Data registration method and system based on multi-beam echo sounder

Abstract

The present invention provides a data registration method and system based on a multi-beam echo sounder, the method comprising the following steps: obtaining observation point cloud data and transforming it to a mobile coordinate system; coarsely registering data of two adjacent frames based on an ICP algorithm; based on the waveform characteristics of the observation point cloud data, screening a feature point set whose curvature meets a threshold value, and constructing a feature matching point pair; establishing an objective function about a transformation matrix based on the maximum value of a correlation entropy model as a target, and completing iterative optimization of precise registration of data. By sequentially transforming the detection data to a reference coordinate system and a mobile coordinate system to eliminate the influence of installation errors and posture errors, the measurement accuracy of the data is improved, and by screening the data to obtain a feature point set whose curvature meets a threshold value, the amount of data processing and data redundancy can be reduced, and then by introducing correlation entropy as an error metric and combining iterative optimization methods, high-precision registration of data can be achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of multi-beam echo sounders, and in particular to a data registration method and system based on a multi-beam echo sounder. Background Art

[0002] In ocean mapping, multi-beam echo sounders are one of the most widely used devices, and their main function is to accurately map the seabed topography.

[0003] The multi-beam echo sounder adopts a one-transmit-multiple-receive strategy. Its array is arranged as a T-shaped transmitting array and receiving array. The transmitting beam and the receiving beam intersect on the seabed. After electronic beam forming at the data end, detection data containing terrain features is obtained.

[0004] In the existing technology, affected by the wind, waves and currents at sea, the posture of the multi-beam echo sounder installation platform has an adverse effect on the sounding accuracy and sounding position, affecting the measurement accuracy and making subsequent alignment and modeling and mapping operations inconvenient. Summary of the invention

[0005] In view of the deficiencies in the prior art, the object of the present invention is to provide a data registration method and system based on a multi-beam echo sounder, aiming to solve the technical problems in the prior art of poor measurement accuracy and inconvenience for subsequent registration.

[0006] In order to achieve the above object, the present invention is implemented by the following technical scheme: a data registration method based on a multi-beam echo sounder, comprising the following steps:

[0007] A detection coordinate system is established based on the measurement coordinates of the multi-beam echo sounder, and observation point cloud data in the detection coordinate system is obtained, and based on the installation error angle between the inertial navigation system and the multi-beam echo sounder, the observation point cloud data is transformed into a reference coordinate system of the inertial navigation system;

[0008] Establishing a mobile coordinate system based on the odometer and the preset navigation trajectory, and transforming the observation point cloud data in the reference coordinate system to the mobile coordinate system based on the attitude data measured by the inertial navigation system;

[0009] Based on the ICP algorithm, the first observation point cloud data and the second observation point cloud data of two adjacent frames in the mobile coordinate system are roughly aligned;

[0010] Based on the waveform characteristics of the observation point cloud data, a target point is selected from the first observation point cloud data that has completed the rough registration, and a curvature characteristic calculation is performed on the target point to screen out a feature point set whose curvature meets a threshold value;

[0011] Selecting points whose Euclidean distances with the feature point set are the minimum from the second observation point cloud data to form a plurality of matching point pairs;

[0012] Selecting, from the plurality of matching point pairs, feature matching point pairs whose Euclidean distance between the matching point pairs is less than a preset distance value and whose normal vector angle is less than a preset angle value;

[0013] Establishing a correlation entropy model of the distance between the first observation point cloud data and the second observation point cloud data, establishing an objective function about the transformation matrix based on the maximum value of the correlation entropy model as a target, and calculating the mean square error between each of the feature matching point pairs according to the objective function;

[0014] If the mean square error is less than a preset threshold, or the number of iterations reaches a preset number, the iteration is stopped to complete the precise registration of the first observation point cloud data and the second observation point cloud data.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: after acquiring detection data based on a detection coordinate system of a multi-beam echo sounder, the detection data is successively transformed into a reference coordinate system and a mobile coordinate system to eliminate the influence of the installation error and attitude error of the multi-beam echo sounder and improve the measurement accuracy of the data; after synchronously transforming two consecutive frames of data into the mobile coordinate system, a rough alignment is first performed using an ICP algorithm, and a feature point set whose curvature meets a threshold is obtained by screening the data, thereby reducing the amount of data processing and data redundancy; data points related to valuable terrain features are retained based on waveform features; and then, by introducing correlation entropy as an error metric and combining an iterative optimization method, high-precision alignment of the data can be achieved.

