Error correction method for wave refraction correction of oval scanning laser radar

Abstract

The application discloses an egg circle type scanning laser radar wave refraction correction error correction method, belongs to the technical field of airborne laser depth sounding, and is used for wave refraction correction of underwater terrain measurement of an airborne laser radar and comprises the following steps: filtering and processing sea surface point clouds by adopting a statistical outlier removal filtering method; extracting a sea surface slope normal vector corresponding to each point by adopting an optimized moving least square method; simplifying an egg circle type scanning laser radar measurement process to be carried out in a carrier coordinate system; constructing a refraction correction model by using Snell's law and a vector decomposition method; and obtaining corrected sea bottom point cloud coordinates. The application removes noise from sea surface point cloud data, fits the measured sea surface point cloud by using a moving least square method, adopts characteristic decomposition of a covariance matrix instead of polynomial fitting, and thus, the calculation is simple, polynomial fitting errors are avoided, the process is greatly simplified by using vector decomposition calculation, and there is only a single solution.

Description

TECHNICAL FIELD

[0001] The application discloses an egg-shaped scanning laser radar wave refraction correction error correction method and belongs to the technical field of airborne laser depth sounding. BACKGROUND

[0002] Airborne laser depth sounding is one of efficient ocean sounding technologies developed in the past two or three decades, integrates a series of frontier technologies such as laser, automatic control, aviation, global navigation satellite system and computer, uses an airplane as a carrying platform, and is an advanced sounding method in which an airborne sensor transmits a laser beam to detect the water depth below the sensor, has the advantages of high speed, high coverage, strong flexibility and the ability to detect underwater areas inaccessible to ships, and has outstanding advantages in describing the seafloor topographic features of shallow sea areas, establishing an underwater three-dimensional obstacle model and island reef monitoring.

[0003] In recent years, the laser radar wave error correction technology is mostly to extract sea surface information, determine sea surface parameters in combination with a wave spectrum, construct an instantaneous sea surface model by simulating the sea surface to correct errors, commonly used wave spectrum models include the Pierson-Moscowitz spectrum, the Joint North Sea Wave Project spectrum and the Apel spectrum, initial parameters of the wave spectrum are determined by extracting the sea surface information, the energy spectrum is converted into an instantaneous wave height by using a fast Fourier transform technology to establish a simulated sea surface model, this method can correct errors caused by the influence of waves on the laser radar to a certain extent, but lacks the spatial and temporal characteristics of the measured sea surface point cloud.

[0004] Since the data collected by the laser radar during work is actually the state of the scanned object at a specific time point, and the real wave changes all the time, to correct the depth sounding error caused by the wave, another important factor is how to track and correct each laser beam. The egg-shaped scanning laser radar has the characteristics of large scanning range and simple mechanical structure, but due to the unique off-axis rotating mirror structure, the scanning shape presents an irregular "egg-shaped" shape, so that the angle of the laser after reflecting the mirror is not as easy to calculate as the regular shape, at present, the refraction correction of the egg-shaped scanning laser radar is to calculate the refraction light vector by constructing a nonlinear equation set according to the angle relationship and coplanar condition, but the calculation process is complex and multiple solutions may occur. SUMMARY

[0005] The application aims to provide an egg-shaped scanning laser radar wave refraction correction error correction method to solve the problems in the prior art that the correction model lacks the spatial and temporal characteristics of the measured sea surface point cloud, the calculation process is complex and multiple solutions may occur.

