An underwater three-dimensional point cloud measurement method, an electronic device, and a storage medium

Abstract

The application discloses a binocular vision underwater three-dimensional point cloud measurement method for sealing a water tank device, electronic equipment and a storage medium, first, a binocular vision underwater optical measurement system is built; a plane calibration plate is used to calibrate the binocular camera system in air, and internal parameters and external parameters of the camera system are obtained; an underwater multi-refraction imaging model is constructed, a method of combining two sets of cameras inside and outside the sealed water tank is adopted, and the spatial position of a refraction plane relative to the camera system inside the sealed water tank is calibrated; the binocular camera system is combined with a guide rail to complete full-circumferential image acquisition of an object surface, and initial images obtained through acquisition are subjected to image enhancement and restoration processing; according to the processed images acquired by the binocular camera, local optical three-dimensional point cloud data are obtained by using a three-dimensional digital image correlation method, and the three-dimensional point cloud is corrected according to the underwater multi-refraction imaging model; the three-dimensional point cloud data obtained at all measurement points of the binocular camera are unified by using coding points, and full-circumferential optical three-dimensional topography measurement of the measured object surface is realized.

Description

Technical Field

[0001] This invention belongs to the field of non-contact underwater high-precision measurement and automated underwater structure monitoring, specifically involving an underwater three-dimensional point cloud measurement method, electronic equipment and storage medium. Background Technology

[0002] Optical-based three-dimensional topography measurement methods include three-dimensional digital image correlation, grating projection, and line structured light methods, which apply the principle of triangulation to measure three-dimensional topography. Due to their advantages such as non-contact measurement, high accuracy, and high speed, optical measurement methods are widely used in scientific and engineering fields.

[0003] The detection of underwater structures mainly relies on experienced divers carrying underwater cameras for manual inspection. However, this method has several drawbacks: (1) Difficulty in observation: The water quality is generally poor, making observation difficult; (2) Inaccuracy in measurement: The assessment of underwater structural defects is subjective due to the lack of engineering experience among divers, and the specific dimensions of the defects cannot be accurately quantified; (3) Incomplete detection: The underwater working environment is unpredictable, leading to missed detections and the inability to accurately determine the morphology and distribution of defects on the underwater structure surface. Currently, the main approach to detecting defects on the surface of underwater structures is to use an effective underwater inspection platform equipped with sonar and optical cameras. Due to the limited scanning resolution, sonar methods can only perform large-scale underwater scanning and cannot effectively obtain information on defects such as cracks and local defects. Furthermore, the accuracy of underwater three-dimensional reconstruction is insufficient. Optical cameras, on the other hand, can construct binocular or multi-view stereo vision to achieve defect detection and three-dimensional reconstruction of the underwater structure surface.

[0004] The main problems faced by underwater structure surface topography measurement technology based on optical measurement are: (1) poor imaging quality in turbid water; (2) cumbersome and complex underwater camera calibration methods; and (3) the refraction of light through different media has a great impact on traditional three-dimensional measurement technology. Proposing reasonable underwater image processing methods, underwater camera calibration methods, and underwater refraction processing methods will promote the development of optical measurement methods in the field of automated underwater structure monitoring. Summary of the Invention

[0005] To address the technical problems mentioned in the background section, this invention proposes an underwater three-dimensional point cloud measurement method, electronic device, and storage medium.

[0006] To achieve the above-mentioned technical objectives, the technical solution of the present invention is as follows:

[0007] A method for underwater three-dimensional point cloud measurement includes the following steps:

[0008] S1. Build an underwater 3D point cloud measurement system, place the binocular camera system in a sealed water tank, and install the sealed water tank on a mechanical guide rail;

[0009] S2. Before the entire device is submerged in water, the camera parameters and refraction parameters of the binocular camera system in the sealed water tank are calibrated, and an underwater imaging multi-refractive correction model is established.

[0010] S3. Based on the size of the underwater object being measured, determine the number of guide rail measuring points required for full-circumference measurement. Using the underwater three-dimensional point cloud measurement system built in step S1, complete the full-circumference surface image acquisition of the object being measured at each measuring point.