[0016] According to one aspect of the above technical solution, the step of calculating the curvature feature of the target point to screen out a set of feature points whose curvature meets the threshold specifically includes:

[0017] Calculate the curvature value and the normal vector angle of the target point, combine the curvature value and the normal vector angle to obtain the curvature feature of the target point, and screen the feature point set whose curvature meets the threshold based on the curvature feature;

[0018] The calculation expression of the curvature characteristic is as follows:

[0019] ;

[0020] Wherein, V is the curvature feature, W is the mean value of the normal vector angle of the target point, and J is the average curvature value of the target point.

[0021] According to one aspect of the above technical solution, the calculation expression of the average curvature value is as follows:

[0022] ;

[0023] In the formula, and is the first-order partial derivative of the surface corresponding to the target point, , and is the second-order partial derivative of the surface corresponding to the target point, J is the mean curvature value, and n is the unit normal vector of the surface corresponding to the target point.

[0024] According to one aspect of the above technical solution, before the step of calculating the curvature value of the target point and the normal vector angle, the method further includes:

[0025] Based on the target point and the neighborhood points of the target point, local surface fitting is performed by least square method;

[0026] The calculation expression of the mean value of the normal vector angle is as follows:

[0027] ;

[0028] In the formula, z is the number of neighborhood points, is the normal vector of the target point, is the normal vector of the tth neighborhood point, and W is the mean angle of the normal vector.

[0029] According to one aspect of the above technical solution, the calculation expression of the objective function is:

[0030] ;

[0031] ;

[0032] In the formula, is the objective function, M is the correlation entropy model, N is the number of feature matching point pairs, R and T are the rotation matrix and translation matrix of the first observation point cloud data transformed to the second observation point cloud data, respectively. and are respectively the i-th feature matching point in the feature matching point pair corresponding to the first observation point cloud data and the second observation point cloud data, is the normal vector of the feature matching point corresponding to the second observation point cloud data, is the kernel length parameter during iteration.

[0033] According to one aspect of the above technical solution, the calculation expression of the mean square error is:

[0034] ;

[0035] In the formula, is the mean square error, and k is the number of iterations.

[0036] According to one aspect of the above technical solution, the coordinate expression for transforming the observed point cloud data into the reference coordinate system of the inertial navigation system is:

[0037] ;

[0038] ;

[0039] ;

[0040] ;

[0041] ;

[0042] ;

[0043] In the formula, , , Transform the observed point cloud data into coordinates in the reference coordinate system, is the eccentricity between the reference coordinate system and the detection coordinate system, is the pitch installation error angle, To measure the roller installation error angle, is the yaw installation error angle, , , is the coordinate of the observed point cloud data in the detection coordinate system, is the transformation matrix between the detection coordinate system and the reference coordinate system.

[0044] According to one aspect of the above technical solution, the coordinate expression for transforming the observation point cloud data in the reference coordinate system to the mobile coordinate system is:

[0045] ;

[0046] ;

[0047] ;

[0048] ;

[0049] ;

[0050] ;

[0051] In the formula, , , Transform the observed point cloud data into coordinates in the mobile coordinate system, is the eccentricity between the mobile coordinate system and the reference coordinate system, is the roll attitude angle, is the pitch attitude angle, is the yaw attitude angle, is the transformation matrix between the reference coordinate system and the moving coordinate system.