[0006] The egg-shaped scanning laser radar wave refraction correction error correction method comprises the following steps:

[0007] S1, the working environment of the airborne laser radar is sea condition less than or equal to three, wind force less than or equal to four, wave height less than or equal to 1.25 m, after obtaining the original sea surface point cloud data, a K-dimensional tree is constructed for the original sea surface point cloud data set, a statistical outlier removal filtering algorithm is used for filtering processing, a density adaptive threshold function is introduced to dynamically correct the filtering processing threshold, and the preprocessed sea surface point cloud data is obtained;

[0008] S2, a K-dimensional tree is constructed for the preprocessed sea surface point cloud data set, a neighborhood radius is set based on the sea surface point cloud density, the neighborhood selection method adopts a fixed radius field, the sea surface point cloud in the neighborhood is weighted least square fitted, a Gaussian kernel function is selected as the weight function, the neighborhood points are centralized, then a covariance matrix is constructed to replace the polynomial fitting process, and feature decomposition is carried out to obtain the sea surface slope information;

[0009] S3, the carrier coordinate system and the auxiliary coordinate system are constructed by changing the angle through the off-axis rotating mirror, the included angle between the laser incidence and the mirror normal vector is calculated, the plane normal vector of the common plane of the laser incidence light, the mirror normal and the reflected light is calculated, the reflected light direction vector equation set is constructed, based on the included angle, the plane normal vector and the equation set, the incident light direction vector for constructing the instantaneous refraction correction model is obtained, the refraction correction model is constructed by using Snell's law and vector decomposition method, and based on the refraction correction model, the corrected sea bottom point cloud coordinates are obtained.

[0010] S1 includes S1.1, the working environment of the airborne laser radar is sea condition less than or equal to three, wind force less than or equal to four, wave height less than or equal to 1.25 m; after obtaining the original sea surface point cloud data, a K-dimensional tree is constructed for the original sea surface point cloud data set;

[0011] S1 includes S1.2, a density adaptive threshold function is introduced on the basis of the statistical outlier removal filtering algorithm, the point density of the surrounding neighborhood points , the basic standard deviation multiplier is dynamically corrected, and the formula is as follows:

[0012] ;

[0013] In the formula, is the corrected dynamic threshold, is the global reference threshold multiplier, is the sensitivity coefficient, is a variable starting from 1 and increasing, and the basic standard deviation multiplier is a multiplier for expanding or adjusting the standard deviation range.

[0014] S2 includes S2.1, constructing K-dimensional tree for the pre-processed sea surface point cloud dataset, accelerating the neighborhood search process, setting the neighborhood radius according to the sea surface point cloud density, and adopting the fixed radius neighborhood for the neighborhood selection method , The neighborhood point set is:

[0015] ;

[0016] In the formula, is the neighborhood point set, is the neighborhood point of is a variable that increases from the starting value 1, is the Euclidean distance, is the neighborhood radius;

[0017] S2 includes S2.2, weighted least squares fitting is performed on the sea surface point cloud in the neighborhood, the fitting weight is inversely proportional to the distance from the target point to the neighborhood point, the weight function is selected as the Gaussian kernel function, and the formula is as follows:

[0018] ;

[0019] In the formula, is the fitting weight, is the natural exponential function, is the Euclidean distance between the point and each point in the neighborhood, is the bandwidth of the Gaussian kernel, and the bandwidth of the Gaussian kernel is a parameter for controlling the smoothness of the Gaussian function;

[0020] S2 includes S2.3, calculating the weighted centroid of the neighborhood point set :

[0021] ;

[0022] In the formula, is the number of neighborhood points;

[0023] S2 includes S2.4, centering the neighborhood points and calculating the offset of the point to the centroid point:

[0024] .

[0025] S2 includes S2.5, constructing a covariance matrix :

[0026] ;

[0027] In the formula, Points to the centroid on the coordinate axis The offset; perform eigenvalue decomposition on the covariance matrix to obtain the eigenvector corresponding to the smallest eigenvalue, the eigenvector being... The corresponding plane normal vector.

[0028] S3 includes S3.1, setting the reflector normal of the off-axis rotating mirror to an angle of 7.5° with the rotation axis, arranging the off-axis rotating mirror at a 45° angle to the horizontal direction, and constructing the carrier coordinate system by incident laser light along the horizontal direction. , The axis is oriented towards the radar's direction of travel. Axially upward, A right-front-upper coordinate system is formed by the axis to the right. Around Rotate the axis counterclockwise by 45° to make Construct an auxiliary coordinate system by aligning the axis with the rotation axis. .

