Dynamic refraction visual correction method for very shallow water blue-green laser sounding

Abstract

The application discloses a method for correcting the vision of laser refraction in extremely shallow water, and belongs to the technical field of marine surveying and mapping and underwater detection, and is used for laser refraction correction. The method comprises the following steps: synchronously collecting binocular images and inertial measurement unit attitude data, and generating three-dimensional point clouds; indexing a light spot area in the point clouds according to a laser emission angle, fitting a micro tangent plane, and obtaining an original normal vector of the water surface; calculating a confidence weight according to the image texture definition, and iteratively calculating an optimal estimated normal vector based on the original normal vector and the weight; calculating a real incident angle according to the optimal estimated normal vector and a laser emission direction; calculating a direction vector of underwater refracted light according to Snell's law; and combining the propagation distances of the laser in the air and in the water and the emission origin coordinates, and calculating three-dimensional coordinates of a target point on the water bottom. The application introduces visual confidence, solves the problem of data failure of traditional visual depth measurement when encountering water surface reflection, glare or broken waves, and significantly improves the robustness under complex sea conditions.

Description

Technical Field

[0001] This invention discloses a dynamic refraction visual correction method for blue-green laser depth sounding in extremely shallow water, belonging to the field of marine surveying and underwater exploration technology. Background Technology

[0002] When conducting underwater topographic mapping in extremely shallow waters (depth < 2m) such as inland waterways and around islands and reefs, the water-air interface exhibits complex dynamic fluctuations due to the influence of wind and waves. Traditional shipborne laser depth sounding systems typically suffer from the following technical drawbacks:

[0003] Severe geometric distortion exists in many existing technologies, which often assume a horizontal water surface (with the normal vector always pointing vertically upward) or compensate for wave height in the vertical direction using only laser displacement sensors. However, the slope of the waves alters the laser's angle of incidence, and according to Snell's law, the direction of the refracted light ray deviates drastically. At a water depth of 1 meter, a 10° water surface tilt can cause the bottom spot position to shift by more than 15 centimeters, severely affecting the planar accuracy of the depth measurement point. Hardware costs are also high; to maintain vertical laser incidence, traditional equipment often requires expensive mechanical stabilization gimbals, and the mechanical response speed struggles to keep up with the vibrations of high-frequency broken waves. In extremely shallow water (0.2 to 1 meter), the echo signal is severely superimposed with surface clutter, and wave-induced refraction errors account for a large proportion in shallow water, making high-precision mapping difficult in this area. Therefore, a method is urgently needed that can capture micro-wave slopes in real time and use software algorithms to replace mechanical gimbals for high-precision refraction correction. Summary of the Invention

[0004] The purpose of this invention is to provide a dynamic refraction visual correction method for blue-green laser depth sounding in extremely shallow water, so as to solve the problems of insufficient accuracy and water surface ripple interference in traditional depth sounding techniques in extremely shallow water.

[0005] A dynamic refraction visual correction method for blue-green laser depth sounding in extremely shallow water includes:

[0006] S1. The unmanned surface vessel (USV) is equipped with a blue-green laser depth sounder, a binocular vision sensor, a synchronization control unit, an inertial measurement unit, and a global navigation satellite system. During the USV's measurement process, the synchronization control unit generates a unified time reference. At the initial moment when the blue-green laser depth sounder emits laser light, the binocular vision sensor performs exposure acquisition, the inertial measurement unit records the attitude angle of the blue-green laser depth sounder, and the global navigation satellite system measures the current coordinate information of the USV to obtain a binocular image sequence.

[0007] S2. Perform stereo matching on the binocular image sequence to generate a disparity map, convert the disparity map into a three-dimensional point cloud, set a calibration matrix, use the calibration matrix and the laser emission angle of the blue-green laser depth sounder to set a preset radius of Euclidean distance and index the laser spot area in the three-dimensional point cloud, use the least squares method to fit the instantaneous tilted spatial plane of the area corresponding to the center of the laser spot, and obtain the original observation normal vector of the water surface.

