An unmanned ship target identification method and device based on formation position intelligent transformation

Abstract

This invention provides a method and apparatus for unmanned surface vessels (USVs) to identify targets based on intelligent formation position transformation. The method includes: obtaining the attitude angle, distance information, environmental information, and image information of the target to be identified; constructing a recognition classifier model; training the recognition classifier model using a sample set of attitude angle, distance information, environmental information, and image information; using the attitude angle, distance information, environmental information, and image information of the target to be identified as inputs to the trained recognition classifier model to obtain a classification result vector; obtaining the optimal attitude angle of the target to be identified based on the classification result vector, then calculating the formation position of the USV formation; intelligently transforming the formation position of the USV crew members; and finally obtaining the recognition result of the target to be identified through the trained recognition classifier model. By using the optimal attitude angle and dynamically transforming the position along the shortest path, the method achieves the technical effect of optimizing the formation cooperative perception strategy and effectively improving the target recognition accuracy.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of automatic identification of marine mobile targets, and particularly relates to an unmanned ship target identification method and device based on formation position intelligent transformation. BACKGROUND

[0002] At present, the formation target cooperative perception method is to comprehensively utilize multi-sensor of the formation target for comprehensive identification, to obtain an identification strategy superior to single target perception, and the comprehensive identification can be roughly divided into two categories, one is a statistical experience strategy-based identification method, and the other is a deep learning-based identification method. Compared with the statistical experience strategy-based identification method, the deep learning-based identification method has higher accuracy and application prospect, and has gradually become a research direction.

[0003] However, the present application inventors found at least the following technical problems in the process of implementing the technical solutions of the embodiments of the present application:

[0004] In the existing method, the photoelectric, motion, environment and other characteristics of the unmanned ship target to be identified are not input as a whole event, and the comprehensive identification performance of different formation position of the unmanned ship is not considered, so that the formation position of the unmanned ship cannot be intelligently adjusted according to the multi-sensor perception information of the unmanned ship, resulting in the technical problem of low accuracy of marine mobile target identification of the unmanned ship. SUMMARY

[0005] The embodiments of the present application provide an unmanned ship target identification method and device based on formation position intelligent transformation, which solves the technical problem in the existing method that the photoelectric, motion, environment and other characteristics of the unmanned ship target to be identified are not input as a whole event, and the comprehensive identification performance of different formation position of the unmanned ship is not considered, so that the formation position of the unmanned ship cannot be intelligently adjusted according to the multi-sensor perception information of the unmanned ship, resulting in low accuracy of marine mobile target identification of the unmanned ship. Based on the different comprehensive identification performance of different formation position of the unmanned ship, the multi-sensor information of the formation of the unmanned ship is comprehensively utilized to obtain the best attitude angle of the target to be identified, and the position is dynamically transformed in real time with the shortest path, so as to achieve the technical effects of guaranteeing timeliness, optimizing formation cooperative perception strategy and effectively improving target identification accuracy.

[0006] In view of the above problems, the embodiments of the present application are proposed to provide an unmanned ship target identification method and device based on formation position intelligent transformation.

[0007] In a first aspect, the present application provides a method for identifying a target by an unmanned ship based on formation position intelligent transformation, the method comprising: obtaining attitude angle and distance information of a target to be identified by a radar sensor; obtaining environment information of the target to be identified; obtaining image information of the target to be identified by an optical sensor, wherein the image information is in one-to-one correspondence with the attitude angle, the distance information and the environment information; constructing an identification classifier model; obtaining an attitude angle, distance information, environment information and image information sample set; training the identification classifier model by using the attitude angle, distance information, environment information and image information sample set to obtain a trained identification classifier model; taking the attitude angle, distance information, environment information and image information of the target to be identified as input items of the trained identification classifier model to obtain a classification result vector; obtaining a best attitude angle of the target to be identified according to the classification result vector; calculating a formation position of an unmanned ship formation according to the best attitude angle; intelligently and dynamically transforming the formation position of the unmanned ship members according to the formation position; and obtaining an identification result of the target to be identified by the trained identification classifier model after the formation position of the unmanned ship members is transformed.

[0008] Preferably, the obtaining of the attitude angle and distance information of the target to be identified by the radar sensor comprises: obtaining a movement direction of the target to be identified by the radar sensor; obtaining position information of the target to be identified and a first member of the unmanned ship formation; obtaining a line direction of the target to be identified and the first member of the unmanned ship formation according to the position information of the target to be identified and the first member of the unmanned ship formation; and calculating an included angle between the movement direction and the line direction to obtain the attitude angle of the target to be identified.

[0009] Preferably, the obtaining of the movement direction of the target to be identified comprises: obtaining first position information of the target to be identified by the radar sensor; obtaining second position information of the target to be identified by the radar sensor; and obtaining the movement direction of the target to be identified according to the first position information and the second position information.