[0029] S3 includes S3.2, determining the laser incident direction vector in the auxiliary coordinate system. and the normal vector of the mirror :

[0030] ;

[0031] ;

[0032] In the formula, This represents the real-time angle of the off-axis rotating mirror;

[0033] Calculate the angle between the incident laser and the normal vector of the mirror. for:

[0034] ;

[0035] In the formula, The length of the vector. This is the transpose symbol.

[0036] S3 includes S3.3, based on the coplanar relationship between the incident laser ray, the mirror normal, and the reflected ray, whereby the normal vector of the coplanar plane is... The formula is as follows:

[0037] ;

[0038] Let the direction vector of the reflected ray be... This yields the following system of equations:

[0039] ;

[0040] S3 includes S3.4, and will around the axis clockwise by 45° to obtain , the incident light direction vector when constructing the instantaneous refraction correction model :

[0041] ;

[0042] wherein, is the angle between the axis, is the angle between the axis, is the angle between the axis, is the angle between the axis, is

[0043] S3 includes S3.5, the instantaneous water surface normal vector of the laser penetration is , the direction vector of the light ray after refraction by the water body is , the cosine value of the incident angle of the light ray before penetrating the water surface is:

[0044] ;

[0045] According to Snell's law and the triangle identity , the sine value of the refraction angle of the light ray after penetrating the water surface is:

[0046] ;

[0047] wherein, is the refractive index of the incident medium, is the refractive index of the refractive medium.

[0048] S3 includes S3.6, according to the vector operation law, decomposes along the normal and tangent of the incident plane, decomposes along the normal and tangent of the incident plane, according to Snell's law, and the parallel components satisfy the relationship , the vertical components are related to the cosine of the refraction angle , and the decomposition formula of the incident light is:

[0049] ;

[0050] ;

[0051] wherein, represents the vertical component, represents the parallel component.

[0052] S3 comprises S3.7, the is modeled as the sum of a vertical component and a parallel component:

[0053] ;

[0054] ;

[0055] After sorting, the refraction vector expression refracted by the water body is obtained:

[0056] ;

[0057] According to the refraction vector expression refracted by the water body, the seabed point cloud coordinates are corrected:

[0058] ;

[0059] In the formula, is the corrected coordinate, is the slant range from the laser emission position to the sea surface point, is the propagation speed of light in air, is the time of light in water.

[0060] Compared with the prior art, the present application has the following beneficial effects: the present application directly processes the sea surface point cloud by using an optimized statistical outlier removal filter, introduces a density adaptive function to retain wave characteristics, can efficiently remove noise points caused by environmental light or scattering, and can retain the space-time characteristics of waves, fits the measured point cloud by using a moving least square method, uses characteristic decomposition of a covariance matrix instead of polynomial fitting, is simple to calculate and avoids polynomial fitting errors, uses vector decomposition calculation to greatly simplify the process, and only has a single solution. BRIEF DESCRIPTION OF DRAWINGS

[0061] Figure 1 is a technical flowchart of the present application;

[0062] Figure 2 is a schematic diagram of the influence of wave refraction on the precision of a laser radar;

[0063] Figure 3 is a sea surface point cloud denoising flowchart;

[0064] Figure 4 is a carrier coordinate system schematic diagram;

[0065] Figure 5 is an instantaneous refraction correction model schematic diagram;

[0066] Figure 6 is a water surface original data point cloud;

[0067] Figure 7 is the filtered sea surface point cloud;

[0068] Figure 8 is the schematic diagram of refraction correction effect. DETAILED DESCRIPTION

[0069] In order to make the objects, technical solutions and advantages of the present application clearer, the technical solutions in the present application are described clearly and completely below. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the protection scope of the present application.

[0070] The oval scanning laser radar wave refraction correction error correction method comprises:

[0071] S1, the working environment of the airborne laser radar is sea state less than or equal to level three, wind force less than or equal to level four, and wave height less than or equal to 1.25 m. After obtaining the original sea surface point cloud data, a K-dimensional tree is constructed for the original sea surface point cloud data set, a statistical outlier removal filtering algorithm is used for filtering processing, a density adaptive threshold function is introduced to dynamically correct the filtering processing threshold, and preprocessed sea surface point cloud data is obtained.

