A strip splicing robust processing method and system for clustered multi-beams

Abstract

The present invention discloses a strip splicing anti-error processing method and system for cluster multi-beam, the method comprising: calculating the curvature and verticality of unmanned boat multi-beam strip data to construct a set of feature points with the same name; weighting the set of feature points with the same name through Huber function to obtain the weight value of the set of feature points with the same name; constructing a seven-parameter transformation model for the set of feature points with the same name, and iteratively solving it through nonlinear least squares adjustment to obtain the solved seven-parameter transformation model for the set of feature points with the same name; based on the solved seven-parameter transformation model for the set of feature points with the same name, the splicing of the unmanned boat multi-beam strip data is realized. The present invention can make full use of the strong matching relationship between the points with the same name, limit the use of the weak matching relationship, and thus improve the accuracy of multi-beam strip splicing. As a strip splicing anti-error processing method and system for cluster multi-beam, the present invention can be widely used in the technical field of multi-beam strip data splicing.

Description

Technical Field

[0001] The present invention relates to the technical field of multi-beam stripe data splicing, and in particular to a stripe splicing robustness processing method and system for clustered multi-beams. Background Art

[0002] In multi-beam data collection based on unmanned boat clusters, due to differences in the size of unmanned boat platforms, equipment models, sea conditions, etc., there is often a certain depth offset between adjacent strips. Therefore, strip splicing is a very critical step in multi-beam data processing. Related technologies, such as the side-scan sonar strip image matching method based on the invariance of ground objects, determine the common coverage area of ​​adjacent strips, use the SURF algorithm to detect feature points in the overlapping area and perform rough matching. Then, based on the invariance of the ground objects, the abnormal matching point pairs are eliminated, and precise matching is performed on this basis. According to the precise matching results, the four-parameter method is used to obtain the overall rigid transformation model. The matching point pairs are classified by K-means, and the local areas containing obvious features are delineated according to the classification results. The thin plate spline function is used to establish a local elastic transformation model for each divided area. The results of rigid transformation first and then elastic transformation are used in the area, and the rigid transformation results are directly used in the interval, so as to realize the splicing of the common coverage area of ​​adjacent strip images. The related technology directly uses the extracted corresponding points for matching transformation, which is a cumbersome process, time-consuming and low in accuracy.

[0003] Therefore, due to the complexity of system errors and the high dynamics of the ocean environment, the corresponding points (or feature points with the same name) extracted by the stitching method of related technologies often have disturbances, resulting in insufficient accuracy of multi-beam strip stitching. Summary of the invention

[0004] In order to solve the above technical problems, the purpose of the present invention is to provide a strip splicing anti-error processing method and system for clustered multi-beam, which can make full use of the strong matching relationship between the same-name points and limit the use of weak matching relationship, thereby improving the accuracy of multi-beam strip splicing.

[0005] The first technical solution adopted by the present invention is: a strip splicing robustness processing method for clustered multi-beams, comprising the following steps:

[0006] Calculate the curvature and verticality of the unmanned boat multi-beam strip data to construct a set of feature points with the same name, where the set of feature points with the same name represents a set of corresponding feature points between adjacent unmanned boat strips;

[0007] The Huber function is used to estimate the weight of the feature point set with the same name, and the weight value of the feature point set with the same name is obtained;

[0008] Combined with the weight values ​​of the feature point set with the same name, a seven-parameter transformation model of the feature point set with the same name is constructed, and the iterative solution is performed through the nonlinear least squares adjustment method to obtain the solved seven-parameter transformation model of the feature point set with the same name;

[0009] The multi-beam strip data of the unmanned boat is spliced ​​based on the solved seven-parameter transformation model of the feature point set with the same name.