[0010] Preferably, the obtaining of the environment information of the target to be identified comprises: obtaining intelligence information by an intelligence information receiver; and obtaining the environment information of the target to be identified according to the intelligence information, wherein the environment information comprises illumination, ocean wind power and sea state information.

[0011] Preferably, the constructed recognition classifier model comprises: a convolution pooling layer configured to perform convolution pooling processing on a plurality of parallel target image information to be recognized to form a first one-dimensional vector, wherein the number of the target image information to be recognized is equal to the number of members of the unmanned ship formation; a binary coding layer configured to convert attitude angle, distance information and environment information of the target to be recognized into a binary array to form a second one-dimensional vector; a full connection layer, input nodes of the full connection layer being composed of the first one-dimensional vector and the second one-dimensional vector in series; and an output node layer, the number of nodes of the output node layer being equal to the number of target types of the target to be recognized.

[0012] Preferably, the method for obtaining the optimal attitude angle of the target to be recognized according to the classification result vector comprises: performing normalization processing on the classification result vector to obtain a plurality of normalized classification result vectors; obtaining a target type corresponding to a maximum value element in each normalized classification result vector, i.e., obtaining a target type of the target to be recognized; obtaining a target type optimal solution vector of the target to be recognized according to the target type of the target to be recognized, wherein a vector element corresponding to the target type is set to 1, and vector elements corresponding to other target types are set to 0; calculating a Euclidean distance between the target type optimal solution vector and each normalized classification result vector; calculating a minimum value of a plurality of Euclidean distances; and obtaining the optimal attitude angle of the target to be recognized according to a classification vector result of the minimum Euclidean distance, wherein the optimal attitude angle and the classification vector result of the minimum Euclidean distance are in a one-to-one correspondence.

[0013] Preferably, the method for calculating the formation position of the unmanned ship formation according to the optimal attitude angle comprises: obtaining a current attitude angle of a first member of the unmanned ship formation and the target to be recognized; calculating an angle difference between the current attitude angle and the optimal attitude angle; calculating a perpendicular point direction and a perpendicular point distance of the first member of the unmanned ship formation to the direction of the optimal attitude angle according to the angle difference and the distance information; performing position transformation of the first member of the unmanned ship formation according to the perpendicular point direction and the perpendicular point distance; calculating the perpendicular point direction and the perpendicular point distance of all members of the unmanned ship formation; and obtaining the formation position of the unmanned ship formation according to the perpendicular point direction and the perpendicular point distance of all members of the unmanned ship formation.

[0014] In a second aspect, the present application provides an unmanned ship target recognition device based on formation position intelligent transformation, which comprises:

[0015] A first obtaining unit is configured to obtain an attitude angle and distance information of a target to be recognized through a radar sensor.

[0016] a second obtaining unit, configured to obtain environment information of the target to be identified;

[0017] a third obtaining unit, configured to obtain image information of the target to be identified by an optical sensor, wherein the image information is in one-to-one correspondence with the attitude angle, the distance information, and the environment information;

[0018] a first constructing unit, configured to construct an identification classifier model;

[0019] a fourth obtaining unit, configured to obtain a sample set of attitude angles, distance information, environment information, and image information;

[0020] a first training unit, configured to train the identification classifier model by using the sample set of attitude angles, distance information, environment information, and image information, to obtain a trained identification classifier model;

[0021] a fifth obtaining unit, configured to take the attitude angle, the distance information, the environment information, and the image information of the target to be identified as input items of the trained identification classifier model, to obtain a classification result vector;

[0022] a sixth obtaining unit, configured to obtain an optimal attitude angle of the target to be identified according to the classification result vector;

[0023] a first calculating unit, configured to calculate a formation position of an unmanned ship formation according to the optimal attitude angle;

[0024] a first transforming unit, configured to intelligently and dynamically transform the formation position of the unmanned ship member according to the formation position;

[0025] a seventh obtaining unit, configured to obtain an identification result of the target to be identified by the trained identification classifier model after the formation position of the unmanned ship member is transformed.

[0026] Preferably, the first obtaining unit comprises:

[0027] an eighth obtaining unit, configured to obtain a moving direction of the target to be identified by a radar sensor;

[0028] a ninth obtaining unit, configured to obtain position information of the target to be identified and a first member of the unmanned ship formation;

[0029] a tenth obtaining unit, configured to obtain a line direction between the to-be-identified target and a first member of the unmanned ship formation according to position information of the to-be-identified target and the first member of the unmanned ship formation;

[0030] a second calculating unit, configured to calculate an included angle between the motion direction and the line direction to obtain a pose angle of the to-be-identified target.

[0031] Preferably, the eighth obtaining unit comprises:

[0032] an eleventh obtaining unit, configured to obtain first position information of the to-be-identified target by the radar sensor;

[0033] a twelfth obtaining unit, configured to obtain second position information of the to-be-identified target by the radar sensor;

[0034] a thirteenth obtaining unit, configured to obtain a motion direction of the to-be-identified target according to the first position information and the second position information.