[0052] On the other hand, the present application also provides a data registration system based on a multi-beam echo sounder, comprising:

[0053] A first coordinate transformation module is used to establish a detection coordinate system based on the measurement coordinates of the multi-beam echo sounder, obtain observation point cloud data in the detection coordinate system, and transform the observation point cloud data into a reference coordinate system of the inertial navigation system based on an installation error angle between the inertial navigation system and the multi-beam echo sounder;

[0054] A second coordinate transformation module is used to establish a mobile coordinate system based on the odometer and the preset navigation trajectory, and transform the observation point cloud data in the reference coordinate system into the mobile coordinate system based on the attitude data measured by the inertial navigation system;

[0055] A coarse registration module, used for coarsely registering the first observation point cloud data and the second observation point cloud data of two adjacent frames in the mobile coordinate system based on an ICP algorithm;

[0056] A curvature module, for selecting a target point from the first observation point cloud data that has completed the rough registration based on the waveform characteristics of the observation point cloud data, and performing curvature characteristic calculation on the target point to screen out a feature point set whose curvature meets a threshold value;

[0057] A first matching module is used to select points whose Euclidean distances with the feature point set are minimum from the second observation point cloud data to form a plurality of matching point pairs;

[0058] A second matching module is used to select, from the plurality of matching point pairs, feature matching point pairs whose Euclidean distance between the matching point pairs is less than a preset distance value and whose normal vector angle is less than a preset angle value;

[0059] A fine registration module, used to establish a correlation entropy model of the distance between the first observation point cloud data and the second observation point cloud data, establish an objective function about the transformation matrix based on the maximum value of the correlation entropy model as a target, and calculate the mean square error between each of the feature matching point pairs according to the objective function;

[0060] The iteration module is used to stop iteration if the mean square error is less than a preset threshold or the number of iterations reaches a preset number, so as to complete the precise alignment of the first observation point cloud data and the second observation point cloud data.

[0061] According to one aspect of the above technical solution, the curvature module is specifically used for:

[0062] Calculate the curvature value and the normal vector angle of the target point, combine the curvature value and the normal vector angle to obtain the curvature feature of the target point, and screen the feature point set whose curvature meets the threshold based on the curvature feature;

[0063] The calculation expression of the curvature characteristic is as follows:

[0064] ;

[0065] Wherein, V is the curvature feature, W is the mean value of the normal vector angle of the target point, and J is the average curvature value of the target point.

[0066] According to one aspect of the above technical solution, the curvature module is further used to: perform local surface fitting by least square method based on the target point and the neighborhood points of the target point;

[0067] The calculation expression of the mean value of the normal vector angle is as follows:

[0068] ;

[0069] In the formula, z is the number of neighborhood points, is the normal vector of the target point, is the normal vector of the tth neighborhood point, and W is the mean angle of the normal vector. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] Figure 1 It is a flow chart of a data registration method based on a multi-beam echo sounder in the first embodiment of the present invention;

[0071] Figure 2 It is a structural block diagram of a data registration system based on a multi-beam echo sounder in a second embodiment of the present invention;

[0072] The following specific implementation manner will further illustrate the present invention in conjunction with the above-mentioned drawings. DETAILED DESCRIPTION

[0073] In order to facilitate the understanding of the present invention, the present invention will be described more fully below with reference to the relevant drawings. Several embodiments of the present invention are given in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, the purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive.

[0074] It should be noted that when an element is referred to as being "fixed to" another element, it may be directly on the other element or there may be a central element. When an element is considered to be "connected to" another element, it may be directly connected to the other element or there may be a central element at the same time. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are for illustrative purposes only.

