A camera attitude angle calibration method using the water edge of river section

Abstract

The present invention discloses a method for calibrating camera attitude angles using the waterside of a river section, comprising: completing the camera's internal parameter calibration to obtain the corresponding internal parameter matrix and distortion parameter matrix; performing distortion correction on the original image; adjusting the azimuth angle so that the marked point in the distortion-corrected image coincides with the longitudinal axis, and setting the azimuth angle to zero; adjusting the pitch angle so that the field of view covers the water surface area to be measured, including the waterside, observing the horizontality of the waterside to level the roll angle, surveying the starting distance and elevation of the camera's optical center, and establishing a water surface photogrammetry model; selecting two points on the waterside of the section in the distortion-corrected image to determine the roll angle; and substituting the sub-pixel image coordinates of the intersection of the line connecting the two points and the longitudinal axis of the image, as well as the world coordinates of the waterside obtained from the water level and cross-sectional topography, into the water surface photogrammetry model to calculate the pitch angle. The present invention can achieve high-precision on-site attitude angle calibration without the assistance of sensors and is suitable for video monitoring systems for river water level, flow velocity, floating objects, etc.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent water conservancy video monitoring, and relates to a camera attitude angle on-site calibration method for river surface photogrammetry, and in particular to a camera attitude angle calibration method using the water edge line of a river section. Background Art

[0002] In recent years, with the rapid development of video surveillance and artificial intelligence technologies, new online monitoring technologies for water levels, flow velocities, and discharges using machine vision have gradually become an important means of modernizing hydrological monitoring. Compared to traditional monitoring methods, they offer advantages such as intuitive, non-contact, safe, efficient, and low-cost. According to the principles of photogrammetry, in order to use ordinary surveillance cameras for water surface measurement, their internal and external parameters must be calibrated to establish a mapping relationship between horizontal plane coordinates and image plane coordinates. Internal parameters include internal orientation elements such as the camera's focal length, image principal point coordinates, and radial and tangential distortion coefficients. External parameters include the camera's three-dimensional world coordinates and six external orientation elements consisting of three-axis attitude angles. Calibration of these parameters is called camera calibration and is a crucial step in image-based flow measurement technology.

[0003] In the measurement of the surface velocity field of laboratory flumes and river engineering models, the camera is usually placed perpendicular to the test plane to shoot, and the scale factor describing the orthographic projection relationship is directly calculated by setting a reference object of known size on the plane to be measured. A more accurate method is to use a calibration target or more than four non-collinear control points, and based on the classic direct linear transformation (DLT) method in the field of close-range photogrammetry, the homography matrix is solved as a whole to establish a reversible coordinate transformation relationship between the image plane and the object plane. This method has the characteristics of a simple model and does not require initial values of internal and external orientation elements. However, it must be ensured that the control points are strictly coplanar with the plane to be measured, otherwise their projection on the image will not reflect the actual water surface elevation. However, this is often difficult to control in actual test environments, resulting in calibration errors.

[0004] In contrast, surveying conditions on large-scale river surfaces are more complex. First, to cover the entire measurement section, cameras are typically mounted on the shore and captured at a low angle to obtain a large field of view of hundreds or even thousands of square meters, resulting in severe image perspective distortion. Image orthorectification can correct this distortion and, to a certain extent, compensate for the loss of far-field spatial resolution. However, this significantly increases the computational and storage requirements for image transformations and may introduce grayscale interpolation errors. Second, river water levels fluctuate significantly, reaching several meters in a short period of time in mountain streams. Changes in river surface elevation alter the image projection relationship, thereby affecting the scale and orientation of the measurement area. Furthermore, the simultaneous placement of control points on the river surface presents practical difficulties. Some studies have attempted to establish projection relationships using floating calibration plates or laser projection points, which can meet the measurement requirements of small-scale flow fields (tens to hundreds of square meters) at large shooting angles. However, these reference objects often have poor visibility under complex lighting conditions such as surface shadows and glare, making automatic detection and precise positioning in the image difficult. When applied to large-scale measurements, this detection error is significantly amplified. Currently, a commonly used velocity calibration scheme is based on the variable-height homography method. This approach, based on a planar homography model, considers the distance between the control point and the water surface and introduces water level and gradient parameters to characterize the dynamic changes in water surface elevation. This approach can effectively improve the measurement accuracy of shore-based LSPIV systems at low inclination angles. However, at least six non-coplanar control points must be evenly distributed along both sides of the river, and their world coordinates must be surveyed using specialized equipment such as total stations and DGPS. This approach is not only time-consuming and labor-intensive, but also poses a serious threat to the safety of surveyors during periods of high flooding. Furthermore, studies have found that the VHH DLT method is sensitive to the number and distribution of control points, detection accuracy, and lens distortion, which can easily lead to non-convergence in the model parameter solution process.