[0010] Furthermore, the step of calculating the curvature and verticality of the multi-beam strip data of the unmanned boat and constructing a set of feature points with the same name specifically includes:

[0011] The multi-beam strip data of the unmanned boat is obtained by using the multi-beam bathymetric equipment carried by the unmanned boat;

[0012] According to the local neighborhood points of the multi-beam strip data of the unmanned boat, a three-dimensional covariance matrix is ​​constructed;

[0013] Determine the characteristic value of each point of the multi-beam strip data of the unmanned boat according to the three-dimensional covariance matrix and calculate the curvature and verticality to obtain the curvature change value of the strip data and the verticality value of the strip data;

[0014] Set the curvature threshold and the verticality threshold, select the points of the strip data that are greater than the curvature threshold and greater than the verticality threshold, and construct a set of significant feature points;

[0015] The KD tree algorithm is used to traverse the set of significant feature points, and the points with the smallest distance between two strip data are selected to construct a set of feature points with the same name.

[0016] Furthermore, the expression of the three-dimensional covariance matrix is ​​specifically as follows:

[0017] ;

[0018] In the above formula, represents the three-dimensional covariance matrix, represents the total number of neighboring points, Indicates Neighboring points, Indicates Neighbors The centroid of the neighborhood.

[0019] Furthermore, the curvature and verticality calculation expressions are specifically as follows:

[0020] ;

[0021] In the above formula, represents the curvature value, Indicates the verticality value, Represents the characteristic value of the multi-beam strip data of the unmanned boat, Represents the normal vector corresponding to the neighboring point.

[0022] Furthermore, the step of estimating the weight of the feature point set with the same name by using the Huber function to obtain the weight value of the feature point set with the same name specifically includes:

[0023] Calculate the distances between several pairs of feature point sets with the same name to obtain the Euclidean distances of the feature point sets with the same name;

[0024] Set the Euclidean distance threshold;

[0025] Based on the Huber function and combined with the Euclidean distance threshold, each pair of feature points with the same name is weighted according to the Euclidean distance of the feature point set with the same name, and the weight value of the feature point set with the same name is obtained.

[0026] Further, the step of combining the weight values ​​of the feature point sets with the same name, constructing a seven-parameter transformation model for the feature point sets with the same name, and iteratively solving the model through the nonlinear least squares adjustment method to obtain the solved seven-parameter transformation model for the feature point sets with the same name specifically includes:

[0027] The transformation relationship of the feature point set with the same name is expressed by a seven-parameter transformation model, and a seven-parameter transformation model of the feature point set with the same name is constructed;

[0028] According to the seven-parameter transformation model of the feature point set with the same name, the error equation of the seven-parameter transformation model of the feature point set with the same name is obtained, and the seven-parameter adjustment value is determined;

[0029] Based on the error equation, combined with the weight value of the feature point set with the same name, and linearized, the linearized error equation is obtained;

[0030] Perform Taylor expansion on the linearized error equation to obtain the linearization result of the first-order term;

[0031] Substitute the linearized error equation into the linearized result of the first-order term to solve and obtain the correction value;

[0032] The seven-parameter adjustment values ​​are substituted into the correction values ​​for iterative calculation until the calculation result meets the preset threshold, and the solved seven-parameter transformation model of the set of feature points with the same name is obtained.

[0033] Furthermore, the expression of the seven-parameter transformation model of the feature point set with the same name is specifically as follows:

[0034] ;

[0035] In the above formula, represents the number of corresponding points obtained, Indicates feature points with the same name, Indicates the coordinate values ​​of feature points with the same name, Represents the difference between the corresponding feature points with the same name, Represents the rotation equations around the coordinate axes.

[0036] Furthermore, the expression of the linearization result of the first-order term is specifically as follows:

[0037] ;

[0038] In the above formula, represents the objective function, represents the weight matrix, represents seven unknown parameters, represents the approximate values ​​of the seven unknown parameters, represents the correction values ​​of seven unknown parameters, A vector of observations representing the coordinates of the strip points.

[0039] Furthermore, the expression of the correction value is specifically as follows:

[0040] ;

[0041] In the above formula, represents the correction values ​​of seven unknown parameters, represents the weight matrix, Represents an intermediate variable parameter.