[0075] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which the present invention belongs. The terms used herein in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0076] Embodiment 1

[0077] See also Figure 1 , which is a flow chart of a data registration method based on a multi-beam echo sounder in a first embodiment of the present invention. As shown in the figure, the method includes the following steps:

[0078] Step S100, a detection coordinate system is established based on the measurement coordinates of the multi-beam echo sounder, and the observation point cloud data in the detection coordinate system is obtained, and the observation point cloud data is transformed into the reference coordinate system of the inertial navigation system based on the installation error angle between the inertial navigation system and the multi-beam echo sounder. Specifically, the above-mentioned detection coordinate system, that is, the coordinate system with the multi-beam echo sounder as the origin, has an installation deviation from the coordinate origin of the inertial navigation system. By converting the detection data into the detection coordinate system, the influence of the installation deviation can be eliminated and the measurement accuracy can be improved.

[0079] Furthermore, in this embodiment, the coordinate expression of the observation point cloud data transformed into the reference coordinate system of the inertial navigation system is:

[0080] ;

[0081] ;

[0082] ;

[0083] ;

[0084] ;

[0085] ;

[0086] In the formula, , , Transform the observed point cloud data into coordinates in the reference coordinate system, is the eccentricity between the reference coordinate system and the detection coordinate system, is the pitch installation error angle, To measure the roller installation error angle, is the yaw installation error angle, , , is the coordinate of the observed point cloud data in the detection coordinate system, is the transformation matrix between the detection coordinate system and the reference coordinate system.

[0087] Step S200, establishing a mobile coordinate system based on the odometer and a preset navigation trajectory, and transforming the observation point cloud data in the reference coordinate system to the mobile coordinate system based on the attitude data measured by the inertial navigation system.

[0088] Furthermore, in this embodiment, the coordinate expression for transforming the observation point cloud data in the reference coordinate system to the mobile coordinate system is:

[0089] ;

[0090] ;

[0091] ;

[0092] ;

[0093] ;

[0094] ;

[0095] In the formula, , , Transform the observed point cloud data into coordinates in the mobile coordinate system, is the eccentricity between the mobile coordinate system and the reference coordinate system, is the roll attitude angle, is the pitch attitude angle, is the yaw attitude angle, is the transformation matrix between the reference coordinate system and the moving coordinate system.

[0096] Specifically, the above-mentioned mobile coordinate system, that is, the coordinate system with the odometer as the origin, has an installation deviation from the coordinate origin of the inertial navigation system. By calibrating the observation point cloud data based on the attitude data of the inertial navigation system and then converting it to the mobile coordinate system, the attitude influence of the carrier can be eliminated and the measurement accuracy can be improved. At the same time, by converting the observation point cloud data to the mobile coordinate system, the observation point cloud data relative to the carrier can be accurately obtained. At the same time, based on the preset navigation trajectory, it can facilitate the alignment between each frame of data.

[0097] Step S300 , coarsely aligning the first observation point cloud data and the second observation point cloud data of two adjacent frames in the mobile coordinate system based on the ICP algorithm.

[0098] Specifically, in this embodiment, due to the accumulated errors of the sensor itself and the influence of environmental noise, it is necessary to align multiple data in the mobile coordinate system to obtain continuous and accurate point cloud data.

[0099] In this step, the above-mentioned coarse registration uses, for example, FPFH (Fast Point Feature Histograms), fast point feature histograms, or conventional ICP algorithms to achieve overall coarse registration between the two point cloud data.

[0100] Step S400, based on the waveform characteristics of the observation point cloud data, select a target point from the first observation point cloud data that has completed the rough registration, and calculate the curvature characteristics of the target point to screen out a feature point set whose curvature meets a threshold.

[0101] Specifically, in this embodiment, the step of calculating the curvature feature of the target point to screen out a feature point set whose curvature meets the threshold value specifically includes:

[0102] Step S410, calculating and obtaining the curvature value and the normal vector angle of the target point, combining the curvature value and the normal vector angle to obtain the curvature feature of the target point, and screening and obtaining a feature point set whose curvature meets a threshold based on the curvature feature;

[0103] The calculation expression of the curvature characteristic is as follows:

[0104] ;

[0105] Wherein, V is the curvature feature, W is the mean value of the normal vector angle of the target point, and J is the average curvature value of the target point.

