal bore based on a binocular camera includes the following steps:

[0058] S1: Determine the position of the tidal bore monitoring high tower according to the tidal bore three-dimensional reconstruction area, and determine the camera position and lens parameters of the binocular camera according to the tidal bore three-dimensional reconstruction area and the position of the tidal bore monitoring high tower;

[0059] S2: Construct a target range, obtain the physical coordinates of the target, cyclically move and shoot the target range according to the camera position parameters and lens parameters to obtain calibration images, write the pixel coordinates of the target points in the calibration images into the reference coordinate system of the first moving shot, associate the pixel coordinates of the target points in the calibration images with the physical coordinates of the targets in the target range, and calculate the internal parameters of the binocular camera;

[0060] S3: Deploy two binocular cameras, measure three co-visible calibration points within the image ranges of the two binocular cameras through a total station, and calculate the external parameters of the two binocular cameras respectively using the parallax method;

[0061] S4: The two binocular cameras collect images periodically at the same time until the end of the tidal bore. Use the camera internal parameter calibration to correct the distortion of the collected images, densely reconstruct the three-dimensional point cloud according to the camera internal parameters and external parameters, and adjust and output the three-dimensional reconstruction model of the tidal bore after evaluating the accuracy of the three-dimensional point cloud region.

[0062] Step S1 specifically includes:

[0063] S11: According to the selected three-dimensional reconstruction area of the tidal bore, determine the position of the tidal bore monitoring tower so that the tidal bore monitoring tower will not shake and is located at the place closest to the river surface when the tidal bore arrives;

[0064] S12: According to the three-dimensional reconstruction area of the tidal bore and the position of the tidal bore monitoring tower, set up a binocular camera for top-down shooting, cover the three-dimensional reconstruction area of the tidal bore with the shooting range, and determine the camera position of the binocular camera. The camera position includes the camera shooting angle and the shooting distance L. Among them, the camera shooting angle ensures that while shooting the river surface from above, the shooting range covers the three-dimensional reconstruction area of the tidal bore, and the shooting distance L is the physical distance between the binocular camera and the center point of the captured image;

[0065] S13: Adjust the camera lens focal length according to the camera shooting angle and the shooting distance L to make the pixel ratio of the three-dimensional reconstruction area of the tidal bore in the captured image the largest. At the same time, adjust the camera lens aperture to ensure clear image formation, and determine the lens parameters of the binocular camera. The lens parameters include the camera lens focal length f and the camera lens aperture. Among them, ensuring clear image formation is comprehensively judged by adjusting the camera lens aperture and observing the change curves of image sharpness and contrast.

[0066] Among them, the height of the tidal bore monitoring tower is 25 meters.

[0067] Camera internal parameter calibration: Use Zhang's calibration method to calibrate the camera internal parameters. Construct a target range, and obtain the physical coordinates of the targets in the target range through measurement. Among them, the target range is a uniformly distributed target range (such as a checkerboard or a dot array), and use a total station to measure the pixel coordinates of each target point.

[0068] Embodiment 2

[0069] Obtain the camera internal parameters:

[0070] Perform cyclic moving shooting according to the shooting distance L of the binocular camera from the target range and the focal length f of the binocular camera lens, record the reference coordinate system at the first moving shooting, and splice all the captured images to obtain a calibration image. Among them, cyclic moving shooting means obtaining an image set by cyclically shooting the target range by moving the binocular camera, and then using the image set to splice and generate a calibration image that covers the entire visual field of the target range, so as to ensure the accuracy of the distortion coefficient calculation; the reference coordinate system is the position of the binocular camera corresponding to the target point at the image center at the first moving shooting. Define a reference coordinate system, which will be used as the reference for all subsequent calculations. This coordinate system defines the x-axis direction with the vector from the binocular camera position to the target point, and uses the vertical direction as the y-axis direction. The target point corresponding to the image center is measured by moving the target to the center point of the image;

[0071] Obtain the physical position of the binocular camera at each shooting and calculate the rotation parameters, and use the transformation relationship to uniformly write the pixel coordinates of the target points in all the captured images into the reference coordinate system of the first moving shooting;

[0072] Associate the pixel coordinates of the target points in the calibration image with the physical coordinates of the targets in the target range, and obtain the distortion coefficient and internal parameter matrix of the binocular camera through distortion coefficient calculation