[0072] S2, a K-dimensional tree is constructed for the preprocessed sea surface point cloud data set, a neighborhood radius is set based on the sea surface point cloud density, a fixed radius field is used for neighborhood selection, a weighted least square fitting is performed on the sea surface point cloud in the neighborhood, a Gaussian kernel function is selected for the weight function, the neighborhood points are centralized, then a covariance matrix is constructed to replace the polynomial fitting process, and feature decomposition is performed to obtain sea surface slope information.

[0073] S3, a carrier coordinate system and an auxiliary coordinate system are constructed by changing the angle through the off-axis rotating mirror, the included angle between the laser incidence and the mirror normal vector is calculated, the plane normal vector of the common plane of the laser incidence light, the mirror normal and the reflected light is calculated, the reflected light direction vector equation set is constructed, the incident light direction vector for constructing the instantaneous refraction correction model is obtained based on the included angle, the plane normal vector and the equation set, the refraction correction model is constructed by using the Snell's law and the vector decomposition method, and the corrected sea bottom point cloud coordinates are obtained based on the refraction correction model.

[0074] S1 comprises S1.1, the working environment of the airborne laser radar is sea state less than or equal to level three, wind force less than or equal to level four, and wave height less than or equal to 1.25 m. After obtaining the original sea surface point cloud data, a K-dimensional tree is constructed for the original sea surface point cloud data set.

[0075] S1 comprises S1.2, a density adaptive threshold function is introduced based on the statistical outlier removal filtering algorithm, the density of the point around the neighborhood points is calculated , the base standard deviation multiplier is dynamically revised, and the formula is as follows:

[0076] ;

[0077] In the formula, is the revised dynamic threshold value, is a global reference threshold multiplier, is a sensitivity coefficient, is a variable that starts from 1 and increases, and the base standard deviation multiplier is a multiplier for expanding or adjusting the standard deviation range.

[0078] S2 includes S2.1, constructing a K-dimensional tree for the pre-processed sea surface point cloud data set to accelerate the neighborhood search process, setting the neighborhood radius according to the sea surface point cloud density, and using a fixed radius neighborhood for neighborhood selection. For a point in the sea surface point cloud data set , The neighborhood point set is:

[0079] ;

[0080] In the formula, is the neighborhood point set, is the neighborhood point of the target point, is a variable that starts from 1 and increases, is the Euclidean distance, is the neighborhood radius;

[0081] S2 includes S2.2, weighted least squares fitting of the neighborhood sea surface point cloud, the fitting weight is inversely proportional to the distance from the neighborhood point to the target point, the weight function is selected as a Gaussian kernel function, and the formula is as follows:

[0082] ;

[0083] In the formula, is the fitting weight, is a natural exponential function, is the Euclidean distance between the point and each point in the neighborhood, is the bandwidth of the Gaussian kernel, and the bandwidth of the Gaussian kernel is a parameter for controlling the smoothness of the Gaussian function;

[0084] S2 includes S2.3, calculating the weighted centroid of the neighborhood point set :

[0085] ;

[0086] In the formula, is the number of neighborhood points;​

[0087] S2 includes S2.4, centering neighborhood points, calculating point offset to the centroid point :

[0088] .

[0089] S2 includes S2.5, constructing a covariance matrix :

[0090] ;

[0091] wherein, are offset of point to the centroid point in the coordinate axis ; eigenvalue decomposition is performed on the covariance matrix to obtain a feature vector corresponding to the minimum eigenvalue, and the feature vector is a corresponding plane normal vector.

[0092] S3 includes S3.1, setting the angle between the normal of the deflection mirror and the rotation axis to 7.5°, arranging the deflection mirror at an angle of 45° with the horizontal direction, and constructing a carrier coordinate system with the laser incident along the horizontal direction , the axis is directed towards the forward direction of the radar, the axis is directed upwards, the axis is directed to the right to form a right front upper coordinate system, and is rotated counterclockwise around the axis by 45°, so that the axis coincides with the rotation axis, and an auxiliary coordinate system is constructed .