[0035] Preferably, the second obtaining unit comprises:

[0036] a fourteenth obtaining unit, configured to obtain intelligence information by an intelligence information receiver;

[0037] a fifteenth obtaining unit, configured to obtain environment information of the to-be-identified target according to the intelligence information, wherein the environment information comprises illumination, ocean wind power, and sea state information.

[0038] Preferably, the first constructing unit comprises:

[0039] a convolution pooling layer, configured to perform convolution pooling processing on a plurality of parallel to-be-identified target image information to form a first one-dimensional vector, wherein the number of the to-be-identified target image information is equal to the number of members of the unmanned ship formation;

[0040] a binary coding layer, configured to convert the pose angle, distance information, and environment information of the to-be-identified target into a binary array to form a second one-dimensional vector;

[0041] a full connection layer, an input node of the full connection layer being composed of a concatenation of the first one-dimensional vector and the second one-dimensional vector;

[0042] an output node layer, a number of nodes of the output node layer being equal to a number of target types of the to-be-identified target.

[0043] Preferably, the sixth obtaining unit comprises:

[0044] a sixteenth obtaining unit, configured to normalize the classification result vector to obtain a plurality of normalized classification result vectors;

[0045] a seventeenth obtaining unit, configured to obtain a target type corresponding to a maximum value element in a vector according to each normalized classification result vector, i.e., to obtain a target type of the target to be recognized;

[0046] an eighteenth obtaining unit, configured to obtain a target type optimal solution vector of the target to be recognized according to the target type of the target to be recognized, wherein a vector element corresponding to the target type is set to 1, and vector elements corresponding to other target types are set to 0;

[0047] a third calculating unit, configured to calculate an Euclidean distance between the target type optimal solution vector and each normalized classification result vector;

[0048] a fourth calculating unit, configured to calculate a minimum value of a plurality of Euclidean distances;

[0049] a nineteenth obtaining unit, configured to obtain a best attitude angle of the target to be recognized according to a classification vector result of the minimum Euclidean distance, wherein the best attitude angle and the classification vector result of the minimum Euclidean distance are in a one-to-one correspondence.

[0050] Preferably, the first calculating unit comprises:

[0051] a twentieth obtaining unit, configured to obtain a current attitude angle of the first member of the unmanned ship formation and the target to be recognized;

[0052] a fifth calculating unit, configured to calculate an angle difference between the current attitude angle and the best attitude angle;

[0053] a sixth calculating unit, configured to calculate a perpendicular point direction and a perpendicular point distance of the first member of the unmanned ship formation to a direction of the best attitude angle according to the angle difference and the distance information;

[0054] a second transforming unit, configured to perform position transformation on the first member of the unmanned ship formation according to the perpendicular point direction and the perpendicular point distance;

[0055] a seventh calculating unit, configured to calculate the perpendicular point direction and the perpendicular point distance of all members of the unmanned ship formation;

[0056] A twenty-first obtaining unit is configured to obtain the formation position of the unmanned ship formation according to the plumb point direction and the plumb point distance of all members of the unmanned ship formation.

[0057] In a third aspect, the present application provides an unmanned ship target recognition device based on formation position intelligent transformation, which comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the following steps when executing the program:

[0058] Obtain the attitude angle and distance information of the target to be recognized through a radar sensor; obtain the environment information of the target to be recognized; obtain the image information of the target to be recognized through an optical sensor, wherein the image information, the attitude angle, the distance information, and the environment information are in one-to-one correspondence; construct an identification classifier model; obtain the attitude angle, distance information, environment information, and image information sample set; train the identification classifier model using the attitude angle, distance information, environment information, and image information sample set to obtain a trained identification classifier model; input the attitude angle, distance information, environment information, and image information of the target to be recognized into the trained identification classifier model as input items to obtain a classification result vector; obtain the best attitude angle of the target to be recognized according to the classification result vector; calculate the formation position of the unmanned ship formation according to the best attitude angle; intelligently and dynamically transform the formation position of the unmanned ship members according to the formation position; and obtain the recognition result of the target to be recognized through the trained identification classifier model after the formation position of the unmanned ship members is transformed.

[0059] In a fourth aspect, the present application provides a computer readable storage medium having a computer program stored thereon, and the program is executed by a processor to implement the following steps:

[0060] The attitude angle and distance information of a target to be identified are obtained through a radar sensor; environment information of the target to be identified is obtained; image information of the target to be identified is obtained through an optical sensor, wherein the image information is in one-to-one correspondence with the attitude angle, the distance information and the environment information; a recognition classifier model is constructed; an attitude angle, distance information, environment information and image information sample set is obtained; the recognition classifier model is trained by using the attitude angle, distance information, environment information and image information sample set, to obtain a trained recognition classifier model; the attitude angle, distance information, environment information and image information of the target to be identified are taken as input items of the trained recognition classifier model, to obtain a classification result vector; the best attitude angle of the target to be identified is obtained according to the classification result vector; the formation position of the unmanned ship formation is calculated according to the best attitude angle; the formation formation position of the unmanned ship members is intelligently and dynamically changed according to the formation position; and the recognition result of the target to be identified is obtained through the trained recognition classifier model after the formation formation position of the unmanned ship members is changed.