[0011] S4. The original image acquired in step S3 is processed using an image restoration and enhancement algorithm;

[0012] S5. First, the images acquired by each binocular camera are restored and enhanced using the method in step S4. Then, based on the camera parameters calibrated in step S1, the local initial three-dimensional point cloud data of the surface of the object being measured is calculated using the three-dimensional digital image correlation method.

[0013] S6. Based on the underwater imaging multi-refractive correction model established in step S2, the local three-dimensional point cloud obtained in step S5 is corrected to obtain the corrected three-dimensional point cloud data.

[0014] S7. Using the principle of close-range photogrammetry, the coordinate system of the three-dimensional point cloud data obtained at all measurement points of the binocular camera system in step S6 is unified, and finally the three-dimensional point cloud data of the entire circumference surface of the object under test is obtained.

[0015] Preferably, the optical measurement method used by the underwater three-dimensional point cloud measurement system in step S1 is a three-dimensional topography measurement method based on the principle of binocular vision, including line structured light method, three-dimensional digital speckle correlation method and grating projection method.

[0016] Preferably, the calibration of camera parameters and refraction parameters of the binocular vision system in the sealed water tank in step S2 includes the calibration of the camera's own parameters and the calibration of the refraction parameters of the sealed water tank.

[0017] Specifically, camera parameter calibration refers to calibrating the intrinsic parameter matrix, extrinsic parameter matrix, and distortion coefficients of each binocular camera using a high-precision checkerboard calibration board.

[0018] The calibration of refraction parameters in a sealed water tank specifically refers to the following: The refraction parameters are calibrated using a combination of two cameras, one inside and one outside the sealed water tank. The camera system inside the sealed water tank is designated as the first binocular camera system, and the camera system outside the sealed water tank is designated as the second binocular camera system. The three-dimensional point cloud of the glass plane of the sealed water tank relative to the left camera of the second binocular camera system is reconstructed using the second binocular camera system. The position transformation relationship between the two binocular camera systems is then calibrated. The three-dimensional point cloud measured by the second binocular camera system is transformed into the coordinate system of the first binocular camera system using the position transformation relationship of the left camera of the two binocular camera systems. Finally, the spatial equations of the inner and outer surfaces of the glass plane of the sealed water tank in the coordinate system of the left camera of the first binocular camera system are calibrated.

[0019] Preferably, the plane equations S1 and S2 of the inner and outer surfaces of the glass relative to the optical center C0 of the left camera of the binocular camera system are expressed as follows:

[0020] S1: z1 = b1 + b2x + b3y

[0021] S2: z2=(b1+d)+b2x+b3y

[0022] Where b1, b2, b3 and d are the undetermined coefficients of the plane equations that need to be calculated to determine the two plane equations S1 and S2;

[0023] The transformation relationship between the second binocular camera system and the first binocular camera system is obtained by combining the left camera of the first binocular camera system and the left camera of the second binocular camera system into a binocular system, and using a high-precision checkerboard calibration board for binocular calibration to obtain the spatial transformation matrix RT of the left camera of the second binocular camera system relative to the left camera of the first binocular camera system. The formula is expressed as follows:

[0024]

[0025] Multiplying any point in the reconstructed 3D point cloud from the second binocular camera system by the RT matrix gives the position of that point in the first binocular camera system.

[0026] Preferably, in order to obtain high-quality images for 3D point cloud computing, image restoration and enhancement processing is performed in step S4;

[0027] The formula for image restoration based on the underwater image degradation model is expressed as follows:

[0028] I c (x, y) = J c (x, y)t c (x, y) + B c [1-t c (x, y)]

[0029] Where Ic(x,y) is the image acquired by the camera, Jc(x,y) is the real image signal, tc(x,y) is the water transmittance, and Bc[1-tc(x,y)] is the background scattered light;

[0030] The image enhancement algorithm uses the standard median filtering algorithm, expressed by the following formula:

[0031] f(i,j)=median{f(r,s)lf(r,s)∈N f(i，j)}

[0032] In the formula, N f(i,j) is the solid neighborhood of f(i,j), and f(r,s) is the gray value of the pixel in the solid neighborhood.