[0042] The second technical solution adopted by the present invention is: a strip splicing robust processing system for clustered multi-beams, comprising:

[0043] The first module is used to calculate the curvature and verticality of the unmanned boat multi-beam strip data and construct a set of feature points with the same name, wherein the set of feature points with the same name represents a set of corresponding feature points between adjacent unmanned boat strips;

[0044] The second module is used to estimate the weight of the feature point set with the same name through the Huber function to obtain the weight value of the feature point set with the same name;

[0045] The third module is used to construct a seven-parameter transformation model of the feature point set with the same name by combining the weight values ​​of the feature point set with the same name, and iteratively solve it by nonlinear least square adjustment method to obtain the solved seven-parameter transformation model of the feature point set with the same name;

[0046] The fourth module is used to realize the splicing of multi-beam strip data of unmanned boats based on the solved seven-parameter transformation model of the same-name feature point set.

[0047] The method and system of the present invention have the following beneficial effects: the present invention calculates the curvature and verticality of the multi-beam strip data of the unmanned boat, constructs a set of feature points with the same name, obtains a set of significant feature points of adjacent strips by setting curvature and verticality thresholds, thereby improving the matching accuracy and speed of the set of feature points with the same name, further estimates the weight of the set of feature points with the same name by using the Huber function, thereby making full use of the strong matching relationship between the points with the same name, limiting the use of the weak matching relationship, and reducing the disturbance of the feature points with the same name, and finally constructs a seven-parameter transformation model of the set of feature points with the same name in combination with the weight value of the set of feature points with the same name, and iteratively solves it by using the nonlinear least squares adjustment method, thereby improving the calculation accuracy of the seven-parameter transformation model of the set of feature points with the same name, and realizes the splicing of the multi-beam strip data of the unmanned boat based on the solved seven-parameter transformation model of the set of feature points with the same name, thereby improving the precision of the splicing of the multi-beam strips. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 It is a flow chart of the steps of a strip splicing robustness processing method for clustered multi-beams of the present invention;

[0049] Figure 2 It is a structural block diagram of a strip splicing robust processing system for clustered multi-beams of the present invention;

[0050] Figure 3 It is a schematic diagram of the process of multi-beam strip splicing provided by a specific embodiment of the present invention. DETAILED DESCRIPTION

[0051] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. The step numbers in the following embodiments are only provided for the convenience of explanation and description, and the order between the steps is not limited in any way. The execution order of each step in the embodiment can be adaptively adjusted according to the understanding of those skilled in the art.

[0052] Reference Figure 1 and Figure 3 The present invention provides a strip splicing robustness processing method for clustered multi-beams, the method comprising the following steps:

[0053] S100, calculating the curvature and verticality of the unmanned boat multi-beam strip data, and constructing a set of feature points with the same name, wherein the set of feature points with the same name represents a set of corresponding feature points between adjacent unmanned boat strips;

[0054] S110, obtaining multi-beam strip data of the unmanned boat by using a multi-beam bathymetric device carried by the unmanned boat;

[0055] In this embodiment, the multi-beam strip data of the unmanned boat is obtained by the multi-beam depth sounding equipment carried by the unmanned boat. The unmanned boat obtains underwater multi-beam data in a cluster manner. The data obtained by a single unmanned boat is a single strip data, and the data obtained by adjacent unmanned boats are adjacent multi-beam strip data.

[0056] S120, constructing a three-dimensional covariance matrix according to the local neighborhood points of the unmanned boat multi-beam strip data;

[0057] In this embodiment, for any point on the multi-beam strip, a given point , select Nearest Neighbor Depending on the respective 3D structure and local point density in the data set, a three-dimensional covariance matrix is ​​constructed, which is expressed as:

[0058] ;

[0059] In the above formula, represents the three-dimensional covariance matrix, represents the total number of neighboring points, Indicates Neighboring points, Indicates Neighbors The centroid of the neighborhood.