[0005] A two-step camera calibration method used in computer vision offers a solution for reducing the reliance on control points during field measurements. It accounts for nonlinear lens distortion and decomposes the homography matrix into intrinsic and extrinsic parameter matrices. By decomposing the photogrammetry task into two steps—in-house calibration of intrinsic parameters and distortion aberrations, and in-field calibration of extrinsic parameters such as translation and rotation—the number of control points required in the field can be reduced to four, increasing system deployment flexibility. For example, a rotational calibration method based on a two-dimensional precision gimbal (PTZ) first calibrates the initial values of the camera's exterior orientation parameters using control points with good near-field visibility. The camera is then rotated to the measurement pose and compensated for changes in azimuth and pitch angles using the gimbal's dials. Because the calibration process requires manual adjustment of the camera pose and readings, automated measurement is not currently supported. If the camera pose changes due to external interference, the existing calibration results become inapplicable. Re-surveying control points and recalibrating the calibration can lead to inconsistent measurement coordinate systems, complicating data utilization. An improved approach involves rigidly connecting a camera to an attitude sensor to form a direct sensor directional photogrammetry system, using the attitude sensor to measure the camera's three-axis attitude angle. However, this method requires not only high measurement accuracy from the attitude sensor but also precise calibration of the eccentricity and eccentricity of the two during system integration. Furthermore, compensation for the measured attitude angle during photogrammetry is required to ensure measurement accuracy. This results in high system cost and calibration complexity. Therefore, developing a simple, easy-to-use, on-site camera attitude angle calibration technology that requires no image control is essential for achieving photogrammetry of river surfaces and automated online monitoring of multiple factors, including water level, flow velocity, flow rate, and floating debris. Summary of the Invention

[0006] Purpose of the invention: To address the problem of on-site calibration of camera attitude angles in river surface photogrammetry systems, a camera attitude angle calibration method using the water edge of a river section is provided. This method can achieve high-precision on-site calibration of attitude angles without the assistance of sensors and is suitable for video monitoring systems for river water level, flow velocity, floating objects, etc.

[0007] Technical solution: To achieve the above-mentioned purpose, the present invention provides a method for calibrating the camera attitude angle using the water edge of a river section, comprising the following steps:

[0008] S1: Complete the camera's intrinsic calibration and obtain the corresponding intrinsic parameter matrix and distortion parameter matrix;

[0009] S2: Perform distortion correction on the original image according to the intrinsic parameter matrix and the distortion parameter matrix;

[0010] S3: Set the measurement section marking points and set up the camera on the extension line of the two points. Adjust the azimuth angle so that the marking points in the distortion correction image coincide with the longitudinal axis, and set the azimuth angle to zero.

[0011] S4: Adjust the pitch angle so that the field of view covers the water surface area to be measured including the water edge, observe the horizontality of the water edge to level the roll angle, survey the starting distance and elevation of the camera's optical center, and establish a water surface photogrammetry model;

[0012] S5: Select two points on the water edge of the cross section in the distortion-corrected image, and determine the roll angle by calculating the slope of the line connecting the two points;

[0013] S6: Substitute the sub-pixel image coordinates of the intersection of the line connecting the two points and the longitudinal axis of the image, as well as the world coordinates of the water edge obtained from the water level and cross-sectional topography, into the water surface photogrammetry model to calculate the pitch angle.