[0061] The one or more technical solutions in the embodiments of the present application have at least one or more of the following technical effects:

[0062] The embodiment of the present application provides a target recognition method and device for unmanned ship based on formation position intelligent transformation, the method comprises the following steps: obtaining the attitude angle and distance information of a target to be recognized through a radar sensor; obtaining the environment information of the target to be recognized; obtaining the image information of the target to be recognized through an optical sensor, wherein the image information, the attitude angle, the distance information and the environment information are in one-to-one correspondence; constructing an identification classifier model; obtaining the attitude angle, distance information, environment information and image information sample set; training the identification classifier model by using the attitude angle, distance information, environment information and image information sample set to obtain a trained identification classifier model; taking the attitude angle, distance information, environment information and image information of the target to be recognized as input items of the trained identification classifier model to obtain a classification result vector; obtaining the best attitude angle of the target to be recognized according to the classification result vector; calculating the formation position of the unmanned ship formation according to the best attitude angle; intelligently and dynamically transforming the formation position of the unmanned ship member according to the formation position; obtaining the recognition result of the target to be recognized through the trained identification classifier model after the formation position of the unmanned ship member is transformed. The technical problem that the recognition accuracy of the unmanned ship marine mobile target is low in the prior art is solved, because the photoelectric, motion and environment characteristics of the target to be recognized of the unmanned ship are not input as a whole event, and the comprehensive recognition performance of different formation positions of the unmanned ship is not considered, so that the formation position of the unmanned ship formation cannot be intelligently adjusted according to the multi-sensor perception information of the unmanned ship, and the technical effects of guaranteeing timeliness, optimizing formation collaborative perception strategy and effectively improving the target recognition accuracy are achieved.

[0063] The above description is only a summary of the technical scheme of the present application, in order to more clearly understand the technical means of the present application, the content of the specification can be implemented, and in order to make the above and other purposes, characteristics and advantages of the present application more obvious and easy to understand, the specific embodiments of the present application are described below. BRIEF DESCRIPTION OF DRAWINGS

[0064] Figure 1 It is a flowchart of the target recognition method for unmanned ship based on formation position intelligent transformation in the embodiment of the present application;

[0065] Figure 2 It is a structure diagram of the target recognition device for unmanned ship based on formation position intelligent transformation in the embodiment of the present application;

[0066] Figure 3A structure schematic diagram of another unmanned ship target recognition device based on formation position intelligent transformation in the embodiment of the application;

[0067] Figure 4 A structure schematic diagram of the identification classifier model in the embodiment of the application;

[0068] Figure 5 A schematic diagram of the attitude angle in the embodiment of the application.

[0069] Label explanation: first obtaining unit 1, second obtaining unit 2, third obtaining unit 3, first constructing unit 4, fourth obtaining unit 5, first training unit 6, fifth obtaining unit 7, sixth obtaining unit 8, first calculating unit 9, first transforming unit 10, seventh obtaining unit 11, bus 300, receiver 301, processor 302, transmitter 303, memory 304, bus interface 306. DETAILED DESCRIPTION

[0070] The embodiment of the application provides a kind of unmanned ship target recognition method and device based on formation position intelligent transformation, solve the technical problem that in existing method, because the photoelectric, motion, environment etc. Characteristic of the target to be identified of unmanned ship is not input as a whole event, and the comprehensive recognition performance of different formation position of unmanned ship is not considered, so as to be unable to adjust the formation position of unmanned ship according to the multi-sensor perception information of unmanned ship, leading to the low accuracy of the sea mobile target recognition of unmanned ship.

[0071] The technical scheme provided by the present application is as follows: obtaining the attitude angle and distance information of a to-be-identified target through a radar sensor; obtaining the environment information of the to-be-identified target; obtaining the image information of the to-be-identified target through an optical sensor, wherein the image information is in one-to-one correspondence with the attitude angle, the distance information and the environment information; constructing an identification classifier model; obtaining an attitude angle, distance information, environment information and image information sample set; training the identification classifier model using the attitude angle, distance information, environment information and image information sample set to obtain a trained identification classifier model; taking the attitude angle, distance information, environment information and image information of the to-be-identified target as input items of the trained identification classifier model to obtain a classification result vector; obtaining the best attitude angle of the to-be-identified target according to the classification result vector; calculating the formation position of the unmanned ship formation according to the best attitude angle; intelligently and dynamically changing the formation position of the unmanned ship members according to the formation position; and obtaining the identification result of the to-be-identified target through the trained identification classifier model after the formation position of the unmanned ship members is changed. The comprehensive identification performance of the unmanned ship at different formation positions is different, the best attitude angle of the to-be-identified target is obtained by comprehensively utilizing the multi-sensor information of the unmanned ship formation, the position is dynamically changed in real time in the shortest path, so that the technical effects of guaranteeing timeliness, optimizing the formation collaborative perception strategy and effectively improving the target identification accuracy are achieved.