[0106] Specifically, in this embodiment, the calculation expression of the average curvature value is as follows:

[0107] ;

[0108] In the formula, and is the first-order partial derivative of the surface corresponding to the target point, , and is the second-order partial derivative of the surface corresponding to the target point, J is the mean curvature value, and n is the unit normal vector of the surface corresponding to the target point.

[0109] In addition, in this embodiment, before the step of calculating the curvature value of the target point and the normal vector angle, the method further includes:

[0110] Based on the target point and the neighborhood points of the target point, local surface fitting is performed by least square method;

[0111] The calculation expression of the mean value of the normal vector angle is as follows:

[0112] ;

[0113] In the formula, z is the number of neighborhood points, is the normal vector of the target point, is the normal vector of the tth neighborhood point, and W is the mean angle of the normal vector.

[0114] Specifically, when the curvature of the target point in the local area is greater, the values ​​of the normal vector angle mean and the average curvature value are greater, and thus the value of the curvature feature becomes larger. In some application scenarios of this embodiment, by calculating the deformation factor of the target point and finding its mean and standard deviation, the threshold is the sum of the mean and the standard deviation, and the corresponding points whose curvature features are greater than the threshold are selected as feature points to obtain a feature point set whose curvature satisfies the threshold.

[0115] For ease of understanding, in the precise registration process of point clouds, it is usually necessary to find the nearest other points corresponding to each point, that is, the nearest neighboring points, and then use the point pairs to perform matrix transformation of the point cloud, so as to solve the optimal matrix change parameters through continuous iteration. The conventional ICP algorithm searches for correspondences by traversing all points in the point cloud. Under ideal conditions, it has high registration accuracy, but in the presence of noise, it is easy to have mismatched points; this application selects relevant feature points with obvious terrain features, so that the selected sampling points have richer terrain information. At the same time, the curvature is obtained based on the local surface composed of multiple neighborhood points, which is less affected by single-point noise and can improve the registration accuracy.

[0116] Step S500 , selecting points whose Euclidean distances with the feature point set are minimum from the second observation point cloud data to form a plurality of matching point pairs.

[0117] Specifically, after obtaining the feature point set from the first observation point cloud data, a KD tree is established based on the first observation point cloud data and the second observation point cloud data, so as to obtain corresponding matching points from the second observation point cloud data.

[0118] Step S600, select from the plurality of matching point pairs the feature matching point pairs whose Euclidean distance between the matching point pairs is less than a preset distance value and whose normal vector angle is less than a preset angle value. It can be understood that after coordinate transformation and rough alignment, setting the above preset distance value and preset angle value can filter out some mismatched pairs and improve matching accuracy.

[0119] Step S700, establish a correlation entropy model of the distance between the first observation point cloud data and the second observation point cloud data, establish an objective function about the transformation matrix based on the maximum value of the correlation entropy model as the target, and calculate the mean square error between each pair of feature matching points according to the objective function.

[0120] Specifically, in this embodiment, the calculation expression of the above objective function is:

[0121] ;

[0122] ;

[0123] In the formula, is the objective function, M is the correlation entropy model, N is the number of feature matching point pairs, R and T are the rotation matrix and translation matrix of the first observation point cloud data transformed to the second observation point cloud data, respectively. and are respectively the i-th feature matching point in the feature matching point pair corresponding to the first observation point cloud data and the second observation point cloud data, is the normal vector of the feature matching point corresponding to the second observation point cloud data, is the kernel length parameter during iteration.

[0124] For ease of understanding, the purpose of the correlation entropy model is to measure the similarity between random variables. The correlation entropy is used as part of the objective function to optimize the transformation parameters (such as rotation matrix and translation vector) between point clouds. By maximizing the correlation entropy, the similarity between the transformed point cloud and the target point cloud can be maximized, while reducing the impact of noise and outliers on the registration results.

[0125] Step S800: If the mean square error is less than a preset threshold, or the number of iterations reaches a preset number, the iteration is stopped to complete the precise registration of the first observation point cloud data and the second observation point cloud data.