[0008] S3. Set the image quality threshold, combine the image texture clarity score of the laser landing point area, calculate the adaptive confidence weight, and calculate the optimal estimated normal vector based on the original observation normal vector of the water surface and the adaptive confidence weight using a recursive formula.

[0009] S4. Calculate the corrected true laser incident angle using the refractive index of the laser in air, the refractive index of the laser in water, and the unit vector of the laser's exit direction in air. Based on the corrected true laser incident angle, calculate the direction vector of the underwater refracted light using Snell's law.

[0010] S5. Based on the direction vector of the underwater refracted light, combined with the laser propagation distance in air, the laser propagation distance in water, and the coordinates of the laser emission origin, calculate the three-dimensional coordinates of the target point underwater.

[0011] S1 includes attitude angle Including roll angle Pitch angle and yaw angle ;

[0012] By using a synchronization control unit to unify the time reference of the left and right eye image frames of the binocular vision sensor, a left eye image frame aligned with the time reference is generated. And the right eye image frame aligned with the time reference ,Will and Merged into a stereo image sequence , .

[0013] S2 includes, S2.1, and... Perform polar correction to make and The corresponding pixels are on the same horizontal line, and a semi-global matching algorithm is used to calculate... and disparity value of corresponding pixel Generate a disparity map by summing the disparity values. The horizontal coordinates of the image, The vertical coordinates of the image.

[0014] S2 includes S2.2, a reprojection matrix based on a binocular vision sensor. Each pixel in the disparity map is mapped to a 3D coordinate in the binocular vision sensor coordinate system to generate a 3D point cloud:

[0015] ;

[0016] In the formula, The x-axis coordinate in the binocular vision sensor coordinate system. The y-axis coordinate in the binocular vision sensor coordinate system. The z-axis coordinate in the binocular vision sensor coordinate system. These are the homogeneous components in the binocular vision sensor coordinate system. It is the transpose symbol;

[0017] A calibration matrix is ​​set up, and the region corresponding to the center of the laser spot is indexed in the three-dimensional point cloud using the calibration matrix and the laser emission angle of the blue-green laser depth sounder. The instantaneous tilted spatial plane of the region corresponding to the center of the laser spot is fitted using the least squares method to obtain the original observation normal vector of the water surface.

[0018] S2 includes S2.3, acquiring the joint calibration matrix of the binocular vision sensor and the blue-green laser depth meter. , Includes rotation matrix Translation vector Based on the current emission angle of the laser depth sounder, the ray path of the laser beam under the binocular vision sensor system is calculated, and the preset Euclidean distance radius is set as follows. Searching for points in a 3D point cloud where the Euclidean distance from the center of the laser spot is less than a preset radius. point set As the laser spot area, .

[0019] S2 includes S2.4. The instantaneous tilted spatial plane calculation process includes utilizing... Construct the objective function :

[0020] ,

[0021] In the formula, For the first The x-axis coordinates of the laser spot area points For the first The y-axis coordinates of the laser spot area points For the first The z-axis coordinates of each laser spot area point;

[0022] Solving using the least squares method Minimize parameters ;

[0023] S2 includes S2.5, the fitted plane parameters. Normalization yields the original observed normal vector of the water surface. :

[0024] .

[0025] S3 includes, S3.1, converting the image region corresponding to the laser spot region into a grayscale image. Calculate using the Sobel operator respectively Gradient in the horizontal direction and the gradient in the vertical direction ,calculate The sum of the squared gradient magnitudes of all pixels is used as the original sharpness value. :

[0026] ;

[0027] Will Mapping to interval Obtain image texture sharpness score :

[0028] ;

[0029] In the formula, The preset maximum sharpness threshold is the one with the highest accuracy. The preset minimum resolution threshold is used.

[0030] S3 includes S3.2, calculation Adaptive confidence weights for time-lapse images :

[0031] ;

[0032] In the formula, The base of the exponential function, To preset the image quality threshold, For sensitivity systems.

[0033] S3 includes S3.3, which calculates the optimal estimated normal vector using a recursive formula. :

[0034] ;

[0035] ;

[0036] In the formula, for The optimal estimated normal vector at time t. The sampling time interval, This is an estimate of the angular velocity of the wave surface.