[0060] S130, determining the characteristic value of each point of the unmanned boat multi-beam strip data according to the three-dimensional covariance matrix and calculating the curvature and verticality to obtain the curvature change value of the strip data and the verticality value of the strip data;

[0061] In this embodiment, the three-dimensional covariance matrix represents a symmetric positive definite matrix, which has three eigenvalues ​​that exist and are non-negative. Define the three eigenvalues ​​of the matrix , used to represent the local three-dimensional shape, where The corresponding eigenvector is the normal vector of the point. The eigenvalue is used to estimate the change of the local surface, that is, the change of curvature. , its expression is as follows:

[0062] ;

[0063] In the above formula, represents the curvature value, Indicates the verticality value, Represents the characteristic value of the multi-beam strip data of the unmanned boat, Represents the normal vector corresponding to the neighboring point.

[0064] S140, setting a curvature threshold and a verticality threshold, selecting points of the strip data that are greater than the curvature threshold and greater than the verticality threshold, and constructing a set of significant feature points;

[0065] S150, traversing the set of significant feature points by using the KD tree algorithm, and selecting the point with the smallest distance between two strip data to construct a set of feature points with the same name.

[0066] In this embodiment, the curvature and verticality thresholds are set. and Then the feature point is a significant feature point, and the strip is obtained from it The set of salient feature points , adjacent strips The set of salient feature points is , use KD tree to traverse the significant feature points of adjacent strips and ,like and The distance is the smallest and is less than 1, then it is considered are feature points with the same name, from which we can get a set of feature points with the same name and significant features , is the number of corresponding points obtained.

[0067] S200, weight estimation is performed on a set of feature points with the same name by using a Huber function to obtain a weight value of the set of feature points with the same name;

[0068] Specifically, the distances between several pairs of feature point sets with the same name are calculated to obtain the Euclidean distance of the feature point sets with the same name; a Euclidean distance threshold is set; based on the Huber function and in combination with the Euclidean distance threshold, each pair of feature points with the same name is given a weight according to the Euclidean distance of the feature point sets with the same name, to obtain the weight value of the feature point sets with the same name.

[0069] In this embodiment, the feature point set with the same name and significant features Not all points in the image are exactly matched. The distance between the corresponding points is divided into two categories: strong matching points and weak matching points. The weights between the corresponding points are calculated by the Huber function to fully strengthen the matching relationship and limit the weak matching relationship. The Euclidean distance between the corresponding points is calculated as , the expression estimated by Huber function is:

[0070] ;

[0071] In the above formula, represents the residual, represents the symbolic function, Represents the distance threshold, expressed as the Euclidean distance between corresponding points.

[0072] Among them, the corresponding weight function As shown below:

[0073] ;

[0074] This means that the greater the distance between two points, the weaker the matching relationship and the smaller the corresponding weight; conversely, the stronger the matching relationship and the greater the corresponding weight.

[0075] S300, combining the weight values ​​of the feature point set with the same name, constructing a seven-parameter transformation model for the feature point set with the same name, and performing iterative solution through a nonlinear least squares adjustment method to obtain a solved seven-parameter transformation model for the feature point set with the same name;

[0076] S310, expressing the transformation relationship of the feature point set with the same name by a seven-parameter transformation model, and constructing a seven-parameter transformation model for the feature point set with the same name;

[0077] In this embodiment, a seven-parameter transformation model is used to represent adjacent strips according to the feature point set with the same name. and The transformation relationship between them, the expression of the seven-parameter transformation model of the feature point set with the same name is:

[0078] ;

[0079] In the above formula, represents the number of corresponding points obtained, Indicates feature points with the same name, Indicates the coordinate values ​​of feature points with the same name, Represents the difference between the corresponding feature points with the same name, Represents the rotation equations around the coordinate axes.

[0080] Among them, it should be noted that:

[0081] ;

[0082] In the above formula, Respectively The point around The angle of rotation.

[0083] S320, obtaining an error equation of the seven-parameter transformation model of the feature point set with the same name according to the seven-parameter transformation model of the feature point set with the same name, and determining the seven-parameter adjustment value;

[0084] In this embodiment, according to the feature point set with the same name , nonlinear least squares adjustment is used to solve the seven parameters of the transformation model in the expression of the seven-parameter transformation model of the set of characteristic points with the same name , the error equation of the seven-parameter model is as follows:

[0085] ;

[0086] In the above formula, The observation vector representing the coordinates of the strip points, Represents a vector of correction values.