[0073] Among them, the calculation of the rotation parameters includes: measuring the coordinates C1 of the binocular camera and the coordinates P1 of the target point corresponding to the image center point by a total station,

[0074] Among them, , represents the coordinates of C1 on the x-axis, y-axis, and z-axis;

[0075] Match the image center point and find the coordinates P1 of the corresponding target point in the world coordinate system,

[0076] Among them, , represents the coordinates of P1 on the x-axis, y-axis, and z-axis;

[0077] Define the x-axis direction of the initial coordinate system with the vector from the camera position of the binocular camera to the target point:

[0078] ;

[0079] ;

[0080] Among them, represents the vector from the camera position to the target point, represents The unitized vector, which points to the x-axis vector of the reference coordinate system;

[0081] Select the vertical direction as the y-axis direction in the plane perpendicular to and denote it as . Calculate the z-axis direction through cross product: to obtain the rotation matrix parameters for the first shot: .

[0082] Among them, the calculation of the distortion coefficients includes:

[0083] Extract the position information of the target in the image through an image processing algorithm, and match it with the coordinates of the target point in the world coordinate system to obtain the homography matrix of the mapping relationship between the calibration image and the target field plane;

[0084] Use the nonlinear least squares method for optimization. Through the coordinates of the calibration points and their projected coordinates in the calibration image, inversely deduce the distortion coefficients and internal parameter matrix of the binocular camera. The distortion coefficients mainly include: radial distortion coefficients k1, k2, k3, etc. and tangential distortion coefficients p1, p2.

[0085] Example 3

[0086] Calculate the external parameters of the camera: Obtain the external parameters of the camera through camera installation and deployment and field calibration point measurement to obtain the external parameter matrix.

[0087] Fix two binocular cameras at both ends of the bracket through an integrated camera bracket, install the camera bracket on the high tower, connect the network cable and power cable to the industrial computer at the bottom of the high tower, and control the camera shooting and 3D reconstruction through the system on the industrial computer. At the same time, the camera bracket ensures that the two cameras are fixed and can only rotate up and down and move left and right simultaneously. Such a hardware design ensures that the relative external parameters of the two cameras do not change, reduces the error caused by measurement, and improves the accuracy. In conventional acquisition and shooting, the camera settings usually use single-point shooting or place the camera at a similar position. However, when dealing with tidal bores, such settings cannot accurately monitor the dynamic of the tidal bore dozens of meters away and are prone to errors. The length of the camera bracket in the present invention is 6 meters, and the baseline distance is also 6 meters. By simultaneously shooting with two binocular cameras within a relatively long baseline range, the accuracy of long-distance 3D reconstruction can be improved, and the baseline length can ensure the synchronous movement of the two binocular cameras, further reducing the monitoring error.

[0088] The field calibration point measurement is to measure three co-visible calibration points within the image range of the two binocular cameras through a total station. Use the parallax method to calculate the spatial physical positions and rotation directions of the two cameras respectively through the calibration points. Since the hardware design ensures that the relative external parameters of the two cameras do not change, averaging the rotation directions can reduce the measurement error, thereby obtaining an accurate external parameter matrix.

[0089] Example 4

[0090] Collect images at the same time: When the tidal bore arrives, the system on the industrial control computer enables the image collection function. Two binocular cameras collect images simultaneously once per second and continuously collect until the tidal bore ends.

[0091] The industrial control computer system of the tidal bore monitoring high tower synchronizes the time of the two binocular cameras using the precise time synchronization protocol to keep them always consistent.

[0092] Verify the time synchronization of the two binocular cameras. Issue a shooting instruction to make the two binocular cameras photograph the running microsecond-level timer at the same time, and determine whether the error of the time alignment result is accurate to the microsecond level. If the error range exceeds the specified threshold, repeat the time synchronization operation until the time alignment result reaches the microsecond level.

[0093] According to the periodic collection requirements, the industrial controller continuously issues image collection instructions at the same time to the two binocular cameras. After receiving the instructions, the two binocular cameras on the left and right perspectives take pictures according to the instruction time, and the results are as Figure 2 、 Figure 3 shown;

[0094] After completing the collection requirements, the industrial control computer stops issuing image collection instructions.