[0014] Furthermore, in step S1, the camera is calibrated with internal parameters in the laboratory according to the Zhang Zhengyou calibration method. After calibration, the pixel size s, the internal parameter matrix K and the distortion parameter matrix D are obtained as follows:

[0015]

[0016] D=[k1 k2 p1 p2] (2)

[0017] Among them, C x Represents the horizontal coordinate of the principal point, C y represents the vertical coordinate of the principal point, f x Indicates the equivalent focal length of the camera on the x-axis of the image plane, f y It represents the equivalent focal length of the camera on the y-axis of the image plane. The focal length f of the camera is calculated based on the pixel size s of the image sensor:

[0018] f=(f x +f y )·s / 2 (3)

[0019] k1 represents the first radial distortion coefficient, p1 represents the first tangential distortion coefficient, k2 represents the second radial distortion coefficient, and p2 represents the second tangential distortion coefficient.

[0020] Furthermore, in step S2, the image is subjected to distortion correction using the following formula:

[0021]

[0022] The camera coordinates (x, y) of the undistorted image are converted to image coordinates (u, v) using the following formula:

[0023]

[0024] The camera coordinates (x′, y′) of the distorted image are converted to image coordinates (u′, v′) in the same way:

[0025]

[0026] At this point, the image distortion correction is completed.

[0027] Furthermore, the process of establishing the water surface photogrammetry model in step S4 is as follows:

[0028] Establish a central projection model, which can be expressed as:

[0029]

[0030]

[0031] Where m1~m 12 is the element of the projection matrix and has no practical meaning; (X, Y, Z) is the three-dimensional space coordinate of the object point, is the pixel coordinate of the image point;

[0032] By introducing the central projection model into the water surface photogrammetry model, the object-image collinearity relationship of the model can be expressed as:

[0033]

[0034]

[0035] This formula uses the gradient coefficient to express the water surface elevation Z, D1 is the gradient coefficient of the river section in the cross-section direction, D2 is the gradient coefficient along the river bank direction, and D3 is the water level coefficient of the cross-section;

[0036] The object coordinate system is established with the origin F, the foot point of the camera's optical center on the plane to be measured:

[0037] According to the water surface photogrammetry model, it is assumed that the object point P (X, Y, Z), the projection center C (X C ,Y C ,Z C ) and the image point p(x, y) are collinear, o and O are the projection points of the projection center on the image plane and the horizontal plane respectively; for an ideal pinhole imaging system without distortion, the image principal point o is located at the center of the image, and its image plane coordinates (mm) are:

[0038]

[0039] Where s (mm) is the pixel size of the image sensor, m × n (pixel) is the image resolution, and the image plane coordinates (x, y) can be expressed using the corresponding image coordinates (i, j) (pixel) as follows:

[0040]

[0041] When the object distance OC is much larger than the image distance oC, the focal length f is approximately equal to the image distance; the image point and the object point satisfy the linear geometric relationship established by similar triangles; if the image plane coordinate system is used as the reference, the collinearity condition equation can be expressed as follows: the elements of the projection matrix (m1~m 12 ) can be replaced by the following coefficients with specific physical meanings:

[0042]

[0043] The above formula describes the plane coordinate transformation relationship from the image point to the object point; where (x o ,y o ) is the image center, f is the camera focal length, ZZ C is the height of the camera relative to the water surface, Y is the y-direction coordinate value of the surface target, and the rotation matrix composed of the coefficients can use the azimuth angle κ, pitch angle ω and roll angle of the camera relative to the horizontal plane express:

[0044]

[0045] According to the corresponding position of the camera and the cross section, X C = 0 and κ = 0, so the rotation vector can be simplified to:

[0046]

[0047] Therefore, only the exterior orientation parameters remain Unknown. In actual measurement, Y C and Z C are the starting distance and elevation of the camera optical center obtained through survey.