[0087] Furthermore, it should be noted that the seven parameter unknowns are expressed as , its approximate value is expressed as , the corresponding correction value is Adjustment value Calculated as .

[0088] S330, based on the error equation, combined with the weight value of the feature point set with the same name, and linearized, to obtain a linearized error equation;

[0089] In this embodiment, according to the least squares principle, the expression of the least squares solution is as follows:

[0090] ;

[0091] In the above formula, represents the corresponding weight matrix, The expression for the least squares solution of the nonlinear equation composed of Huber weights is as follows:

[0092] ;

[0093] The expression of the least square solution of the nonlinear equation is linearized to obtain the linearized error equation, which is expressed as follows:

[0094] ;

[0095] In the above formula, represents seven unknown parameters, represents the approximate values ​​of the seven unknown parameters, Represents the correction values ​​of seven unknown parameters.

[0096] S340, performing Taylor expansion on the linearized error equation to obtain a linearization result of the first-order term;

[0097] In this embodiment, when Taylor expansion is used for the linearized error equation, the linearization result of the first-order term is as follows:

[0098] ;

[0099] In the above formula, represents the objective function, represents the weight matrix, represents seven unknown parameters, represents the approximate values ​​of the seven unknown parameters, represents the correction values ​​of seven unknown parameters, A vector of observations representing the coordinates of the strip points.

[0100] make , both of which are A cubic matrix of order , where is the number of observation equations, is the number of basic observations, This holds true for all equations.

[0101] S350, substituting the linearized error equation into the linearized result of the first-order term to solve it and obtain a correction value;

[0102] S360, substituting the seven-parameter adjustment value into the correction value for iterative calculation until the calculation result meets the preset threshold value, thereby obtaining the solved seven-parameter transformation model of the feature point set with the same name.

[0103] In this embodiment, by substituting the linearized error equation into the linearized result of the first-order term and taking into account the above signs, the correction value The representation is as follows:

[0104] ;

[0105] Correction value The expression can be simplified to:

[0106] ;

[0107] The adjustment value Approximately equal to value . Adjust the value Substitute the correction value The expression is calculated iteratively until The axis is less than a specified threshold, thus deriving a solution using a nonlinear least-squares adjustment , so as to solve the parameters of the seven-parameter transformation model, and then use the model to complete the splicing of adjacent multi-beam strip data.

[0108] S400, based on the solved seven-parameter transformation model of the feature point set with the same name, realize the splicing of the multi-beam strip data of the unmanned boat.

[0109] In summary, the embodiment of the present invention obtains a set of feature points with the same name by calculating the curvature and verticality of each point and setting corresponding thresholds; according to the Euclidean distance between the feature points with the same name, Huber is used to determine the corresponding weights for the points with the same name, so as to make full use of the strong matching relationship between the points with the same name and limit the use of weak matching relationships; a seven-parameter transformation model is constructed, and the seven parameters of the model are iteratively solved by a nonlinear least squares adjustment method.

[0110] It should be further explained that, in the embodiment of the present invention, when obtaining a set of feature points with the same name By setting the curvature and verticality Threshold, when satisfied and , get the set of significant feature points of adjacent strips and Then, the KD tree is used to traverse the set of significant feature points of adjacent strips to obtain the corresponding set of feature points with the same name. When solving the constructed seven-parameter transformation model, the correspondence between the solved points with the same name is considered. The weight between each corresponding point is determined by the Huber function. The distance between The corresponding points are divided into two categories: strong matching correspondence and weak matching correspondence, so as to make full use of the strong matching information and limit the weak matching information.