[0037] S4 includes, S4.1, based on Calculate the corrected true laser incident angle ;

[0038] ;

[0039] In the formula, It is an inverse cosine function. Let be the unit vector of the laser's exit direction in air. It is the absolute value;

[0040] S4 includes S4.2, which calculates the direction vector of the underwater refracted light rays. :

[0041] ;

[0042] In the formula, Let be the refractive index of the laser in air. Let be the refractive index of the laser in water.

[0043] S5 includes calculating the three-dimensional coordinates of the target point underwater. :

[0044] ;

[0045] In the formula, The coordinates of the laser emission origin are: This is the distance the laser travels in the air. This represents the distance the laser travels in water. The coordinates of the laser emission center in the geodetic coordinate system at the moment of laser emission are given.

[0046] Compared with existing technologies, this invention has the following advantages: By introducing visual confidence, this invention solves the problem of data failure in traditional visual depth measurement when encountering water surface reflection, glare, or breaking waves, significantly improving robustness in complex sea conditions; by reconstructing the micro-wave morphology corresponding to each laser pulse in real time, it corrects the tens of centimeter-level planar position error caused by wave tilt, achieving high-precision measurement without blind spots in extremely shallow waters; without relying on expensive mechanically stabilized gimbals, stable underwater topographic data can be obtained even when the unmanned vessel is sailing at high speed or swaying violently through algorithm-level spatiotemporal filtering and refraction correction, filling the operational blind spots of conventional multibeam sonar and traditional lidar. Attached Figure Description

[0047] Figure 1 This is a schematic diagram of the overall process of the method of the present invention. Detailed Implementation

[0048] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention are described clearly and completely below. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0049] A dynamic refraction visual correction method for blue-green laser depth sounding in extremely shallow water includes:

[0050] S1. The unmanned surface vessel (USV) is equipped with a blue-green laser depth sounder, a binocular vision sensor, a synchronization control unit, an inertial measurement unit, and a global navigation satellite system. During the USV's measurement process, the synchronization control unit generates a unified time reference. At the initial moment when the blue-green laser depth sounder emits laser light, the binocular vision sensor performs exposure acquisition, the inertial measurement unit records the attitude angle of the blue-green laser depth sounder, and the global navigation satellite system measures the current coordinate information of the USV to obtain a binocular image sequence.

[0051] S2. Perform stereo matching on the binocular image sequence to generate a disparity map, convert the disparity map into a three-dimensional point cloud, set a calibration matrix, use the calibration matrix and the laser emission angle of the blue-green laser depth sounder to set a preset radius of Euclidean distance and index the laser spot area in the three-dimensional point cloud, use the least squares method to fit the instantaneous tilted spatial plane of the area corresponding to the center of the laser spot, and obtain the original observation normal vector of the water surface.

[0052] S3. Set the image quality threshold, combine the image texture clarity score of the laser landing point area, calculate the adaptive confidence weight, and calculate the optimal estimated normal vector based on the original observation normal vector of the water surface and the adaptive confidence weight using a recursive formula.

[0053] S4. Calculate the corrected true laser incident angle using the refractive index of the laser in air, the refractive index of the laser in water, and the unit vector of the laser's exit direction in air. Based on the corrected true laser incident angle, calculate the direction vector of the underwater refracted light using Snell's law.

[0054] S5. Based on the direction vector of the underwater refracted light, combined with the laser propagation distance in air, the laser propagation distance in water, and the coordinates of the laser emission origin, calculate the three-dimensional coordinates of the target point underwater.

[0055] S1 includes attitude angle Including roll angle Pitch angle and yaw angle ;

[0056] By using a synchronization control unit to unify the time reference of the left and right eye image frames of the binocular vision sensor, a left eye image frame aligned with the time reference is generated. And the right eye image frame aligned with the time reference ,Will and Merged into a stereo image sequence , .

[0057] S2 includes, S2.1, and... Perform polar correction to make and The corresponding pixels are on the same horizontal line, and a semi-global matching algorithm is used to calculate... and disparity value of corresponding pixel Generate a disparity map by summing the disparity values. The horizontal coordinates of the image, The vertical coordinates of the image.