[0072] The technical scheme of the present application will be described in detail below with reference to the drawings and specific embodiments. It should be understood that the embodiments and specific features in the embodiments of the present application are detailed descriptions of the technical scheme of the present application, and are not limitations of the technical scheme of the present application. In the case of no conflict, the technical features in the embodiments and the embodiments of the present application can be combined with each other.

[0073] The term "and / or" in this paper is only a description of the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B, which can represent the three cases of A alone, A and B together, and B alone. In addition, the character " / " in this paper generally represents that the front and rear associated objects are in an "or" relationship.

[0074] Embodiment one

[0075] Figure 1 A flowchart of a method for identifying a target of an unmanned ship based on intelligent transformation of formation position in an embodiment of the present application. As shown in Figure 1 The method comprises the following steps:

[0076] Step S1: obtaining the attitude angle and distance information of a to-be-identified target through a radar sensor;

[0077] Specifically, in the embodiments of the present application, the definition of the attitude angle of the target to be identified is the included angle between the moving direction of the target to be identified and the direction of the unmanned ship observing the target to be identified. In short, the direction of the unmanned ship observing the target to be identified is the direction of the line between the unmanned ship and the target to be identified, so that the attitude angle of each member in the unmanned ship formation observing the target to be identified is different. If the number of members in the unmanned ship formation is n, then n attitude angle information of the same target to be identified observed by the members of the unmanned ship formation at a certain moment will be generated. The distance information of the target to be identified is the current distance information between the target to be identified and a certain member of the unmanned ship formation, and the attitude angle and distance information of the target to be identified can be obtained in real time through the radar sensor loaded on the unmanned ship.

[0078] In step S1, the method for obtaining the attitude angle of the target to be identified is described in detail. First, the moving direction of the target to be identified is obtained through the radar sensor on the unmanned ship, and then the current position information of the target to be identified and any member of the unmanned ship formation is obtained, so as to obtain the direction of the line between the target to be identified and any member of the unmanned ship formation. Finally, the included angle between the moving direction and the line direction is calculated to obtain the attitude angle of the target to be identified.

[0079] Further, the method for obtaining the moving direction of the target to be identified comprises: obtaining first position information of the target to be identified through the radar sensor; obtaining second position information of the target to be identified through the radar sensor; and obtaining the moving direction of the target to be identified according to the first position information and the second position information.

[0080] Specifically, the embodiments of the present application obtain the position information of the target to be identified at two different times, so as to achieve the technical effect of calculating the moving direction of the target to be identified. Specifically, the first position information and the second position information of the target to be identified are also monitored by the radar sensor loaded on the unmanned ship, wherein the first position information is the current time position information of the target to be identified, and the second position information is the next time position information of the target to be identified. The line direction between the current time position information and the next time position information is the moving direction of the target to be identified.

[0081] Step S2: obtaining the environment information of the target to be identified;

[0082] Further, the method for obtaining the environment information of the target to be identified comprises: obtaining intelligence information through an intelligence information receiver; and obtaining the environment information of the target to be identified according to the intelligence information, wherein the environment information includes illumination, ocean wind power, and sea state information.

[0083] Specifically, in the process of identifying the maneuvering target by the unmanned ship, in addition to the factors such as the attitude angle and the target image information, the environment in which the target to be identified is located also has a great influence on the cooperative perception recognition rate, wherein the environment information includes light, ocean wind, sea state and other information. An intelligence information receiver is arranged on the unmanned ship, and various environment information of the target to be identified is received through the intelligence information receiver, so as to achieve the technical effect of increasing the dimension of input information.

[0084] Step S3: obtaining image information of the target to be identified through an optical sensor, wherein the image information has a one-to-one correspondence with the attitude angle, the distance information and the environment information;

[0085] Specifically, the image information of the target to be identified is obtained by the optical sensor loaded on each member of the unmanned ship formation. Each formation member will obtain the image information of the target to be identified at a certain moment. Due to the different attitude angles, distance information and environment information, the image information of the target to be identified obtained by the formation members is also different, so that the image information has a one-to-one correspondence with the attitude angle, the distance information and the environment information. According to the attitude angle, the distance information and the environment information, the image information is numbered, and the attitude angle, the distance information, the environment information and the image information are associated to form multi-source heterogeneous information of the target to be identified, which is used as an input item of a deep learning classifier model, thereby further achieving the technical effects of real-time acquisition of multi-feature information of the target to be identified and effective improvement of cooperative perception recognition accuracy of the formation.