[0095] Image distortion correction: Use the distortion coefficients obtained from the camera internal parameter calibration to correct the distortion of the collected images. The correction results are as Figure 4 、 Figure 5 shown.

[0096] Example 5

[0097] Dense three-dimensional point cloud reconstruction: Perform dense three-dimensional point cloud reconstruction on the images after distortion correction. Using the internal parameters of the binocular cameras and the external parameters between the two binocular cameras, use the PatchMatch algorithm to perform dense matching reconstruction on the images. The PatchMatch algorithm estimates the disparity of each pixel in the multi-view images through three steps: random initialization, neighborhood propagation, and random search disparity correction. In the neighborhood propagation of the PatchMatch algorithm, a "greedy" strategy (GreedyStrategy) is adopted, that is, the disparity of the current pixel will be replaced by the disparity with the smallest matching cost among its neighboring pixels.

[0098] Assume that the disparity of the current pixel point p is , and the disparities of its neighboring pixel points and are and respectively. During the neighborhood propagation, the algorithm will compare the cost function of pixel point p with and Based on the size, select the disparity with the smallest cost as the new disparity value:

[0099] ;

[0100] In cost calculation, the photometric consistency of the local area is used to measure the similarity between two patches. Denote and as the images from two viewpoints, p and q represent the pixel point and its neighborhood respectively, d is the disparity value, and the definition of photometric consistency is as follows:

[0101] ;

[0102] Through the internal and external parameters of the binocular camera, the triangulation method can be used to convert the disparity of the pixel point into depth information, and then a dense point cloud can be obtained. As Figure 6 is the distribution map of the tidal bore point cloud from the top-down view, and the height of the point cloud and the distance from the shore can be accurately monitored.

[0103] Accuracy evaluation of the three-dimensional reconstruction area of the tidal bore: Conduct accuracy evaluation on the three-dimensional reconstruction area of the tidal bore to obtain the mean absolute error, mean height error, and mean horizontal error within the current area.

[0104] The method of accuracy evaluation refers to conducting a ship verification in the three-dimensional reconstruction area of the tidal bore. Place RTK equipment and a target on the ship. When the water surface is calm, evenly set 20 control point positions in the three-dimensional reconstruction area of the tidal bore. Sail the ship to each control point and stay, continuously record the RTK values and time. At the same time, take 10 photos with the binocular camera at a rate of 1 photo per second. Compare the target positions in the 10 groups of images. If the average pixel deviation of the target positions in the 10 groups of images is less than 2, and at the same time intercept the RTK values during this time period and calculate the RTK average value. If the absolute value of the difference between all RTK values and the RTK average value is less than 2 cm, then the data is available. Then conduct three-dimensional reconstruction on the taken images and calculate the physical coordinates of the control points. Through 20 loops, obtain the RTK average values of the 20 control points and the physical coordinates of the three-dimensional reconstruction by the binocular camera. Calculate the mean absolute error, mean height error, and mean horizontal error through the comparison of the two coordinates.

[0105] The mean absolute error refers to the average value of the error between the RTK measurement and the physical coordinates of the three-dimensional reconstruction. The mean height error is the average value of the error in height, and the mean horizontal error is the average value of the error in the horizontal direction.

[0106] The judgment threshold for accuracy evaluation is 20 cm. When the mean absolute error of accuracy evaluation is greater than 20 cm, the three-dimensional reconstruction area of the tidal bore is modified again, the camera position and lens parameters are adjusted, and the operations in steps S2 to S4 are performed again until the mean absolute error of accuracy evaluation does not exceed the specified threshold range, and the three-dimensional reconstruction model of the tidal bore is output.

[0107] Taking the first shooting of the left-view binocular camera at 20 control point positions as an example, it is denoted as camera 1, and the control point information collected by camera 1 is shown in Table 1.

[0108] Table 1: Control point information collected by camera 1 for the first time

[0109]

[0110] Perform three-dimensional reconstruction on all the collected control point information. Taking the first shooting at 20 control point positions as an example, the reconstruction results are shown in Table 2;

[0111] Table 2: Reconstruction results of the information collected by camera 1 for the first time

[0112]

[0113] Calculate the error between the two sets of data, and the calculation results are shown in Table 3;

[0114] Table 3: Error calculation results

[0115]

[0116] The mean absolute error in Table 3 is 8.1 cm, which is less than 20 cm. Therefore, the model accuracy meets the requirements, and thus the three-dimensional reconstruction of the tidal bore model is completed.