[0111] Therefore, compared with the prior art, the advantages of the embodiments of the present invention are that the speed of splicing strip multi-beams is faster, the splicing accuracy is higher, and the anti-interference ability is stronger. It can be applied to the multi-beam data strips collected by unmanned boat clusters to efficiently obtain high-precision data splicing results that meet engineering requirements. This patent sets the curvature and verticality thresholds to perform high-precision matching of feature points with the same name and quickly obtain a set of feature points with the same name. In addition, a seven-parameter transformation model is adopted to determine the weights of the corresponding feature points through the Huber function, fully utilize the strong matching correspondence of the points with the same name, limit the use of weak matching correspondence, improve the anti-interference ability of the model, and further improve the accuracy of the model parameters.

[0112] Reference Figure 2 , a strip splicing robust processing system for clustered multi-beams, comprising:

[0113] The first module 201 is used to calculate the curvature and verticality of the unmanned boat multi-beam strip data and construct a set of feature points with the same name, wherein the set of feature points with the same name represents a set of corresponding feature points between adjacent unmanned boat strips;

[0114] The second module 202 is used to estimate the weight of the feature point set with the same name by using the Huber function to obtain the weight value of the feature point set with the same name;

[0115] The third module 203 is used to construct a seven-parameter transformation model of the feature point set with the same name in combination with the weight value of the feature point set with the same name, and iteratively solve it by a nonlinear least squares adjustment method to obtain the solved seven-parameter transformation model of the feature point set with the same name;

[0116] The fourth module 204 is used to realize the splicing of the multi-beam strip data of the unmanned boat based on the solved seven-parameter transformation model of the feature point set with the same name.

[0117] The contents of the above method embodiments are all applicable to the present system embodiments. The functions specifically implemented by the present system embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0118] The above is a specific description of the preferred implementation of the present invention, but the invention is not limited to the embodiments. Those skilled in the art may make various equivalent modifications or substitutions without violating the spirit of the present invention. These equivalent modifications or substitutions are all included in the scope defined by the claims of this application.

Claims

1. A strip splicing robustness processing method for clustered multi-beams, characterized in that: The following steps are involved: Calculate the curvature and verticality of the unmanned boat multi-beam strip data to construct a set of feature points with the same name, where the set of feature points with the same name represents a set of corresponding feature points between adjacent unmanned boat strips; The Huber function is used to estimate the weight of the feature point set with the same name, and the weight value of the feature point set with the same name is obtained; The transformation relationship of the feature point set with the same name is expressed by a seven-parameter transformation model, and a seven-parameter transformation model of the feature point set with the same name is constructed; The expression of the seven-parameter transformation model of the same-name feature point set is specifically as follows: ； In the above formula, represents the number of corresponding points obtained, Indicates feature points with the same name, Indicates the coordinate values ​​of feature points with the same name, Represents the difference between the corresponding feature points with the same name, Represents the rotation parameter around the coordinate axis; According to the seven-parameter transformation model of the feature point set with the same name, the error equation of the seven-parameter transformation model of the feature point set with the same name is obtained, and the seven-parameter adjustment value is determined; Based on the error equation, combined with the weight value of the feature point set with the same name, and linearized, the linearized error equation is obtained; Perform Taylor expansion on the linearized error equation to obtain the linearization result of the first-order term; Substitute the linearized error equation into the linearized result of the first-order term to solve and obtain the correction value; Substitute the seven-parameter adjustment value into the correction value for iterative calculation until the calculation result meets the preset threshold, and obtain the solved seven-parameter transformation model of the feature point set with the same name; The multi-beam strip data of the unmanned boat is spliced ​​based on the solved seven-parameter transformation model of the feature point set with the same name.

2. According to claim 1, a strip splicing robustness processing method for clustered multi-beams is characterized in that: The step of calculating the curvature and verticality of the unmanned boat multi-beam strip data and constructing a set of feature points with the same name specifically includes: The multi-beam strip data of the unmanned boat is obtained by using the multi-beam bathymetric equipment carried by the unmanned boat; According to the local neighborhood points of the multi-beam strip data of the unmanned boat, a three-dimensional covariance matrix is ​​constructed; Determine the characteristic value of each point of the multi-beam strip data of the unmanned boat according to the three-dimensional covariance matrix and calculate the curvature and verticality to obtain the curvature change value of the strip data and the verticality value of the strip data; Set the curvature threshold and the verticality threshold, select the points of the strip data that are greater than the curvature threshold and greater than the verticality threshold, and construct a set of significant feature points; The KD tree algorithm is used to traverse the set of significant feature points, and the points with the smallest distance between two strip data are selected to construct a set of feature points with the same name.