[0086] Step S4: constructing an identification classifier model;

[0087] Specifically, in step S4, as shown in the figure, Figure 4 the construction of the identification classifier model includes: a convolution pooling layer, which is used for convolution pooling processing on a plurality of parallel target to be identified image information to form a first one-dimensional vector, wherein the number of target to be identified image information is equal to the number of members of the unmanned ship formation; a binary encoding layer, which is used for converting the attitude angle, distance information and environment information of the target to be identified into a binary array to form a second one-dimensional vector; a full connection layer, the input nodes of which are composed of the first one-dimensional vector and the second one-dimensional vector in series; and an output node layer, the number of nodes of which is equal to the number of target types of the target to be identified.

[0088] Step S5: obtaining a sample set of attitude angle, distance information, environment information and image information;

[0089] Step S6: training the recognition classifier model by using the attitude angle, distance information, environment information and image information sample set, to obtain a trained recognition classifier model;

[0090] Specifically, after the recognition classifier model is constructed, a large amount of historical data set of the attitude angle, distance information, environment information and image information of the to-be-recognized target needs to be collected, so as to iteratively train the recognition classifier model until the recognition classifier model converges. Further, the technical effect of training the recognition classifier model is achieved.

[0091] Step S7: taking the attitude angle, distance information, environment information and image information of the to-be-recognized target as input items of the trained recognition classifier model, to obtain a classification result vector;

[0092] Specifically, the attitude angle, distance information, environment information and image information of the to-be-recognized target are input into the trained recognition classifier model, and after deep learning, the classification result vector of the to-be-recognized target is obtained. Due to the complexity of the actual situation, the classification result vector is not the optimal solution vector in theory, and the values on each target type node are between 0 and 1.

[0093] Step S8: obtaining the optimal attitude angle of the to-be-recognized target according to the classification result vector;

[0094] Further, the optimal attitude angle of the to-be-recognized target is obtained according to the classification result vector, including: normalizing the classification result vector to obtain a plurality of normalized classification result vectors; obtaining the target type corresponding to the maximum value element in the vector according to each normalized classification result vector, that is, obtaining the target type of the to-be-recognized target; obtaining the optimal solution vector of the target type of the to-be-recognized target, wherein the vector element corresponding to the target type is set to 1, and the vector element corresponding to other target types is set to 0; calculating the Euclidean distance between the optimal solution vector of the target type and each normalized classification result vector; calculating the minimum value of a plurality of Euclidean distances; obtaining the optimal attitude angle of the to-be-recognized target according to the classification vector result of the minimum Euclidean distance, wherein the optimal attitude angle and the classification vector result of the minimum Euclidean distance are in one-to-one correspondence.

[0095] Specifically, first, the classification result vector is normalized to obtain a plurality of normalized classification result vectors. It is assumed that the value on the i-th node is x i The normalization formula is:

[0096]

[0097] wherein n represents the number of types of the target to be identified, and then the target type corresponding to the maximum element in each of the normalized classification result vectors is calculated as the target type of the target to be identified. For example, the target type of the target to be identified can be a yacht, a motorboat, a fishing boat, a cruise ship, a pirate ship, a patrol boat, etc., if the classification result vector output by the identification classifier model is [0.99, 0.11, 0.03, 0.25, 0.05, 0.16], the maximum element in the vector is 0.99, and the target type corresponding to the first output node is a yacht, then the target type of the target to be identified is a yacht. Then, according to the target type of the target to be identified, a target type optimal solution vector is obtained. In the embodiment of the present application, the definition of the optimal solution vector is that in the most ideal case, the value of the correct solution on the output node of the identification classifier model should be 1, and the value of the incorrect solution should be 0. In short, the optimal solution vector should be a one-dimensional vector composed of multiple 0s and one 1, and the optimal solution vector of the aforementioned target to be identified should be [1, 0, 0, 0, 0, 0]. Then, after calculating the Euclidean distance between the target type optimal solution vector and each of the normalized classification result vectors, a minimum value of the Euclidean distance is obtained, and finally, according to the classification vector result of the minimum Euclidean distance, the best attitude angle of the target to be identified is obtained, that is, the attitude angle in each of the classification result vectors of the minimum Euclidean distance is the best attitude angle of the target to be identified, wherein the best attitude angle is the attitude angle of the unmanned ship formation member when the cooperative perception target recognition rate is the highest.