[0048] Furthermore, the method for determining the roll angle in step S5 includes:

[0049] In the distortion-free image after distortion correction, two points p1(x1, y1) and p2(x2, y2) on the water edge of the cross section are selected. The slope of the water edge in the image can be manually marked to determine the roll angle.

[0050]

[0051] At this point, the calculation of the roll angle is completed.

[0052] Furthermore, the method for calculating the pitch angle in step S6 includes:

[0053] Calculate the sub-pixel image coordinate y0 of the intersection of the two annotated points p1(x1,y1) and p2(x2,y2) with the image's vertical axis:

[0054]

[0055] According to the cross-sectional terrain data at the camera installation location and the water level Z0, the world coordinate Y0 of the water edge is calculated and substituted into the model to solve the pitch angle ω:

[0056]

[0057] At this point, the three-axis attitude angle measurement is completed.

[0058] This paper assumes a fixed-focus camera with fixed internal parameters, a fixed installation position, and a precisely measured cross-sectional topography of the river channel to be measured. Based on this premise, the camera's internal parameters are first calibrated in the laboratory. Then, on-site, a laser rangefinder or total station is used to accurately measure the camera's positioning parameters, such as the starting point distance and elevation. Finally, based on a water surface photogrammetry model, the transformation relationship between the waterside coordinates and the attitude angle is derived. The camera's azimuth, roll, and pitch angles are then calculated using the straight waterside of the river channel section, achieving calibration of the three-axis orientation parameters.

[0059] Beneficial effects: Compared with the existing technology, the present invention can achieve high-precision on-site calibration of attitude angles without the assistance of sensors. It not only solves the problem that the existing measurement method needs to rely on attitude sensors to achieve measurement calibration, and requires secondary calibration, resulting in unstable calibration accuracy, but the entire process only requires one calibration to complete. The calibration operation is simple and easy to implement, which solves the problems of high system cost and high calibration complexity in the existing measurement method, reduces the system cost, and ensures the stability of the calibration effect. It is suitable for video monitoring systems for river water level, flow rate, floating objects, etc. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 is a flow chart of the method of the present invention;

[0061] Figure 2 This is a schematic diagram of the camera installation location;

[0062] Figure 3 This is a schematic diagram of camera installation in a specific implementation plan;

[0063] Figure 4 is the original image captured in the specific implementation scheme;

[0064] Figure 5 is a distortion-corrected image in a specific implementation scheme. DETAILED DESCRIPTION

[0065] The present invention is further illustrated below with reference to the accompanying drawings and specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and are not used to limit the scope of the present invention. After reading the present invention, modifications of various equivalent forms of the present invention made by those skilled in the art all fall within the scope defined by the claims attached to this application.

[0066] The present invention provides a camera attitude angle calibration method using the water edge line of the river section. Figure 1 As shown, it includes the following steps:

[0067] S1: Perform intrinsic calibration of the camera in the laboratory using the Zhang Zhengyou calibration method. After calibration, the pixel size s, intrinsic parameter matrix K, and distortion parameter matrix D are obtained as follows;

[0068]

[0069] D=[k1 k2 p1 p2] (2)

[0070] Among them, C x Represents the horizontal coordinate of the principal point, C y represents the ordinate of the principal point, fx represents the equivalent focal length of the camera on the x-axis of the image plane, fy represents the equivalent focal length of the camera on the y-axis of the image plane, and the focal length f of the camera is calculated based on the pixel size s of the image sensor:

[0071] f=(f x +f y )·s / 2 (3)

[0072] k1 represents the first radial distortion coefficient, p1 represents the first tangential distortion coefficient, k2 represents the second radial distortion coefficient, and p2 represents the second tangential distortion coefficient.

[0073] S2: Perform distortion correction on the original image:

[0074] The distortion parameters are used for nonlinear distortion correction of the image, where (x′, y′) and (x, y) are the distorted and undistorted camera coordinates, respectively.