3. According to claim 2, a strip splicing robustness processing method for clustered multi-beams is characterized in that: The expression of the three-dimensional covariance matrix is ​​specifically as follows: ； In the above formula, represents the three-dimensional covariance matrix, represents the total number of neighboring points, Indicates Neighboring points, Indicates Neighbors The centroid of the neighborhood.

4. According to claim 3, a strip splicing robustness processing method for clustered multi-beams is characterized in that: The curvature and verticality calculation expressions are specifically as follows: ； In the above formula, represents the curvature value, Indicates the verticality value, Represents the characteristic value of the multi-beam strip data of the unmanned boat, Represents the normal vector corresponding to the neighboring point.

5. According to claim 4, a strip splicing robustness processing method for clustered multi-beams is characterized in that: The step of estimating the weight of the feature point set with the same name by using the Huber function to obtain the weight value of the feature point set with the same name specifically includes: Calculate the distances between several pairs of feature point sets with the same name to obtain the Euclidean distances of the feature point sets with the same name; Set the Euclidean distance threshold; Based on the Huber function and combined with the Euclidean distance threshold, each pair of feature points with the same name is weighted according to the Euclidean distance of the feature point set with the same name, and the weight value of the feature point set with the same name is obtained.

6. According to claim 5, a strip splicing robustness processing method for clustered multi-beams is characterized in that: The expression of the linearization result of the first-order term is specifically as follows: ； In the above formula, represents the objective function, represents the weight matrix, represents seven unknown parameters, represents the approximate values ​​of the seven unknown parameters, represents the correction values ​​of seven unknown parameters, A vector of observations representing the coordinates of the strip points.

7. The strip splicing robustness processing method for clustered multi-beam according to claim 6, characterized in that: The expression of the correction value is specifically as follows: ； In the above formula, represents the correction values ​​of seven unknown parameters, represents the weight matrix, Represents an intermediate variable parameter.

8. A strip splicing robust processing system for clustered multi-beams, characterized in that: Includes the following modules: The first module is used to calculate the curvature and verticality of the unmanned boat multi-beam strip data and construct a set of feature points with the same name, wherein the set of feature points with the same name represents a set of corresponding feature points between adjacent unmanned boat strips; The second module is used to estimate the weight of the feature point set with the same name through the Huber function to obtain the weight value of the feature point set with the same name; The third module is used to express the transformation relationship of the feature point set with the same name through a seven-parameter transformation model, and construct a seven-parameter transformation model for the feature point set with the same name; The expression of the seven-parameter transformation model of the same-name feature point set is specifically as follows: ； In the above formula, represents the number of corresponding points obtained, Indicates feature points with the same name, Indicates the coordinate values ​​of feature points with the same name, Represents the difference between the corresponding feature points with the same name, Represents the rotation parameter around the coordinate axis; According to the seven-parameter transformation model of the feature point set with the same name, the error equation of the seven-parameter transformation model of the feature point set with the same name is obtained, and the seven-parameter adjustment value is determined; Based on the error equation, combined with the weight value of the feature point set with the same name, and linearized, the linearized error equation is obtained; Perform Taylor expansion on the linearized error equation to obtain the linearization result of the first-order term; Substitute the linearized error equation into the linearized result of the first-order term to solve and obtain the correction value; Substitute the seven-parameter adjustment value into the correction value for iterative calculation until the calculation result meets the preset threshold, and obtain the solved seven-parameter transformation model of the feature point set with the same name; The fourth module is used to realize the splicing of multi-beam strip data of unmanned boats based on the solved seven-parameter transformation model of the same-name feature point set.

Images