[0093] S3 includes S3.2, determining the laser incident direction vector and the mirror normal vector :

[0094] ;

[0095] ;

[0096] wherein, is the real-time angle of the deflection mirror;

[0097] the angle between the laser incident direction and the mirror normal vector is calculated as :

[0098] ;

[0099] wherein, is the vector length, is the transpose symbol.

[0100] S3 includes S3.3, based on the coplanar relationship of the laser incident light, the mirror normal and the reflected light, the normal vector of the coplanar plane is , the formula is as follows:

[0101] ;

[0102] Let the direction vector of the reflected light be , and the equation group is obtained:

[0103] ;

[0104] S3 includes S3.4, rotating around the axis clockwise by 45° to obtain , is the incident light direction vector when constructing the instantaneous refraction correction model :

[0105] ;

[0106] In the formula, is the angle between and the axis, is the angle between and the axis, is the angle between and the axis.

[0107] S3 includes S3.5, let the instantaneous water surface normal vector of the laser penetration be , the direction vector of the light after refraction by the water body is , and the cosine value of the incident angle of the light before penetrating the water surface is:

[0108] ;

[0109] According to Snell's law and the triangle identity , the sine value of the refraction angle of the light after penetrating the water surface is:

[0110] ;

[0111] In the formula, is the refractive index of the incident medium, is the refractive index of the refractive medium.

[0112] S3 includes S3.6, according to the vector operation rule, decompose along the normal and tangent of the incident plane, and Decomposed along the normal and tangential of the incident plane, according to Snell's law, and The parallel component satisfies the relationship The vertical component is related to the cosine of the refraction angle The decomposition formula of the incident light is:

[0113] ;

[0114] ;

[0115] In the formula, represents the vertical component, represents the parallel component.

[0116] S3 includes S3.7, modeling as the sum of the vertical component and the parallel component:

[0117] ;

[0118] ;

[0119] After sorting, the refraction vector expression refracted by the water body is obtained:

[0120] ;

[0121] According to the refraction vector expression refracted by the water body, the coordinates of the seabed point cloud are corrected:

[0122] ;

[0123] In the formula, is the corrected coordinate, is the slant range from the laser emission position to the sea surface point, is the propagation speed of light in air, is the time of light in water.

[0124] To solve the technical problems of the present application, two aspects are taken as the starting point. First, analyze the causes of wave error. When the light beam contacts the water-air surface, due to the different refractive indices of light in different media, refraction effect will occur. As shown in Figure 2 , which can be expressed by Snell's law as:

[0125] ;

[0126] Among them , are the refractive indices of the two media before and after the incident, , are the propagation speeds of light in the two media before and after the incident, is the incident angle, is the reflection angle.

[0127] However, due to the actual sea surface under the action of temperature, wind and other natural conditions, complex wave changes will occur, resulting in an angle deviation between the tangential plane of the laser incident on the sea surface and the horizontal plane, thereby causing a deviation of the normal vector of the laser incident surface, and further causing a change in the refraction angle, and this influence increases linearly with water depth.

[0128] Secondly, based on the current development of laser radar equipment gradually developing towards high resolution and high frequency collection, in order to correct the depth error caused by waves, each laser needs to be accurately corrected, that is, by establishing an instantaneous refraction correction model for each sea surface laser foot point in the coordinate system, as shown in Figure 5 , the following technical route is realized, as shown in Figure 1 .