[0033] Preferably, step S6 specifically refers to: Let Q(X, Y, Z) be the previous target point in the initial 3D point cloud, and Q′(Xa, Ya, Za) be the point reconstructed based on the air imaging model; n1, n2, and n3 represent the refractive indices of light in air, glass, and water, respectively; C0(x0, y0, z0) be the optical center of the left camera, and C1(x1, y1, z1) be the optical center of the right camera; point P1(A1, B1, C1) be the intersection of the light ray with the air-glass refraction surface (refractive plane S1); and P2(A2, B2, C2) be the intersection of the light ray with the glass-water refraction surface (refractive plane S2). Point N = (e1, e2, e3) is the normal vector of the refraction plane S1. When the refraction planes S1 and S2 are completely parallel, in the refraction correction imaging model, the light ray originates from a point on the object being measured. It undergoes the first refraction at the interface between water and glass, with the incident angle and refraction angle being θ3 and θ2, respectively. A second refraction occurs at the interface between glass and air, with the incident angle and refraction angle being θ2 and θ1, respectively. Finally, the light ray enters the camera and focuses at the camera's focal point to form an image. Lines L1, L1′, L2, L2′, L3, and L3′ represent the light paths in the three media. The equation of line L is expressed as follows:

[0034]

[0035] Where (α1, β1, γ1) is the direction vector of line L1;

[0036] Solving the system of equations L and S, we obtain the coordinates (A1, B1, C1) of point P1, expressed by the following formula:

[0037]

[0038] According to Snell's law of refraction, the relationship between the direction vectors (α1, β1, γ1) and (α2, β2, γ2) of lines L1 and L2 is as follows:

[0039]

[0040] The equation of line L2 is obtained based on the relationship between the direction vectors:

[0041]

[0042] Based on the above method, the coordinates of point P2 are obtained by combining the equation of line L2 and the refraction plane S2. The relationship between the direction vectors of line L2 and line L3 is obtained according to Snell's law of refraction, and finally the equation of line L3 is obtained. Similarly, the equation of line L3′ is obtained by analyzing the optical path of the right camera. The equations of lines L3 and L3′ are combined, and the actual three-dimensional coordinates of the target on the underwater object are obtained by solving the least squares method.

[0043] Preferably, step S7 specifically refers to:

[0044] Using the principle of close-range photogrammetry, coded markers are pasted on the surface of the object being measured;

[0045] Then, use a DSLR camera to take pictures of the coding point from multiple angles and poses to reconstruct the three-dimensional spatial position of the coding marker point relative to the DSLR camera, and use the spatial coordinate system where the coding point is located as the global coordinate system.

[0046] Next, a binocular camera system is used to photograph the coded markers at each measurement point and reconstruct the spatial position of the coded points relative to the binocular system. By utilizing the consistency of the coded point information, the spatial transformation relationship between the binocular system coordinate system and the global coordinate system at all measurement points is calculated.

[0047] Finally, the coordinate system of the three-dimensional point cloud data obtained at all measurement points of the binocular camera system in step S6 is unified to obtain the three-dimensional point cloud data of the entire circumference surface of the object being measured.

[0048] An electronic device includes a memory and a processor, the memory storing a computer program executable by the processor, the processor executing the computer program to implement the above-described underwater three-dimensional point cloud measurement method.

[0049] A storage medium storing a computer program, which, when read and executed, implements the above-described underwater three-dimensional point cloud measurement method.

[0050] The beneficial effects of adopting the above technical solution are as follows:

[0051] Image restoration based on an underwater image degradation model and image enhancement based on median filtering improve the quality of underwater image acquisition. By employing a combination of two cameras inside and outside a sealed tank, the positional relationship of the refractive plane relative to the camera system is calibrated, establishing an underwater imaging multi-refractive correction model. All calibration tasks are completed in air, avoiding underwater camera calibration and improving underwater measurement efficiency. The underwater imaging multi-refractive correction model is used to correct the initial 3D point cloud, overcoming the influence of light refraction through different media on the reconstructed 3D point cloud. A multi-point local 3D point cloud stitching method based on close-range photogrammetry principles achieves high-precision measurement of the 3D point cloud morphology of underwater structures. This invention solves the problems of poor image quality in underwater optical measurement, cumbersome camera calibration, the influence of underwater light refraction, and low precision of multi-point local 3D point cloud stitching methods, further improving the measurement efficiency and quality of underwater optical measurement methods in automated underwater structure monitoring scenarios. Attached Figure Description