[0117] Embodiment 6

[0118] As Figure 7 、 Figure 8 shown, a device for three-dimensional reconstruction of a tidal bore based on binocular cameras includes: two binocular cameras 1 and a tidal bore monitoring high tower 3. The tidal bore monitoring high tower 3 is installed on the seawall 4 and is located at the position closest to the river surface 5 without shaking when the tidal bore arrives. A camera support 2 is provided on the tidal bore monitoring high tower 3, and the two binocular cameras 1 are respectively installed at both ends of the camera support 2. The three-dimensional reconstruction area 6 of the tidal bore is smaller than the overlapping area of the shooting ranges of the two binocular cameras 1. The tidal bore monitoring high tower is 25 m, and the camera support is 6 m.

[0119] At the bottom of the tidal bore monitoring tower 3, there is a distribution box 9. Inside the distribution box 9, there is a power supply 12, an industrial control computer 10 and a network router 13. The industrial control computer 10 and the network router 13 are electrically connected to the power supply 12. The industrial control computer 10 is communicatively connected to the network router 13. An image acquisition card 11 is provided on the industrial control computer 10. The industrial control computer 10 is communicatively connected to two binocular cameras 1 through a camera network cable 7, for controlling the two binocular cameras 1 to achieve precise synchronization. The power supply 12 is electrically connected to the two binocular cameras 1 through a camera power cable 8.

[0120] The present invention adopts binocular vision three-dimensional imaging technology and self-developed devices, solves the image distortion problem through camera internal parameter calibration and image distortion correction, solves the problem of simultaneous shooting of tidal bore images through microsecond-level time synchronization, solves the problem of sparse point clouds in three-dimensional reconstruction of a small number of images through dense three-dimensional point cloud reconstruction, and finally constructs a high-precision three-dimensional point cloud model to achieve high-precision three-dimensional reconstruction of the tidal bore morphology.

[0121] The above embodiments only represent several implementation manners of the present invention. The description is relatively specific and detailed, but it should not be construed as a limitation on the scope of the present invention patent. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several deformations and improvements can still be made, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the present invention patent should be subject to the appended claims.

Claims

1. A method for three-dimensional reconstruction of tidal bores based on binocular cameras, characterized in that the steps Including: S1: Determine the position of the tidal bore monitoring tower according to the three-dimensional reconstruction area of the tidal bore, and determine the camera position and lens parameters of the binocular camera according to the three-dimensional reconstruction area of the tidal bore and the position of the tidal bore monitoring tower; S2: Construct a target range, obtain the physical coordinates of the target, cyclically move and shoot the target range according to the camera position parameters and lens parameters to obtain calibration images, write the pixel coordinates of the target points in the calibration images into the reference coordinate system of the first moving shot, associate the pixel coordinates of the target points in the calibration images with the physical coordinates of the targets in the target range, and calculate the internal parameters of the binocular camera; S3: Deploy two binocular cameras, measure three co-visible calibration points within the image range of the two binocular cameras by total station, and calculate the external parameters of the two binocular cameras respectively using the parallax method; S4: The two binocular cameras collect images periodically at the same time until the tidal bore ends, use the camera internal parameter calibration to correct the distortion of the collected images, densely reconstruct the three-dimensional point cloud according to the camera internal parameters and external parameters, adjust and output the three-dimensional reconstruction model of the tidal bore after evaluating the accuracy of the three-dimensional point cloud area; The acquisition of the camera internal parameters specifically includes: cyclically moving and shooting according to the camera position parameters and lens parameters, and splicing all the captured images to obtain calibration images; Obtain the physical position of the binocular camera during each moving shot and calculate the rotation parameters, and use the transformation relationship to uniformly write the pixel coordinates of the target points in all the captured images into the reference coordinate system of the first moving shot; Associate the pixel coordinates of the target points in the calibration images with the physical coordinates of the targets in the target range, and calculate the distortion coefficient and internal parameter matrix of the binocular camera through the distortion coefficient calculation; The calculation of the rotation parameters includes: measuring the coordinates C1 of the binocular camera and the coordinates P1 of the target point corresponding to the image center point by total station; Match the image center point and find the coordinates P1 of the corresponding target point in the world coordinate system; Define the x-axis direction of the initial coordinate system with the vector from the camera position of the binocular camera to the target point: v x = P1 - C1; Among them, v x represents the vector from the camera position to the target point, and e x represents the vector after normalizing v x and points to the x-axis vector of the reference coordinate system; Select the vertical direction as the y-axis direction in the plane perpendicular to e x , denoted as e y . Calculate the z-axis direction through cross product: e z = e x × e y . Obtain the rotation matrix parameters for the first shot: R1 = [e x e y e z ; The accuracy evaluation includes conducting a ship verification in the three-dimensional reconstruction area of the tidal bore, placing RTK equipment and targets on the ship, setting multiple control point positions in the three-dimensional reconstruction area of the tidal bore, sailing the ship to each control point and staying, recording the time interval between the first stay point and the last stay point and continuously recording the RTK values, and the two binocular cameras simultaneously shoot multiple groups of images at equal time intervals; Compare the target positions in multiple groups of images, calculate the average pixel deviation of the target positions in the images, intercept all the RTK values during the time interval, and calculate the RTK average value; Judge whether the absolute values of the pixel average deviation, the difference between all the RTK values and the RTK average value are all less than the threshold. If all the judgment conditions are met, conduct three-dimensional reconstruction on the captured images; Obtain the RTK average value of the control points that meet the judgment conditions among all the control points and the three-dimensional reconstruction coordinates of the binocular camera, and compare the RTK average value and the three-dimensional reconstruction coordinates to obtain the mean absolute error, mean height error and mean horizontal error.