[0075] The image is distorted using the following formula:

[0076]

[0077] The camera coordinates (x, y) of the undistorted image are converted to image coordinates (u, v) using the following formula:

[0078]

[0079] The camera coordinates (x′, y′) of the distorted image are converted to image coordinates (u′, v′) in the same way:

[0080]

[0081] At this point, the image distortion correction is completed.

[0082] S3: If Figure 2 As shown, set the measuring section marking points on both sides of the section to be measured and set the camera on the extension line of the two points. Adjust the azimuth angle so that the marking points in the distortion correction image coincide with the longitudinal axis, and set the azimuth angle to zero. The equipment on-site installation is shown in Figure 3 shown.

[0083] S4: As Figure 2 As shown, adjust the pitch angle so that the field of view covers the water surface area to be measured including the water edge. Observe the horizontality of the water edge to level the roll angle, survey the starting distance and elevation of the camera's optical center, and establish a water surface photogrammetry model:

[0084] The process of establishing the water surface photogrammetry model is as follows:

[0085] Establish a central projection model, which can be expressed as:

[0086]

[0087]

[0088] Where m1~m 12 is the element of the projection matrix and has no practical meaning; (X, Y, Z) is the three-dimensional space coordinate of the object point, is the pixel coordinate of the image point;

[0089] By introducing the central projection model into the water surface photogrammetry model, the object-image collinearity relationship of the model can be expressed as:

[0090]

[0091]

[0092] This formula uses the gradient coefficient to express the water surface elevation Z, D1 is the gradient coefficient of the river section in the cross-section direction, D2 is the gradient coefficient along the river bank direction, and D3 is the water level coefficient of the cross-section;

[0093] The object coordinate system is established with the origin F, the foot point of the camera's optical center on the plane to be measured:

[0094] According to the water surface photogrammetry model, it is assumed that the object point P (X, Y, Z), the projection center C (X C ,Y C ,Z C ) and the image point p(x, y) are collinear, o and O are the projection points of the projection center on the image plane and the horizontal plane respectively; for an ideal pinhole imaging system without distortion, the image principal point o is located at the center of the image, and its image plane coordinates (mm) are:

[0095]

[0096] Where s (mm) is the pixel size of the image sensor, m × n (pixel) is the image resolution, and the image plane coordinates (x, y) can be expressed using the corresponding image coordinates (i, j) (pixel) as follows:

[0097]

[0098] When the object distance OC is much larger than the image distance oC, the focal length f is approximately equal to the image distance; the image point and the object point satisfy the linear geometric relationship established by similar triangles; if the image plane coordinate system is used as the reference, the collinearity condition equation can be expressed as follows: the elements of the projection matrix (m1~m 12 ) can be replaced by the following coefficients with specific physical meanings:

[0099]

[0100] The above formula describes the plane coordinate transformation relationship from the image point to the object point; where (x o ,y o ) is the image center, f is the camera focal length, ZZ C is the height of the camera relative to the water surface, Y is the y-direction coordinate value of the surface target, and the rotation matrix composed of the coefficients can use the azimuth angle κ, pitch angle ω and roll angle of the camera relative to the horizontal plane express:

[0101]

[0102] According to the corresponding position of the camera and the cross section, X C = 0 and κ = 0, so the rotation vector can be simplified to:

[0103]

[0104] Therefore, only the exterior orientation parameters remain Unknown. In actual measurement, Y C and Z C are the starting distance and elevation of the camera optical center obtained through survey.

[0105] S5: Take the original image of the complete river section, such as Figure 3 As shown in , the obtained scene picture is subjected to distortion correction, and then two points p1(x1, y1) and p2(x2, y2) on the water edge of the cross section are selected in the distortion-free image after distortion correction, as shown in Figure 4 As shown, mark the coordinates of the image points on both sides of the straight water edge line. The larger the horizontal distance between the two points, the better, to prevent human errors caused by selecting points too close. The roll angle is determined by manually marking the slope of the water edge line in the image.

[0106]

[0107] At this point, the roll angle is completed Calculation.