[0129] (1) First of all, due to the single-band laser radar carrying 532nm wavelength laser emitting blue-green laser with strong penetration to water body, and due to the influence of environmental light, aerosol and water body scattering during medium propagation, there are many noise points in the collected sea surface point cloud, which cannot extract the correct sea surface point cloud, so it is necessary to first carry out denoising processing on the sea surface point cloud to extract high-quality sea surface point cloud data set;

[0130] (2) After obtaining the sea surface point cloud data, in order to retain the time and space characteristics of the collected sea surface, the sea surface slope information is directly extracted from the processed sea surface point cloud, so a suitable method is needed to extract accurate sea surface slope information;

[0131] (3) Due to the irregular scanning trajectory of the oval scanning laser radar, the light incidence angle is different from other scanning methods, so a light ray tracing method suitable for oval scanning laser radar is needed to correct each laser.

[0132] Therefore, the present application obtains the following technical scheme:

[0133] (1) Sea surface point cloud preprocessing. Generally speaking, the working environment of airborne laser radar is recommended to be no more than three sea states, i.e. wind force level 3-4, wave height 0.5m-1.25m. Under this working environment, there is some surge on the sea surface, but no large wave change, so the noise point density is much smaller than the sea surface point cloud density, and the overall sea surface point cloud data set conforms to the normal distribution. After obtaining the original sea surface point cloud data set, a K-dimensional tree is constructed to speed up the neighborhood search. The K-dimensional tree is a binary tree data structure for dividing the point set in K-dimensional space with K-dimensional hyperplane. A statistical outlier removal filtering method is used to filter the sea surface point cloud, and a point is searched in the data set, and then k+1 adjacent points are calculated The average distance to neighboring points is used to dynamically adjust the threshold using a density adaptive function. This process assesses wave crest characteristics and removes outliers. Points with a distance greater than the threshold are marked as outliers and filtered out. Points with a distance less than the threshold are output as point cloud coordinates to obtain the processed sea surface point cloud. Figure 3 As shown;

[0134] (2) Sea surface slope information extraction. The key to refraction correction lies in whether the sea surface slope normal vector corresponding to each point can be extracted from the preprocessed sea surface point cloud. Since the sea surface wave is essentially a non-linear changing surface, it is necessary to perform local surface fitting for each sea surface laser point and calculate the normal vector. This invention uses the optimized moving least squares (MLS) method to achieve this function;

[0135] (3) Instantaneous sea surface correction model based on oval scanning lidar. Compared with other scanning methods, oval scanning lidar has the advantages of large scanning range and simple mechanical structure. However, due to its irregular scanning shape, the incident angle of the laser when penetrating the water surface changes nonlinearly. Therefore, in order to determine the incident light direction vector when the laser enters the water surface, the whole process can be simplified to the carrier coordinate system, such as... Figure 4 As shown. Finally, using Snell's law and vector decomposition, a refraction correction model is constructed to calculate the refracted ray direction vector and the corrected seabed point cloud coordinates, as shown. Figure 5 As shown, Let be the instantaneous water surface normal vector at which the laser penetrates. The direction vector of the light rays after refraction through the water body. Let be the incident light direction vector. The angle of incidence of light before it penetrates the water. The angle of refraction of light after it passes through the water surface. Let be the refractive index of the incident medium. Let be the refractive index of the refracting medium. for The vertical component, for Parallel components, for The vertical component, for Parallel components.

[0136] To verify the performance of this invention in correcting wave refraction errors, a single-band airborne lidar developed by the team was used for testing at a comprehensive surveying and mapping test site. The equipment was used to scan the water body and underwater targets in the pool to obtain the original point cloud dataset. The original dataset was then processed based on this invention to verify its performance.

[0137] Self-developed lidar scans of raw point clouds on the sea surface, such as... Figure 6As shown, it is obvious that there are more noise points due to water scattering and environmental light effects, and after processing by the density adaptive statistical outlier removal filtering algorithm, the water surface point cloud image is as shown in Figure 7 As shown, the noise points have been removed, and the data comparison before and after filtering is shown in Table 1:

[0138] Table 1, data comparison before and after filtering

[0139] .

[0140] The wind level on the test day was 1~2, and the elevation change of the water surface point cloud data after filtering met the conditions.