[0052] Figure 1 This is a flowchart of the invention method;

[0053] Figure 2 Diagram of a multi-refractive correction model for underwater imaging;

[0054] Figure 3 This is a schematic diagram of the method for calibrating the refractive parameters of a sealed water tank. Detailed Implementation

[0055] The invention will be further described below with reference to specific implementation examples. This invention discloses an underwater three-dimensional point cloud measurement method, wherein the optical measurement method used is a three-dimensional digital image correlation method based on the principle of binocular vision, such as... Figure 1 As shown, it includes the following steps:

[0056] (1) Build an underwater three-dimensional point cloud measurement system, place the binocular camera system in a sealed water tank, and install the sealed water tank on a mechanical guide rail;

[0057] (2) Before the entire device is submerged in water, the camera parameters and refraction parameters of the binocular camera system in the sealed water tank are calibrated, and a system is established as follows: Figure 2 The underwater imaging multi-refractive correction model is shown.

[0058] The camera's own parameters are calibrated using a high-precision checkerboard calibration plate, including the intrinsic parameter matrix, extrinsic parameter matrix, and distortion coefficients.

[0059] The refractive parameters mainly include the plane equations of the inner and outer surfaces of the sealed water tank glass relative to the optical center C0 of the left camera of the binocular system, expressed by the following formula:

[0060] S1: z1 = b1 + b2x + b3y

[0061] S2: z2=(b1+d)+b2x+b3y

[0062] Where b1, b2, b3 and d are the undetermined coefficients of the plane equations that need to be calculated to determine the two plane equations S1 and S2.

[0063] The calibration method for the refractive plane of a sealed water tank is as follows: Figure 3 As shown. The sealed water tank is designated as the first binocular camera system, and the area outside the sealed water tank is designated as the second binocular camera system. The 3D point cloud of the sealed water tank's glass plane relative to the left camera of system 2 is reconstructed using the second binocular camera system. The positional transformation relationship between the two binocular camera systems is then calibrated. The 3D point cloud measured by system 2 is transferred to system 1 using the positional transformation relationship of the left cameras of the two systems. Finally, the spatial equation of the sealed water tank's glass plane in the coordinate system of the left camera of system 1 is calibrated. The transformation relationship of system 2 relative to system 1 is obtained by combining the left cameras of system 1 and system 2 into a single binocular system. A high-precision checkerboard calibration board is used for binocular calibration to obtain the spatial transformation matrix RT of the left camera of system 2 relative to the left camera of system 1. The formula is expressed as follows:

[0064]

[0065] Multiplying any point in the reconstructed 3D point cloud in System 2 by the RT matrix gives the point's position in System 1. The purpose of the underwater imaging multi-refractive correction model is to correct the virtual image of a point on the surface of the object being measured in the water to the actual position of that point on the surface of the object.

[0066] Let Q(X, Y, Z) be the previous target point in the initial 3D point cloud, and Q′(Xa, Ya, Za) be the point reconstructed based on the air imaging model; nl, n2, and n3 represent the refractive indices of light in air, glass, and water, respectively; C0(x0, y0, z0) is the optical center of the left camera, and C1(x1, y1, z1) is the optical center of the right camera; point P1(A1, B1, C1) is the intersection of the ray with the air-glass refraction surface (refractive plane S1); P2(A2, B2, C2) is the intersection of the ray with the glass-water refraction surface (refractive plane S2); N = (e (e1, e2, e3) are the normal vectors of the refraction plane S1. When the refraction planes S1 and S2 are completely parallel, in the refraction correction imaging model, the light ray originates from a point on the object being measured. It undergoes the first refraction at the interface between water and glass, with the incident angle and refraction angle being θ3 and θ2, respectively. A second refraction occurs at the interface between glass and air, with the incident angle and refraction angle being θ2 and θl, respectively. Finally, the light ray enters the camera and is focused at the camera's focal point to form an image. Lines L1, L1′, L2, L2′, L3, and L3′ represent the light paths in the three media. The equation of line L is expressed as follows:

[0067]

[0068] Where (α1, β1, γ1) is the direction vector of line L1;

[0069] Solving the system of equations L and S, we obtain the coordinates (A1, B1, C1) of point P1, expressed by the following formula:

[0070]

[0071] According to Snell's law of refraction, the relationship between the direction vectors (α1, β1, γ1) and (α2, β2, γ2) of lines L1 and L2 is as follows:

[0072]

[0073] The equation of line L2 is obtained based on the relationship between the direction vectors:

[0074]

[0075] Based on the above method, the coordinates of point P2 are obtained by combining the equation of line L2 and the refraction plane S2. The relationship between the direction vectors of line L2 and line L3 is obtained according to Snell's law of refraction, and finally the equation of line L3 is obtained. Similarly, the equation of line L3′ is obtained by analyzing the optical path of the right camera. The equations of lines L3 and L3′ are combined, and the actual three-dimensional coordinates of the target on the underwater object are obtained by solving the least squares method.

[0076] (3) Based on the size of the underwater object being measured, determine the number of guide rail measuring points required for full-circumference measurement. By combining the binocular system with the guide rail, complete the full-circumference surface image acquisition of the object being measured at each measuring point, and ensure that the binocular vision system has an overlap of 1 / 3 to 1 / 2 between each measuring point to ensure the continuity of the measurement results.

[0077] (4) The original images are processed using image restoration and enhancement algorithms. First, image restoration is performed, and then image enhancement is performed.

[0078] Image restoration processing is based on an underwater image degradation model, expressed by the following formula:

[0079] I c (x, y) = J c (x, y)t c (x, y) + B c [1-t c (x, y)]

[0080] Where Ic(x,y) is the image acquired by the camera, Jc(x,y) is the real image signal, tc(x,y) is the water transmittance, and Bc[1-tc(x,y)] is the background scattered light;

[0081] The image enhancement algorithm uses the standard median filtering algorithm, expressed by the following formula:

[0082] f(i,j)=median{f(r,s)|f(r,s)∈N f(i，j)}

[0083] In the formula, N f(i，j) f(i,j) is the solid neighborhood of f(i,j), and f(r,s) is the gray value of the pixel within the solid neighborhood.

[0084] (5) First, the images acquired by each binocular camera are restored and enhanced using the method in step (4). Then, based on the camera internal and external parameters calibrated in step (1), the local initial three-dimensional point cloud data of the surface of the object being measured is calculated using the three-dimensional digital image correlation method.

[0085] (6) Based on the underwater imaging multi-refractive correction model established in S1, the local three-dimensional point cloud obtained in S5 is corrected point by point to obtain the corrected three-dimensional point cloud data.

[0086] (7) Using the principle of close-range photogrammetry, coded markers are pasted on the surface of the object to be measured. Then, a single-lens reflex camera is used to take pictures of the coded markers from multiple angles and poses to reconstruct the three-dimensional spatial position of the coded markers relative to the single-lens reflex camera. The spatial coordinate system where the coded markers are located is used as the global coordinate system. Then, a binocular camera system is used to take pictures of the coded markers at each measurement point and reconstruct the spatial position of the coded markers relative to the binocular system. Using the consistency of the coded marker information, the spatial transformation relationship between the binocular system coordinate system and the global coordinate system at all measurement points is calculated to unify the coordinate system of the three-dimensional point cloud data obtained at all binocular camera system measurement points in step (6). Finally, the three-dimensional point cloud data of the entire circumference surface of the object to be measured is obtained.

[0087] The present invention also discloses an electronic device, comprising: a memory and a processor, wherein the memory stores a computer program executable by the processor, and the processor executes the computer program to implement the above-described underwater three-dimensional point cloud measurement method. A storage medium is also disclosed, on which a computer program is stored; when the computer program is read and executed, the above-described underwater three-dimensional point cloud measurement method is implemented.