[0126] Specifically, in some application scenarios of this embodiment, the registration iteration process is as follows:

[0127] 1. Initialization: Select an initial transformation parameter (such as a rotation matrix and a translation vector), which can be obtained through methods such as coarse registration.

[0128] 2. Establishing a corresponding relationship: Searching for matching points with the minimum Euclidean distance between each other and meeting the preset distance value and preset angle value in the feature point set and the second observation point cloud data to form a feature matching point pair;

[0129] 3. Calculate transformation parameters: Based on the transformed correspondence, calculate new transformation parameters to minimize the above mean square error, that is, to maximize the similarity corresponding to the correlation entropy, that is, the above objective function.

[0130] 3. Update transformation parameters: Apply the calculated new transformation parameters to the first observation point cloud data to update its position.

[0131] 4. Iteration: Repeat steps 2 to 3 until the convergence condition is met (such as the mean square error is less than the preset threshold or the number of iterations reaches the preset upper limit).

[0132] In summary, the data registration method and system based on the multi-beam echo sounder in the above-mentioned embodiments of the present invention, after obtaining the detection data of the detection coordinate system based on the multi-beam echo sounder, sequentially transforms the detection data to the reference coordinate system and the mobile coordinate system to eliminate the influence of the installation error and posture error of the multi-beam echo sounder and improve the measurement accuracy of the data. After synchronously transforming the data of two consecutive frames to the mobile coordinate system, the ICP algorithm is first used for coarse registration, and the data is screened to obtain a feature point set whose curvature meets the threshold, which can reduce the data processing amount and data redundancy, retain the relevant data points of valuable terrain features based on the waveform characteristics, and then introduce the correlation entropy as the error metric, combined with the iterative optimization method, to achieve high-precision data registration.

[0133] Embodiment 2

[0134] like Figure 2 As shown, the second embodiment of the present application provides a data registration system based on a multi-beam echo sounder, comprising:

[0135] A first coordinate transformation module 100 is used to establish a detection coordinate system based on the measurement coordinates of the multi-beam echo sounder, obtain observation point cloud data in the detection coordinate system, and transform the observation point cloud data into a reference coordinate system of the inertial navigation system based on an installation error angle between the inertial navigation system and the multi-beam echo sounder;

[0136] A second coordinate transformation module 200 is used to establish a mobile coordinate system based on the odometer and the preset navigation trajectory, and transform the observation point cloud data in the reference coordinate system into the mobile coordinate system based on the attitude data measured by the inertial navigation system;

[0137] A coarse registration module 300 is used to perform coarse registration of the first observation point cloud data and the second observation point cloud data of two adjacent frames in the mobile coordinate system based on an ICP algorithm;

[0138] The curvature module 400 is used to select a target point from the first observation point cloud data that has completed the rough registration based on the waveform characteristics of the observation point cloud data, and perform curvature characteristic calculation on the target point to screen out a feature point set whose curvature meets a threshold value;

[0139] A first matching module 500 is used to select points whose Euclidean distances with the feature point set are minimum from the second observation point cloud data to form a plurality of matching point pairs;

[0140] The second matching module 600 is used to select, from the plurality of matching point pairs, feature matching point pairs whose Euclidean distance between the matching point pairs is less than a preset distance value and whose normal vector angle is less than a preset angle value;

[0141] A fine registration module 700 is used to establish a correlation entropy model of the distance between the first observation point cloud data and the second observation point cloud data, establish an objective function about the transformation matrix based on the maximum value of the correlation entropy model as a target, and calculate the mean square error between each of the feature matching point pairs according to the objective function;

[0142] The iteration module 800 is used to stop iteration if the mean square error is less than a preset threshold or the number of iterations reaches a preset number, so as to complete the precise registration of the first observation point cloud data and the second observation point cloud data.