[0141] The comparison of the seabed point cloud shape before and after correction is as shown in Figure 8 As shown, the blue point cloud and the yellow point cloud are water surface point clouds, the red point cloud is the original water bottom point cloud without correction, and the green point cloud is the water bottom point cloud after wave refraction correction. It can be seen that the water depth deviation is obviously corrected. The bottom structure of the pool is in a three-step ladder type, and the distance from the deepest part of the pool bottom to the water surface is about 3.85m measured on the test day. 10 groups of points were randomly taken from the water surface points and the water bottom points at the deepest part of the pool respectively to calculate the water depth data, and the comparison before and after correction is shown in Table 2:

[0142] Table 2, refraction correction data comparison

[0143] .

[0144] According to the data comparison, the water depth deviation caused by the wave is well corrected, and the error is better than 5cm.

[0145] The above examples are only used to illustrate the technical solutions of the present application, and not to limit it. Although the present application has been described in detail with reference to the foregoing examples, those skilled in the art should understand that the technical solutions recorded in the foregoing examples can still be modified, or some or all of the technical features can be replaced by equivalents, and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for correcting errors in wave refraction correction for an oval scan laser radar, characterized by, The application relates to a method for correcting seabed point cloud data of an airborne laser radar. S1, the working environment of the airborne laser radar is that the sea state is less than or equal to level three, the wind force is less than or equal to level four, and the wave height is less than or equal to 1.25 m; after original sea surface point cloud data are obtained, a K-dimensional tree is constructed for the original sea surface point cloud data set, a statistical outlier removal filtering algorithm is adopted for filtering processing, a density self-adaptive threshold function is introduced to dynamically correct the filtering processing threshold, and preprocessed sea surface point cloud data are obtained; S2, a K-dimensional tree is constructed for the preprocessed sea surface point cloud data set, a neighborhood radius is set based on the sea surface point cloud density, a fixed radius field is adopted as the neighborhood selection mode, sea surface point cloud in the neighborhood is subjected to weighted least square fitting, a Gaussian kernel function is selected as the weight function, the neighborhood points are centralized, then a covariance matrix is constructed to replace the polynomial fitting process, and feature decomposition is carried out to obtain sea surface slope information; S3, a carrier coordinate system and an auxiliary coordinate system are constructed by means of a deflection mirror transformation angle, the included angle between the laser incidence and the mirror normal vector is calculated, the plane normal vector of the common plane of the laser incidence light, the mirror normal and the reflected light is calculated, a reflected light direction vector equation set is constructed, the incident light direction vector for constructing an instantaneous refraction correction model is obtained based on the included angle, the plane normal vector and the equation set, a refraction correction model is constructed by means of the Snell law and vector decomposition, and the corrected seabed point cloud coordinates are obtained based on the refraction correction model.

2. The oval scan LiDAR wave refraction correction error correction method according to claim 1, characterized in that, S1 includes that the working environment of the airborne laser radar is that the sea state is less than or equal to level three, the wind force is less than or equal to level four, and the wave height is less than or equal to 1.25 m; After the original sea surface point cloud data are obtained, a K-dimensional tree is constructed for the original sea surface point cloud data set; S1 includes, S1.2, introducing a density adaptive threshold function on the basis of statistical outlier removal filtering algorithm, calculating the point Density of surrounding neighborhood points , the basic standard deviation multiplier is dynamically corrected, the formula is as follows: ； wherein is the modified dynamic threshold, is the global reference threshold multiplier, is the sensitivity coefficient, is a variable that is incremented from a starting value of 1, and the base standard deviation multiplier is a multiplier that expands or adjusts the range of the standard deviation.