[0108] S6: Substitute the sub-pixel image coordinates of the intersection of the line connecting the two points and the longitudinal axis of the image, as well as the world coordinates of the water edge obtained from the water level and cross-sectional topography, into the water surface photogrammetry model to calculate the pitch angle:

[0109] Calculate the sub-pixel image coordinate y0 of the intersection of the two annotated points p1(x1,y1) and p2(x2,y2) with the image's vertical axis:

[0110]

[0111] According to the cross-sectional terrain data at the camera installation location and the water level Z0, the world coordinate Y0 of the water edge is calculated and substituted into the model to solve the pitch angle ω:

[0112]

[0113] At this point, the three-axis attitude angle measurement is completed.

[0114] Based on the above scheme, in order to verify the effectiveness of the method of the present invention, the above method of the present invention is applied to an example, as follows:

[0115] From the above step S1, the camera intrinsic parameter matrix and distortion parameter matrix are first obtained, as shown in the following table.

[0116] Table 1

[0117] parameter Calibration results m(pixel) 3840 n(pixel) 2160 <![CDATA[f x (pixel)]]> 2876.5066 <![CDATA[f y (pixel)]]> 2884.6309 C 1947.3818 <![CDATA[C y ]]> 1043.7429 <![CDATA[k1]]> -0.4074 <![CDATA[k2]]> 0.0004 <![CDATA[p1]]> 0.0007 <![CDATA[p2]]> -0.0004

[0118] In step S4, the camera is first set up and the original image is captured. Figure 4 As shown, then measure the distance Y between the camera's starting point and C , elevation Z C , and the water level data Z0 when the original cross-section image was taken, as shown in Table 2.

[0119] Table 2

[0120]

[0121]

[0122] Based on the cross-sectional topographic data and the water level value, step S5 calculates the Y0 coordinate of the sub-pixel image coordinate y0 of the intersection of the two marked points and the longitudinal axis of the image in the world coordinate system, and the result is Y0 = 187.61 meters.

[0123] Finally, according to the data in Table 1, the original cross-section image is distorted to obtain a distortion-free image. Then, the straight water edge coordinates p1(x1, y1) and p2(x2, y2) of the cross-section are manually selected from the distortion-free image, as shown in the following example: Figure 5 Then, according to the calibration results in Table 2, substitute them into steps S5 and S6 to obtain the final actual roll angle and pitch angle parameters of the camera as shown in Table 3.

[0124] Table 3

[0125]

[0126] At this point, the camera roll angle and pitch angle calculation are completed.

[0127] Based on the above content, it can be seen that compared with the existing calibration methods, the method of the present invention can achieve high-precision on-site calibration of attitude angles without the assistance of sensors, while the existing calibration methods require secondary calibration through attitude sensors, and the calibration accuracy depends on the accuracy of the attitude sensors themselves.