[0143] Preferably, in this embodiment, the curvature module 400 is specifically used for:

[0144] Calculate the curvature value and the normal vector angle of the target point, combine the curvature value and the normal vector angle to obtain the curvature feature of the target point, and screen the feature point set whose curvature meets the threshold based on the curvature feature;

[0145] The calculation expression of the curvature characteristic is as follows:

[0146] ;

[0147] Wherein, V is the curvature feature, W is the mean value of the normal vector angle of the target point, and J is the average curvature value of the target point.

[0148] Preferably, in this embodiment, the curvature module 400 is further used to: perform local surface fitting based on the target point and the neighborhood points of the target point by using the least square method;

[0149] The calculation expression of the mean value of the normal vector angle is as follows:

[0150] ;

[0151] In the formula, z is the number of neighborhood points, is the normal vector of the target point, is the normal vector of the tth neighborhood point, and W is the mean angle of the normal vector.

[0152] The technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0153] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A data registration method based on a multi-beam echo sounder, characterized in that: The following steps are involved: A detection coordinate system is established based on the measurement coordinates of the multi-beam echo sounder, and observation point cloud data in the detection coordinate system is obtained, and based on the installation error angle between the inertial navigation system and the multi-beam echo sounder, the observation point cloud data is transformed into a reference coordinate system of the inertial navigation system; Establishing a mobile coordinate system based on the odometer and the preset navigation trajectory, and transforming the observation point cloud data in the reference coordinate system to the mobile coordinate system based on the attitude data measured by the inertial navigation system; Based on the ICP algorithm, the first observation point cloud data and the second observation point cloud data of two adjacent frames in the mobile coordinate system are roughly aligned; Based on the waveform characteristics of the observation point cloud data, a target point is selected from the first observation point cloud data that has completed the rough registration, and a curvature characteristic calculation is performed on the target point to screen out a feature point set whose curvature meets a threshold value; Selecting points whose Euclidean distances with the feature point set are the minimum from the second observation point cloud data to form a plurality of matching point pairs; Selecting, from the plurality of matching point pairs, feature matching point pairs whose Euclidean distance between the matching point pairs is less than a preset distance value and whose normal vector angle is less than a preset angle value; Establishing a correlation entropy model of the distance between the first observation point cloud data and the second observation point cloud data, establishing an objective function about the transformation matrix based on the maximum value of the correlation entropy model as a target, and calculating the mean square error between each of the feature matching point pairs according to the objective function; If the mean square error is less than a preset threshold, or the number of iterations reaches a preset number, the iteration is stopped to complete the precise registration of the first observation point cloud data and the second observation point cloud data.

2. The data registration method based on multi-beam echo sounder according to claim 1, characterized in that: The step of calculating the curvature feature of the target point to screen out a feature point set whose curvature meets a threshold specifically includes: Calculate the curvature value and the normal vector angle of the target point, combine the curvature value and the normal vector angle to obtain the curvature feature of the target point, and screen the feature point set whose curvature meets the threshold based on the curvature feature; The calculation expression of the curvature characteristic is as follows: ； Wherein, V is the curvature feature, W is the mean value of the normal vector angle of the target point, and J is the average curvature value of the target point.

3. The data registration method based on multi-beam echo sounder according to claim 2, characterized in that: The calculation expression of the mean curvature value is as follows: ； In the formula, and is the first-order partial derivative of the surface corresponding to the target point, , and is the second-order partial derivative of the surface corresponding to the target point, J is the mean curvature value, and n is the unit normal vector of the surface corresponding to the target point.

4. The data registration method based on multi-beam echo sounder according to claim 2, characterized in that: Before the step of calculating the curvature value of the target point and the normal vector angle, the method further includes: Based on the target point and the neighborhood points of the target point, local surface fitting is performed by least square method; The calculation expression of the mean value of the normal vector angle is as follows: ； In the formula, z is the number of neighborhood points, is the normal vector of the target point, is the normal vector of the tth neighborhood point, and W is the mean angle of the normal vector.