3. The oval scan LiDAR wave refraction correction error correction method of claim 2, wherein, S2 includes S2.1, constructing a K-dimensional tree for the pre-processed sea surface point cloud dataset, accelerating the neighborhood search process, setting the neighborhood radius according to the sea surface point cloud density, and adopting a fixed radius neighborhood for neighborhood selection , The neighborhood point set is: ； wherein is a neighborhood point set, is a neighborhood point of is a variable that is incremented from a starting value of 1, is the Euclidean distance, is the radius of the field; S2 includes that sea surface point cloud in the neighborhood is subjected to weighted least square fitting, the fitting weight is inversely proportional to the distance of the neighborhood point to the target point, a Gaussian kernel function is selected as the weight function, and the formula is as follows: ； wherein is a fitting weight, is a natural exponential function, is a point is the Euclidean distance to each point in the neighborhood, is a bandwidth of the Gaussian kernel, the bandwidth of the Gaussian kernel being a parameter that controls the degree of smoothing of the Gaussian function; S2 comprises S2.3, computing a weighted centroid of the neighborhood point set : ； In the formula, is the number of neighborhood points; S2 comprises S2.4, centering the neighborhood points, computing the point offset to the centroid point : 。 4. The oval scan lidar wave refraction correction error correction method according to claim 3, characterized by, S2 comprises S2.5, constructing a covariance matrix : ； In the formula, respectively, the offset of the point to the centroid point in the coordinate axis ; the covariance matrix is eigenvalue decomposition, and a feature vector corresponding to the minimum eigenvalue is obtained, and the feature vector is a corresponding plane normal vector.

5. The oval scan lidar wave refraction correction error correction method according to claim 4, characterized by, S3 includes S3.1, the included angle between the mirror normal of the off-axis rotating mirror and the rotating shaft is 7.5°, the off-axis rotating mirror is arranged at an angle of 45° with the horizontal direction, and the laser is incident along the horizontal direction to construct the carrier coordinate system , The axis is towards the radar advancing direction, The axis is upwards, The axis is right to constitute a right front upper coordinate system, Around The axis is counterclockwise rotated by 45°, so that The axis coincides with the rotating shaft, and an auxiliary coordinate system is constructed .

6. The oval scan lidar wave refraction correction error correction method according to claim 5, wherein, S3 comprises S3.2, determining a laser incidence direction vector within the auxiliary coordinate system and the mirror normal vector : ； ； In the formula, is the real-time angle of the off-axis turning mirror; calculating the angle between the laser incidence and the mirror normal vector is: ； wherein is the vector length, is the transpose symbol.

7. The oval scan lidar wave refraction correction error correction method according to claim 6, characterized by, S3 comprises S3.3, based on the coplanar relationship of the laser incident light, the mirror normal and the reflected light, the normal vector of the coplanar plane is , the formula is as follows: ； Let the direction vector of the reflected light ray be , we obtain the system of equations: ； S3 comprises S3.4, to around clockwise by 45° to obtain , is the incident light direction vector when constructing the instantaneous refractive correction model : ； wherein is and the angle between the axes, is and the angle between the axes, is and the angle between the axes.

8. The oval scan lidar wave refraction correction error correction method according to claim 7, characterized by, S3 includes S3.5, the instantaneous water surface normal vector of the laser penetration is , the direction vector of the light ray after refraction by the water body is , and the cosine value of the incident angle of the light ray penetrating the water surface is . ； According to Snell's law and the triangle identity , the refractive angle of the light ray after penetrating the water surface The sine value is: ； wherein n0is the refractive index of the incident medium, n1is the refractive index of the refractive medium.

9. The oval scan lidar wave refraction correction error correction method according to claim 8, characterized by, S3 includes S3.6, according to the vector operation law, the parallel component of decomposition along the incident plane normal and tangential, according to Snell's law, decomposition along the incident plane normal and tangential, according to Snell's law, and the parallel component satisfies the relationship , the vertical component is related to the cosine of the refraction angle The decomposition formula of the incident light is: ； ； wherein represents the perpendicular component, represents the parallel component.

10. The oval scan lidar wave refraction correction error correction method according to claim 9, wherein, S3 comprises S3.7, to Modelled as the sum of a perpendicular component and a parallel component: ； ； After the arrangement, the refraction vector expression after water refraction is obtained: ； According to the refraction vector expression after water refraction, the seabed point cloud coordinates are corrected: ； wherein is the corrected back coordinate, is the slant range from the laser launch position to the sea surface point, is the speed of light in air, is the time of flight of the light in water.

Images