2. The method for three-dimensional reconstruction of a tidal bore based on a binocular camera according to claim 1, characterized in that The calculation of the distortion coefficient includes: Extract the position information of the target in the image through the image processing algorithm, and match it with the coordinates of the target point in the world coordinate system to obtain the homography matrix of the mapping relationship between the calibration image and the target field plane; Derive the distortion coefficients and internal parameter matrix of the binocular camera in reverse through the coordinates of the calibration points and their projection coordinates in the calibration image.

3. The method for three-dimensional reconstruction of tidal bores based on a binocular camera according to claim 1, wherein The deployment of the binocular camera includes: fixing two binocular cameras at both ends of the camera bracket respectively, and installing the camera bracket on the tidal bore monitoring high tower.

4. A method for three-dimensional reconstruction of a tidal bore based on a binocular camera according to claim 1, characterized in that, The specific image acquisition includes: The industrial control computer system of the tidal bore monitoring high tower synchronizes the time of the two binocular cameras using the precise time synchronization protocol to keep them always consistent; According to the periodic acquisition requirements, the industrial controller continuously sends the acquisition image instructions at the same moment to the two binocular cameras. After receiving the instructions, the binocular cameras take pictures according to the instruction moment.

5. A method for three-dimensional reconstruction of tidal bores based on binocular cameras according to claim 1, characterized in that, The dense reconstruction three-dimensional point cloud includes: Use the PatchMatch algorithm to estimate the disparity of each pixel in the two-view images for the internal and external parameters of the camera, and replace the disparity of the current pixel with the disparity with the minimum matching cost among the neighboring pixels; Measure the similarity between two Patches, and use triangulation to convert the disparity of the pixel points into depth information to obtain a dense three-dimensional point cloud.

6. A device for three-dimensional reconstruction of a tidal bore based on a binocular camera, applicable to the method for three-dimensional reconstruction of a tidal bore based on a binocular camera according to any one of claims 1 to 5, characterized in that, Include: Two binocular cameras and a tidal bore monitoring high tower. The tidal bore monitoring high tower is located on the seawall. There is a camera bracket on the tidal bore monitoring high tower. The two binocular cameras are respectively arranged at both ends of the camera bracket. The tidal bore three-dimensional reconstruction area is smaller than the overlapping area of the shooting ranges of the two binocular cameras.

7. The device for three-dimensional reconstruction of tidal bores based on a binocular camera according to claim 6, characterized in that, A distribution box is provided at the bottom of the tidal bore monitoring high tower. A power supply, an industrial control computer and a network router are provided in the distribution box. The industrial control computer and the network router are electrically connected to the power supply. An image acquisition card is provided on the industrial control computer. The industrial control computer is communicatively connected to the two binocular cameras. The power supply is electrically connected to the two binocular cameras.

Images