[0098] Step S9: calculating the formation position of the unmanned ship formation according to the best attitude angle;

[0099] Further, the calculation of the formation position of the unmanned ship formation according to the best attitude angle comprises: obtaining the current attitude angle of the first member of the unmanned ship formation and the target to be identified; calculating the angle difference between the current attitude angle and the best attitude angle; calculating the perpendicular point direction and the perpendicular point distance of the first member of the unmanned ship formation to the direction of the best attitude angle according to the angle difference and the distance information; performing position transformation of the first member of the unmanned ship formation according to the perpendicular point direction and the perpendicular point distance; calculating the perpendicular point direction and the perpendicular point distance of all members of the unmanned ship formation; and obtaining the formation position of the unmanned ship formation according to the perpendicular point direction and the perpendicular point distance of all members of the unmanned ship formation.

[0100] Specifically, as Figure 5As shown, for any member of the unmanned ship formation, first, the radar sensor loaded thereby obtains the current attitude angle of the to-be-identified target, that is, the included angle formed by the direction of the line connecting the unmanned ship formation member and the to-be-identified target at the current moment and the motion direction of the to-be-identified target, and then the angle difference between the current attitude angle and the optimal attitude angle is calculated. If the angle difference is 0, the current attitude angle is the optimal attitude angle, and the members of the unmanned ship formation do not need to change the formation position, and at this time, the cooperative perception target recognition rate is optimal. If the angle difference is not 0, the current attitude angle is not the optimal attitude angle of the unmanned ship formation member to identify the target, and at this time, the problem of how to make the first member of the unmanned ship formation change from the current attitude angle to the optimal attitude angle is considered, starting from the change direction and the change distance, such as Figure 5 As shown, the embodiments of the present application calculate the perpendicular point direction and the perpendicular point distance of the first member of the unmanned ship formation to the direction of the optimal attitude angle, wherein the starting point of the perpendicular point direction is the current position of the first member of the unmanned ship formation in the direction of the current attitude angle, and the included angle between the perpendicular point direction and the direction of the current attitude angle is defined as θ 夹角 , the angle difference between the current attitude angle and the optimal attitude angle is defined as θ 角度差 , and the calculation formula of θ 夹角 is θ 夹角 = 90°- θ 角度差 , the perpendicular point distance is the product of the distance information of the to-be-identified target and the sine value of the angle difference, and the perpendicular point distance is defined as l 垂点距离 , the distance information of the to-be-identified target is defined as l 目标 , and the angle difference is defined as θ 角度差 , and the calculation formula is: l 垂点距离 = l 目标 × Sin(θ 角度差 ). Then, the first member of the unmanned ship formation moves a certain distance (the perpendicular point distance) along the perpendicular point direction, that is, the formation position change is completed, and the other members of the unmanned ship formation also change the position in the same way, and all the members change the position synchronously to further obtain the formation position of the unmanned ship formation.

[0101] Step S10: intelligently and dynamically changing the formation position of the unmanned ship members according to the formation position;

[0102] Step S11: after the formation position of the unmanned ship members is changed, the recognition result of the to-be-identified target is obtained through the trained recognition classifier model.

[0103] Specifically, according to the formation position, the members of the unmanned ship formation dynamically change positions in real time, and then the real-time acquired attitude angle, distance information, environmental information and image information are input into the trained identification classifier model, so as to obtain the target type of the to-be-identified target, and further realize the optimization of the formation cooperative perception strategy, dynamically change the formation in the shortest path, and effectively improve the target recognition accuracy.

[0104] Embodiment two

[0105] Based on the same inventive concept as the unmanned ship target recognition method based on intelligent formation position transformation in the foregoing embodiments, the present application also provides an unmanned ship target recognition device based on intelligent formation position transformation, as shown in Figure 2 The device comprises:

[0106] A first obtaining unit 1 is configured to obtain the attitude angle and distance information of a to-be-identified target through a radar sensor;

[0107] A second obtaining unit 2 is configured to obtain the environmental information of the to-be-identified target;

[0108] A third obtaining unit 3 is configured to obtain the image information of the to-be-identified target through an optical sensor, wherein the image information is in one-to-one correspondence with the attitude angle, the distance information and the environmental information;

[0109] A first constructing unit 4 is configured to construct an identification classifier model;

[0110] A fourth obtaining unit 5 is configured to obtain a sample set of attitude angle, distance information, environmental information and image information;

[0111] A first training unit 6 is configured to train the identification classifier model by using the sample set of attitude angle, distance information, environmental information and image information, to obtain a trained identification classifier model;

[0112] A fifth obtaining unit 7 is configured to take the attitude angle, distance information, environmental information and image information of the to-be-identified target as input items of the trained identification classifier model, to obtain a classification result vector;

[0113] A sixth obtaining unit 8 is configured to obtain the best attitude angle of the to-be-identified target according to the classification result vector;

[0114] A first calculating unit 9 is configured to calculate the formation position of the unmanned ship formation according to the best attitude angle;

[0115] a first transformation unit 10, configured to intelligently and dynamically transform the formation position of the unmanned ship member according to the formation position;

[0116] a seventh obtaining unit 11, configured to obtain, by the trained identification classifier model, an identification result of the to-be-identified target after the formation position of the unmanned ship member is transformed.