Claims

1. A camera attitude angle calibration method using the water edge of a river section, characterized in that: The steps include: S1: Complete the camera's intrinsic calibration and obtain the corresponding intrinsic parameter matrix and distortion parameter matrix; S2: Perform distortion correction on the original image according to the intrinsic parameter matrix and the distortion parameter matrix; S3: Set the measurement section marking points and set up the camera on the extension line of the two points. Adjust the azimuth angle so that the marking points in the distortion correction image coincide with the longitudinal axis, and set the azimuth angle to zero. S4: Adjust the pitch angle so that the field of view covers the water surface area to be measured including the water edge, observe the horizontality of the water edge to level the roll angle, survey the starting distance and elevation of the camera's optical center, and establish a water surface photogrammetry model; S5: Select two points on the water edge of the cross section in the distortion-corrected image, and determine the roll angle by calculating the slope of the line connecting the two points; S6: Substitute the sub-pixel image coordinates of the intersection of the line connecting the two points and the longitudinal axis of the image, as well as the world coordinates of the water edge obtained from the water level and cross-sectional topography, into the water surface photogrammetry model to calculate the pitch angle; The process of establishing the water surface photogrammetry model in step S4 is as follows: Establish a central projection model, which is expressed as: Where m1~m 12 It is the element of the projection matrix and has no practical meaning; (X, Y, Z) is the three-dimensional space coordinate of the object point, is the pixel coordinate of the image point; The central projection model is introduced into the water surface photogrammetry model, and the object-image collinearity relationship of the model is expressed as: This formula uses the gradient coefficient to express the water surface elevation Z, D1 is the gradient coefficient of the river section in the cross-section direction, D2 is the gradient coefficient along the river bank direction, and D3 is the water level coefficient of the cross-section; The object coordinate system is established with the origin F, the foot point of the camera's optical center on the plane to be measured: According to the water surface photogrammetry model, it is assumed that the object point P (X, Y, Z), the projection center C (X C ,Y C ,Z C ) and the image point p(x,y) are collinear, o and O are the projection points of the projection center on the image plane and the horizontal plane respectively; for an ideal pinhole imaging system without distortion, the principal point o is located at the center of the image, and its image plane coordinates are: Where s is the pixel size of the image sensor, m×n (pixel) is the image resolution, and the image plane coordinate (x, y) is expressed as the corresponding image coordinate (i, j) (pixel): When the object distance OC is much larger than the image distance oC, the focal length f is approximately equal to the image distance; the image point and the object point satisfy the linear geometric relationship established by similar triangles; if the image plane coordinate system is used as the reference, the collinearity condition equation is expressed in the following form, the elements of the projection matrix (m1~m 12 ) is replaced by the following coefficients with specific physical meanings: The above formula describes the plane coordinate transformation relationship from the image point to the object point; where (x o ,y o ) is the image center, f is the camera focal length, ZZ C is the height of the camera relative to the water surface, Y is the y-direction coordinate value of the surface target, and the rotation matrix composed of the coefficients uses the azimuth angle κ, pitch angle ω and roll angle of the camera relative to the horizontal plane express: According to the corresponding position of the camera and the cross section, X C = 0 and κ = 0, so the rotation matrix is simplified to: Y C and Z C are the starting point distance and elevation of the camera optical center obtained through survey respectively; The method for calculating the pitch angle in step S6 includes: Calculate the sub-pixel image coordinate y0 of the intersection of the two annotated points p1(x1,y1) and p2(x2,y2) with the image's vertical axis: According to the cross-sectional terrain data at the camera installation location and the water level Z0, the world coordinate Y0 of the water edge is calculated and substituted into the model to solve the pitch angle ω: Complete three-axis attitude angle measurement.

2. The camera attitude angle calibration method using the water edge of a river section according to claim 1 is characterized in that: In step S1, the camera is calibrated with internal parameters in the laboratory according to the Zhang Zhengyou calibration method. After calibration, the pixel size s, the internal parameter matrix K and the distortion parameter matrix D are obtained, which are as follows: D=[k1 k2 p1 p2] (2) Among them, C x Represents the horizontal coordinate of the principal point, C y represents the vertical coordinate of the principal point, f x Indicates the equivalent focal length of the camera on the x-axis of the image plane, f y It represents the equivalent focal length of the camera on the y-axis of the image plane. The focal length f of the camera is calculated based on the pixel size s of the image sensor: f=(f x +f y )·s / 2 (3) k1 represents the first radial distortion coefficient, p1 represents the first tangential distortion coefficient, k2 represents the second radial distortion coefficient, and p2 represents the second tangential distortion coefficient.

3. The camera attitude angle calibration method using the water edge of a river section according to claim 2 is characterized in that: In step S2, the image is corrected for distortion using the following formula: The camera coordinates (x, y) of the undistorted image are converted to image coordinates (u, v) using the following formula: The camera coordinates (x′, y′) of the distorted image are converted to image coordinates (u′, v′) in the same way: Complete the image distortion correction.

4. The camera attitude angle calibration method using the water edge of a river section according to claim 1 is characterized in that: The method for determining the roll angle in step S5 includes: In the distortion-free image after distortion correction, two points p1(x1, y1) and p2(x2, y2) on the water edge of the cross section are selected, that is, the slope of the water edge in the image is manually marked to determine the roll angle. Complete the calculation of the roll angle.

Images