[0088] The embodiments are merely illustrative of the technical concept of the present invention and should not be construed as limiting the scope of protection of the present invention. Any modifications made to the technical solution based on the technical concept proposed in this invention shall fall within the scope of protection of this invention.

Claims

1. A method for underwater three-dimensional point cloud measurement, characterized in that, Includes the following steps: S1. Build an underwater 3D point cloud measurement system, place the binocular camera system in a sealed water tank, and install the sealed water tank on a mechanical guide rail; S2. Before the entire device is submerged in water, the camera parameters and refraction parameters of the binocular camera system in the sealed water tank are calibrated, and an underwater imaging multi-refractive correction model is established. S3. Based on the size of the underwater object being measured, determine the number of guide rail measuring points required for full-circumference measurement, and complete the full-circumference surface image acquisition of the object being measured at each measuring point; S4. The image acquired in step S3 is processed using an image restoration and enhancement algorithm; S5. First, the images acquired by each binocular camera are restored and enhanced using the method in step S4. Then, based on the camera parameters calibrated in step S1, the local initial three-dimensional point cloud data of the surface of the object being measured is calculated using the three-dimensional digital image correlation method. S6. Based on the underwater imaging multi-refractive correction model established in step S2, the local three-dimensional point cloud obtained in step S5 is corrected to obtain the corrected three-dimensional point cloud data. S7. Using the principle of close-range photogrammetry, the coordinate system of the three-dimensional point cloud data obtained at all measurement points of the binocular camera system in step S6 is unified, and finally the three-dimensional point cloud data of the entire circumference surface of the object under test is obtained. Step S6 specifically refers to: Let Q(X, Y, Z) be the previous target point in the initial 3D point cloud, Q ′ (X a Y a Z a Points are reconstructed based on the air imaging model; n1, n2, and n3 represent the refractive indices of light in air, glass, and water, respectively; C0(x0, y0, z0) is the optical center of the left camera, and C1(x1, y1, z1) is the optical center of the right camera; point P1(A1, B1, C1) is the intersection of the ray with the air-glass refraction surface S1; P2(A2, B2, C2) is the intersection of the ray with the glass-water refraction surface S2; N = (e1, e2, e3) is the normal vector of the refraction plane S1. When the refraction planes S1 and S2 are completely parallel, in the refraction correction imaging model, the ray originates from a point on the object being measured, undergoes the first refraction at the water-glass interface, with the incident angle and refraction angle being θ3 and θ2, respectively, and undergoes the second refraction at the glass-air interface, with the incident angle and refraction angle being θ2 and θ1, respectively. Finally, the ray enters the camera, focuses at the camera's focal point, and forms an image. Lines L1 and L2 are also reconstructed. 1′ L2, L 2′ L3 and L 3′ The equation for the straight line L, representing the path of light in the three media, is expressed as follows: Where (α1, β1, γ1) is the direction vector of line L1; Solving the system of equations L and S, we obtain the coordinates (A1, B1, C1) of point P1, expressed by the following formula: According to Snell's law of refraction, the relationship between the direction vectors (α1, β1, γ1) and (α2, β2, γ2) of lines L1 and L2 is as follows: The equation of line L2 is obtained based on the relationship between the direction vectors: By combining the equation of line L2 and the refraction plane S2, we can obtain the coordinates of point P2. According to Snell's law of refraction, we can obtain the relationship between the direction vectors of line L2 and line L3, and finally obtain the equation of line L3. Based on the above method, the optical path of the right camera is analyzed to obtain the equation of line L3′. By combining the equations of lines L3 and L3′, the actual three-dimensional spatial coordinates of the target on the underwater object are obtained by solving the least squares method.

2. The underwater three-dimensional point cloud measurement method according to claim 1, characterized in that, The optical measurement method used in the underwater three-dimensional point cloud measurement system in step S1 is a three-dimensional topography measurement method based on the principle of binocular vision, including the line structured light method, the three-dimensional digital speckle correlation method, and the grating projection method.