[0058] S2 includes S2.2, a reprojection matrix based on a binocular vision sensor. Each pixel in the disparity map is mapped to a 3D coordinate in the binocular vision sensor coordinate system to generate a 3D point cloud:

[0059] ;

[0060] In the formula, The x-axis coordinate in the binocular vision sensor coordinate system. The y-axis coordinate in the binocular vision sensor coordinate system. The z-axis coordinate in the binocular vision sensor coordinate system. These are the homogeneous components in the binocular vision sensor coordinate system. It is the transpose symbol;

[0061] A calibration matrix is ​​set up, and the region corresponding to the center of the laser spot is indexed in the three-dimensional point cloud using the calibration matrix and the laser emission angle of the blue-green laser depth sounder. The instantaneous tilted spatial plane of the region corresponding to the center of the laser spot is fitted using the least squares method to obtain the original observation normal vector of the water surface.

[0062] S2 includes S2.3, acquiring the joint calibration matrix of the binocular vision sensor and the blue-green laser depth meter. , Includes rotation matrix Translation vector Based on the current emission angle of the laser depth sounder, the ray path of the laser beam under the binocular vision sensor system is calculated, and the preset Euclidean distance radius is set as follows. Searching for points in a 3D point cloud where the Euclidean distance from the center of the laser spot is less than a preset radius. point set As the laser spot area, .

[0063] S2 includes S2.4. The instantaneous tilted spatial plane calculation process includes utilizing... Construct the objective function :

[0064] ,

[0065] In the formula, For the first The x-axis coordinates of the laser spot area points For the first The y-axis coordinates of the laser spot area points For the first The z-axis coordinates of each laser spot area point;

[0066] Solving using the least squares method Minimize parameters ;

[0067] S2 includes S2.5, the fitted plane parameters. Normalization yields the original observed normal vector of the water surface. :

[0068] .

[0069] S3 includes, S3.1, converting the image region corresponding to the laser spot region into a grayscale image. Calculate using the Sobel operator respectively Gradient in the horizontal direction and the gradient in the vertical direction ,calculate The sum of the squared gradient magnitudes of all pixels is used as the original sharpness value. :

[0070] ;

[0071] Will Mapping to interval Obtain image texture sharpness score :

[0072] ;

[0073] In the formula, The preset maximum sharpness threshold is the one with the highest accuracy. The preset minimum resolution threshold is used.

[0074] S3 includes S3.2, calculation Adaptive confidence weights for time-lapse images :

[0075] ;

[0076] In the formula, The base of the exponential function, To preset the image quality threshold, For sensitivity systems.

[0077] S3 includes S3.3, which calculates the optimal estimated normal vector using a recursive formula. :

[0078] ;

[0079] ;

[0080] In the formula, for The optimal estimated normal vector at time t. The sampling time interval, This is an estimate of the angular velocity of the wave surface.

[0081] S4 includes, S4.1, based on Calculate the corrected true laser incident angle ;

[0082] ;

[0083] In the formula, It is an inverse cosine function. Let be the unit vector of the laser's exit direction in air. It is the absolute value;

[0084] S4 includes S4.2, which calculates the direction vector of the underwater refracted light rays. :

[0085] ;

[0086] In the formula, Let be the refractive index of the laser in air. Let be the refractive index of the laser in water.

[0087] S5 includes calculating the three-dimensional coordinates of the target point underwater. :

[0088] ;

[0089] In the formula, The coordinates of the laser emission origin are: This is the distance the laser travels in the air. This represents the distance the laser travels in water. The coordinates of the laser emission center in the geodetic coordinate system at the moment of laser emission are given.

[0090] Each frame of the binocular vision sensor contains an exposure timestamp for that moment, which is used for index matching with the laser emission time.

[0091] The instantaneously tilted spatial plane refers to a plane with the center point of the laser spot indexed in the 3D point cloud as the origin and a preset neighborhood radius ( Extract a local point cloud set and fit it using the least squares method to obtain a spatial plane that can characterize the instantaneous tilt state of the local water surface.