5. The data registration method based on multi-beam echo sounder according to claim 1, characterized in that: The calculation expression of the objective function is: ； ； In the formula, is the objective function, M is the correlation entropy model, N is the number of feature matching point pairs, R and T are the rotation matrix and translation matrix of the first observation point cloud data transformed to the second observation point cloud data, respectively. and are respectively the i-th feature matching point in the feature matching point pair corresponding to the first observation point cloud data and the second observation point cloud data, is the normal vector of the feature matching point corresponding to the second observation point cloud data, is the kernel length parameter during iteration.

6. The data registration method based on multi-beam echo sounder according to claim 5, characterized in that: The calculation expression of the mean square error is: ； In the formula, is the mean square error, and k is the number of iterations.

7. The data registration method based on multi-beam echo sounder according to claim 1, characterized in that: The coordinate expression of the observation point cloud data transformed into the reference coordinate system of the inertial navigation system is: ； ； ； ； ； ； In the formula, , , Transform the observed point cloud data into coordinates in the reference coordinate system, is the offset between the reference coordinate system and the detection coordinate system, is the pitch installation error angle, To measure the roller installation error angle, is the yaw installation error angle, , , is the coordinate of the observed point cloud data in the detection coordinate system, is the transformation matrix between the detection coordinate system and the reference coordinate system.

8. The data registration method based on multi-beam echo sounder according to claim 7, characterized in that: The coordinate expression for transforming the observation point cloud data in the reference coordinate system to the mobile coordinate system is: ； ； ； ； ； ； In the formula, , , Transform the observed point cloud data into coordinates in the mobile coordinate system, is the offset between the mobile coordinate system and the reference coordinate system, is the roll attitude angle, is the pitch attitude angle, is the yaw attitude angle, is the transformation matrix between the reference coordinate system and the moving coordinate system.

9. A data registration system based on a multi-beam echo sounder, characterized in that: include: A first coordinate transformation module is used to establish a detection coordinate system based on the measurement coordinates of the multi-beam echo sounder, obtain observation point cloud data in the detection coordinate system, and transform the observation point cloud data into a reference coordinate system of the inertial navigation system based on an installation error angle between the inertial navigation system and the multi-beam echo sounder; A second coordinate transformation module is used to establish a mobile coordinate system based on the odometer and the preset navigation trajectory, and transform the observation point cloud data in the reference coordinate system into the mobile coordinate system based on the attitude data measured by the inertial navigation system; A coarse registration module, used for coarsely registering the first observation point cloud data and the second observation point cloud data of two adjacent frames in the mobile coordinate system based on an ICP algorithm; A curvature module, for selecting a target point from the first observation point cloud data that has completed the rough registration based on the waveform characteristics of the observation point cloud data, and performing curvature characteristic calculation on the target point to screen out a feature point set whose curvature meets a threshold value; A first matching module is used to select points whose Euclidean distances with the feature point set are minimum from the second observation point cloud data to form a plurality of matching point pairs; A second matching module is used to select, from the plurality of matching point pairs, feature matching point pairs whose Euclidean distance between the matching point pairs is less than a preset distance value and whose normal vector angle is less than a preset angle value; A fine registration module, used to establish a correlation entropy model of the distance between the first observation point cloud data and the second observation point cloud data, establish an objective function about the transformation matrix based on the maximum value of the correlation entropy model as a target, and calculate the mean square error between each of the feature matching point pairs according to the objective function; The iteration module is used to stop iteration if the mean square error is less than a preset threshold or the number of iterations reaches a preset number, so as to complete the precise alignment of the first observation point cloud data and the second observation point cloud data.

10. The data registration system based on multi-beam echo sounder according to claim 9, characterized in that: The curvature module is specifically used for: Calculate the curvature value and the normal vector angle of the target point, combine the curvature value and the normal vector angle to obtain the curvature feature of the target point, and screen the feature point set whose curvature meets the threshold based on the curvature feature; The calculation expression of the curvature characteristic is as follows: ； Wherein, V is the curvature feature, W is the mean value of the normal vector angle of the target point, and J is the average curvature value of the target point.

Images