3. The underwater three-dimensional point cloud measurement method according to claim 1, characterized in that, In step S2, the binocular vision system in the sealed water tank is calibrated for camera parameters and refraction parameters, including the calibration of the camera's own parameters and the calibration of the refraction parameters of the sealed water tank. Specifically, camera parameter calibration refers to calibrating the intrinsic parameter matrix, extrinsic parameter matrix, and distortion coefficients of each binocular camera using a high-precision checkerboard calibration board. The calibration of refraction parameters in a sealed water tank specifically refers to the following: The refraction parameters are calibrated using a combination of two cameras, one inside and one outside the sealed water tank. The camera system inside the sealed water tank is designated as the first binocular camera system, and the camera system outside the sealed water tank is designated as the second binocular camera system. The three-dimensional point cloud of the glass plane of the sealed water tank relative to the left camera of the second binocular camera system is reconstructed using the second binocular camera system. The position transformation relationship between the two binocular camera systems is then calibrated. The three-dimensional point cloud measured by the second binocular camera system is transformed into the coordinate system of the first binocular camera system using the position transformation relationship of the left camera of the two binocular camera systems. Finally, the spatial equations of the inner and outer surfaces of the glass plane of the sealed water tank in the coordinate system of the left camera of the first binocular camera system are calibrated.

4. The underwater three-dimensional point cloud measurement method according to claim 3, characterized in that, The plane equations S1 and S2 of the inner and outer surfaces of the glass relative to the optical center C0 of the left camera in the binocular camera system are expressed as follows: S1:z1=b1+b2x+b3y S2:z2=(b1+d)+b2x+b3y Where b1, b2, b3 and d are the undetermined coefficients of the plane equations that need to be calculated to determine the two plane equations S1 and S2; The transformation relationship between the second binocular camera system and the first binocular camera system is obtained by combining the left camera of the first binocular camera system and the left camera of the second binocular camera system into a binocular camera system, and using a high-precision checkerboard calibration board for binocular calibration to obtain the spatial transformation matrix RT of the left camera of the second binocular camera system relative to the left camera of the first binocular camera system. The formula is expressed as follows: Multiplying any point in the reconstructed 3D point cloud from the second binocular camera system by the RT matrix gives the position of that point in the first binocular camera system.

5. The underwater three-dimensional point cloud measurement method according to claim 1, characterized in that, To obtain high-quality images for 3D point cloud computing, image restoration and enhancement processing is performed in step S4; The formula for image restoration based on the underwater image degradation model is expressed as follows: I c (x,y)=J c (x,y)t c (x,y)+B c [1-t c (x,y)] Among them, I c (x,y) is the image acquired by the camera, J c (x,y) represents the real image signal, t c (x,y) represents the transmittance of water, plus B c [1-t c [x,y] represents the background scattered light; The image enhancement algorithm uses the standard median filtering algorithm, expressed by the following formula: f(i,j)=median{f(r,s)|f(r,s)∈N f(i,j) } In the formula, N f(i,j) f(i,j) is the solid neighborhood of f(i,j), and f(r,s) is the gray value of the pixel within the solid neighborhood.

6. The underwater three-dimensional point cloud measurement method according to claim 1, characterized in that, Step S7 specifically refers to: Using the principle of close-range photogrammetry, coded markers are pasted on the surface of the object being measured; Then, use a DSLR camera to take pictures of the coding point from multiple angles and poses to reconstruct the three-dimensional spatial position of the coding marker point relative to the DSLR camera, and use the spatial coordinate system where the coding point is located as the global coordinate system. Next, a binocular camera system is used to photograph the coded markers at each measurement point and reconstruct the spatial position of the coded points relative to the binocular system. By utilizing the consistency of the coded point information, the spatial transformation relationship between the binocular system coordinate system and the global coordinate system at all measurement points is calculated. Finally, the coordinate system of the three-dimensional point cloud data obtained at all measurement points of the binocular camera system in step S6 is unified to obtain the three-dimensional point cloud data of the entire circumference surface of the object being measured.

7. An electronic device, characterized in that, include: A memory and a processor, the memory storing a computer program executable by the processor, the processor executing the computer program to implement the underwater three-dimensional point cloud measurement method according to any one of claims 1-6.

8. A storage medium, characterized in that, The storage medium stores a computer program, which, when read and executed, implements the underwater three-dimensional point cloud measurement method according to any one of claims 1-6.

Images