[0092] The laser spot area is obtained by setting a preset Euclidean distance radius. Searching for points in a 3D point cloud where the Euclidean distance from the center of the laser spot is less than a preset radius. (like The set of points is used as the laser spot region (ROI).

[0093] and The preset maximum / minimum clarity thresholds (corresponding to clear, calm water and highly turbid water).

[0094] The three-dimensional coordinates of the laser emission center in the geodetic coordinate system at the moment of laser emission are given. At the instant of laser emission, the GNSS / IMU observations are converted to the three-dimensional coordinates of the laser emission center in the geodetic coordinate system using the pre-measured lever vector.

[0095] The following description, in conjunction with the accompanying drawings and embodiments, provides further details. Figure 1 As shown, firstly, nanosecond-level hard-triggered synchronous acquisition of multi-source sensor data is performed, then local 3D reconstruction of the water surface is carried out based on binocular vision, the laser impact point area is extracted and the normal vector of the water surface micro-cut plane is calculated; adaptive filtering of the water surface normal vector based on visual confidence and dynamic correction of the laser water entry refraction path based on the water surface normal vector are performed, and finally, high-precision 3D coordinates of the bottom of the water are output.

[0096] An unmanned surface vessel (USV) equipped with a blue-green laser depth sounder, a high-frame-rate binocular vision sensor, an IMU (Inertial Measurement Unit), and GNSS (Global Navigation Satellite System) hovers over the waters to be measured. At the start of the measurement, a unified nanosecond-level clock reference is generated by the synchronization control unit. At the instant the laser depth sounder emits its probe pulse (t_0), a hardware trigger signal synchronously triggers the binocular vision sensor to perform exposure acquisition, while simultaneously recording the carrier attitude angle of the IMU. This mechanism ensures that the time synchronization error between the visual image, laser emission, and attitude data is less than 1 millisecond, fundamentally eliminating spatiotemporal misalignment errors caused by the vessel's motion.

[0097] The forward-looking scanning operation of an unmanned surface vessel was simulated in an experimental water tank, and a typical moment affected by "breaking wave interference" was selected. The invention performs a full-process calculation of single-point data to demonstrate how it avoids incorrect refraction correction and ensures the accuracy of underwater coordinates when laser is incident at a large angle and there is visual noise on the water surface.

[0098] exist At any given time, the system collects raw data from multiple sensors, including laser depth sounder data (forward scan): the scanning mode is with the laser tilted forward. Launch, simulating forward-looking scan; launch direction vector is The one-way distance corresponding to the total optical path measured during flight time is approximately... (Among which the air section is approximately The middle section of the water ).

[0099] Visual observation data was interfered with: the environmental condition was that a cluster of white foam caused by ship disturbance existed precisely at the laser's impact point; the original observation normal vector, due to the abnormal texture of the foam, was misled by the binocular vision algorithm, resulting in an incorrect calculation of an extremely large wave angle. (This corresponds to the water surface tilting backward by approximately 15°, which will lead to a significant error in refraction calculation); Image texture quality score, (below the set threshold) This indicates low visual credibility.

[0100] Historical state data: The optimal normal vector estimate at the previous time step is... The actual physical situation is that at that moment, the water surface has only slight undulations and is close to a horizontal plane.

[0101] set up calculate :

[0102] ;

[0103] The value of 0.12 indicates that the algorithm accurately identified the "bubble" interference and decided to adopt only 12% of the current visual measurement value, while relying on historical inertia (the best estimate at the previous moment) for 88%, thus effectively suppressing noise.

[0104] Will Substitute calculation :

[0105] ;

[0106] The corrected normal vector corresponds to an inclination angle of only about Successfully filtered out the original observations The artificial wave noise restores the true, calm water surface.

[0107] Known air refractive index water refractive index ;

[0108] First, the direct visual correction method is used to calculate the affected visual observations. (tilt )and Incident ) Calculation, the sum of the angles (incident angles) between the two is as high as Calculate the laser incident angle :

[0109] ;

[0110] The direction vector of the refracted light rays underwater was calculated. ;

[0111] Calculate the Y-coordinate of the seabed :

[0112] .

[0113] Calculations are performed using the method of this invention, and the modified version is used. (Approximate level) and Incident Calculations show that the true angle of incidence is approximately... The refraction vector is calculated by substituting it into the vector formula. Its light continues to propagate diagonally forward, which is consistent with physical facts; calculate the angle of incidence. :

[0114] ;

[0115] The direction vector of the refracted light rays underwater was calculated. ;

[0116] Calculate the Y-coordinate of the seabed :

[0117] .

[0118] Set underwater propagation distance Calculate the displacement of the underwater light spot in the horizontal direction (Y-axis). (Relative to the point of entry) First, calculate the true value of the calm water surface theoretical calculation. :

[0119] ;

[0120] The values ​​calculated using the direct visual correction method introduce errors. Angle of incidence, drastic change in the direction of the ray after refraction, calculated The error is ;

[0121] The value calculated using the method of this invention is based on the corrected normal vector. The error is .

[0122] Experimental data clearly show that oblique scanning is performed by unmanned surface vessels (USVs). During incident (incident) operations, the impact of local noise on the refraction path (such as foam) is geometrically amplified. Existing direct vision correction methods, when encountering such interference, misjudge the water surface normal and refraction angle, leading to an approximately [missing information - likely a missing word or phrase] error in the calculated underwater position. The horizontal displacement deviation is unacceptable in precise underwater topographic mapping. This invention introduces an "adaptive filtering mechanism based on image quality evaluation," which automatically reduces the weight of visual observations when they fail, resulting in a stable normal vector. The error was successfully converged to The entire single-point calculation process strongly demonstrates that the present invention has significant robustness and sub-centimeter-level high precision advantages compared with existing methods in complex dynamic water surface and oblique scanning scenarios.

[0123] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A dynamic refraction visual correction method for blue-green laser depth sounding in extremely shallow water, characterized in that, include: S1. The unmanned surface vessel (USV) is equipped with a blue-green laser depth sounder, a binocular vision sensor, a synchronization control unit, an inertial measurement unit, and a global navigation satellite system. During the USV's measurement process, the synchronization control unit generates a unified time reference. At the initial moment when the blue-green laser depth sounder emits laser light, the binocular vision sensor performs exposure acquisition, the inertial measurement unit records the attitude angle of the blue-green laser depth sounder, and the global navigation satellite system measures the current coordinate information of the USV to obtain a binocular image sequence. S2. Perform stereo matching on the binocular image sequence to generate a disparity map, convert the disparity map into a three-dimensional point cloud, set a calibration matrix, use the calibration matrix and the laser emission angle of the blue-green laser depth sounder to set a preset radius of Euclidean distance and index the laser spot area in the three-dimensional point cloud, use the least squares method to fit the instantaneous tilted spatial plane of the area corresponding to the center of the laser spot, and obtain the original observation normal vector of the water surface. S3. Set the image quality threshold, combine the image texture clarity score of the laser landing point area, calculate the adaptive confidence weight, and calculate the optimal estimated normal vector based on the original observation normal vector of the water surface and the adaptive confidence weight using a recursive formula. S4. Calculate the corrected true laser incident angle using the refractive index of the laser in air, the refractive index of the laser in water, and the unit vector of the laser's exit direction in air. Based on the corrected true laser incident angle, calculate the direction vector of the underwater refracted light using Snell's law. S5. Based on the direction vector of the underwater refracted light, combined with the laser propagation distance in air, the laser propagation distance in water, and the coordinates of the laser emission origin, calculate the three-dimensional coordinates of the target point underwater. S3 includes, S3.1, converting the image region corresponding to the laser spot region into a grayscale image. Calculate using the Sobel operator respectively Gradient in the horizontal direction and the gradient in the vertical direction ,calculate The sum of the squared gradient magnitudes of all pixels is used as the original sharpness value. : ； Will Mapping to interval Obtain image texture sharpness score : ； In the formula, The preset maximum sharpness threshold is the one with the highest accuracy. The preset minimum resolution threshold is used. S3 includes S3.2, calculation Adaptive confidence weights for time-lapse images : ； In the formula, The base of the exponential function, To preset the image quality threshold, Sensitivity system; S3 includes S3.3, which calculates the optimal estimated normal vector using a recursive formula. : ； ； In the formula, for The optimal estimated normal vector at time t. The sampling time interval, This is an estimate of the angular velocity of the wave surface.

2. The dynamic refraction visual correction method for ultra-shallow water blue-green laser depth sounding according to claim 1, characterized in that, S1 includes attitude angle Including roll angle Pitch angle and yaw angle ; By using a synchronization control unit to unify the time reference of the left and right eye image frames of the binocular vision sensor, a left eye image frame aligned with the time reference is generated. And the right eye image frame aligned with the time reference ,Will and Merged into a stereo image sequence , .

3. The dynamic refraction visual correction method for ultra-shallow water blue-green laser depth measurement according to claim 2, characterized in that, S2 includes, S2.1, and... Perform polar correction to make and The corresponding pixels are on the same horizontal line, and a semi-global matching algorithm is used to calculate... and disparity value of corresponding pixel Generate a disparity map by summarizing the disparity values. The horizontal coordinates of the image, The vertical coordinates of the image.

4. The dynamic refraction visual correction method for ultra-shallow water blue-green laser depth measurement according to claim 3, characterized in that, S2 includes S2.2, a reprojection matrix based on a binocular vision sensor. Each pixel in the disparity map is mapped to a 3D coordinate in the binocular vision sensor coordinate system to generate a 3D point cloud: ； In the formula, The x-axis coordinate in the binocular vision sensor coordinate system. The y-axis coordinate in the binocular vision sensor coordinate system. The z-axis coordinate in the binocular vision sensor coordinate system. These are the homogeneous components in the binocular vision sensor coordinate system. It is the transpose symbol; A calibration matrix is ​​set up, and the region corresponding to the center of the laser spot is indexed in the three-dimensional point cloud using the calibration matrix and the laser emission angle of the blue-green laser depth sounder. The instantaneous tilted spatial plane of the region corresponding to the center of the laser spot is fitted using the least squares method to obtain the original observation normal vector of the water surface.

5. The dynamic refraction visual correction method for ultra-shallow water blue-green laser depth sounding according to claim 4, characterized in that, S2 includes S2.3, acquiring the joint calibration matrix of the binocular vision sensor and the blue-green laser depth meter. , Includes rotation matrix Translation vector Based on the current emission angle of the laser depth sounder, the ray path of the laser beam under the binocular vision sensor system is calculated, and the preset Euclidean distance radius is set as follows. Searching for points in a 3D point cloud where the Euclidean distance from the center of the laser spot is less than a preset radius. point set As the laser spot area, .

6. The dynamic refraction visual correction method for ultra-shallow water blue-green laser depth sounding according to claim 5, characterized in that, S2 includes S2.

4. The instantaneous tilted spatial plane calculation process includes utilizing... Construct the objective function : ， In the formula, For the first The x-axis coordinates of the laser spot area points For the first The y-axis coordinates of the laser spot area points For the first The z-axis coordinates of each laser spot area point; Solving using the least squares method Minimize parameters ; S2 includes S2.5, the plane parameters obtained by fitting Normalize to obtain the original observed normal vector of the water surface : 。 7. The dynamic refraction visual correction method for ultra-shallow water blue-green laser depth sounding according to claim 6, characterized in that, S4 includes S4.1, based on calculate the corrected true laser incident angle ; ； In the formula, It is an inverse cosine function. Let be the unit vector of the laser's exit direction in air. It is the absolute value; S4 includes S4.2, which calculates the direction vector of the underwater refracted light rays. : ； In the formula, Let be the refractive index of the laser in air. Let be the refractive index of the laser in water.

8. The dynamic refraction visual correction method for ultra-shallow water blue-green laser depth sounding according to claim 7, characterized in that, S5 includes calculating the three-dimensional coordinates of the target point underwater. : ； In the formula, The coordinates of the laser emission origin are: This is the distance the laser travels in the air. This represents the distance the laser travels in water. The coordinates of the laser emission center in the geodetic coordinate system at the moment of laser emission are given.

Images