[0117] Further, the first obtaining unit 1 comprises:

[0118] an eighth obtaining unit, configured to obtain the motion direction of the to-be-identified target by a radar sensor;

[0119] a ninth obtaining unit, configured to obtain the position information of the to-be-identified target and the first member of the unmanned ship formation;

[0120] a tenth obtaining unit, configured to obtain the connecting line direction of the to-be-identified target and the first member of the unmanned ship formation according to the position information of the to-be-identified target and the first member of the unmanned ship formation;

[0121] a second calculating unit, configured to calculate the included angle between the motion direction and the connecting line direction to obtain the attitude angle of the to-be-identified target.

[0122] Further, the eighth obtaining unit comprises:

[0123] an eleventh obtaining unit, configured to obtain the first position information of the to-be-identified target by the radar sensor;

[0124] a twelfth obtaining unit, configured to obtain the second position information of the to-be-identified target by the radar sensor;

[0125] a thirteenth obtaining unit, configured to obtain the motion direction of the to-be-identified target according to the first position information and the second position information.

[0126] Further, the second obtaining unit 2 comprises:

[0127] a fourteenth obtaining unit, configured to obtain intelligence information by an intelligence information receiver;

[0128] a fifteenth obtaining unit, configured to obtain the environment information of the to-be-identified target according to the intelligence information, wherein the environment information comprises illumination, ocean wind power, and sea state information.

[0129] Further, the first construction unit 4 comprises:

[0130] a convolution pooling layer, configured to perform convolution pooling processing on a plurality of parallel target image information to be identified to form a first one-dimensional vector, wherein the number of the target image information to be identified is equal to the number of members in the unmanned ship formation;

[0131] a binary coding layer, configured to convert the attitude angle, distance information and environment information of the target to be identified into a binary array to form a second one-dimensional vector;

[0132] a full connection layer, wherein the input nodes of the full connection layer are composed of the first one-dimensional vector and the second one-dimensional vector in series;

[0133] an output node layer, wherein the number of nodes of the output node layer is equal to the number of target types of the target to be identified.

[0134] Further, the sixth obtaining unit 8 comprises:

[0135] a sixteenth obtaining unit, configured to perform normalization processing on the classification result vector to obtain a plurality of normalized classification result vectors;

[0136] a seventeenth obtaining unit, configured to obtain the target type corresponding to the maximum value element in the vector according to each normalized classification result vector, that is, to obtain the target type of the target to be identified;

[0137] an eighteenth obtaining unit, configured to obtain a target type optimal solution vector of the target to be identified according to the target type of the target to be identified, wherein the vector element corresponding to the target type is set to 1, and the vector elements corresponding to other target types are set to 0;

[0138] a third calculation unit, configured to calculate the Euclidean distance between the target type optimal solution vector and each normalized classification result vector;

[0139] a fourth calculation unit, configured to calculate the minimum value of a plurality of Euclidean distances;

[0140] a nineteenth obtaining unit, configured to obtain the best attitude angle of the target to be identified according to the classification vector result of the minimum Euclidean distance, wherein the best attitude angle and the classification vector result of the minimum Euclidean distance are in one-to-one correspondence.

[0141] Further, the first calculation unit 9 comprises:

[0142] a twentieth obtaining unit, configured to obtain a current attitude angle of the unmanned ship formation first member and the to-be-identified target;

[0143] a fifth calculating unit, configured to calculate an angle difference between the current attitude angle and the optimal attitude angle;

[0144] a sixth calculating unit, configured to calculate, according to the angle difference and the distance information, a foot-of-perpendicular direction and a foot-of-perpendicular distance of the unmanned ship formation first member to a direction of the optimal attitude angle;

[0145] a second transforming unit, configured to perform position transformation of the unmanned ship formation first member according to the foot-of-perpendicular direction and the foot-of-perpendicular distance;

[0146] a seventh calculating unit, configured to calculate the foot-of-perpendicular direction and the foot-of-perpendicular distance of all members of the unmanned ship formation;

[0147] a twenty-first obtaining unit, configured to obtain a formation position of the unmanned ship formation according to the foot-of-perpendicular direction and the foot-of-perpendicular distance of all members of the unmanned ship formation.

[0148] the foregoing Figure 1 The various manners of variation and specific examples of the unmanned ship target identification method based on formation position intelligent transformation in Embodiment One are also applicable to the unmanned ship target identification device based on formation position intelligent transformation in the present embodiment. Through the foregoing detailed description of the unmanned ship target identification method based on formation position intelligent transformation, those skilled in the art can clearly understand the implementation method of the unmanned ship target identification device based on formation position intelligent transformation in the present embodiment. Therefore, for the sake of brevity of the description, the detailed description is not repeated here.

[0149] Embodiment Three

[0150] Based on the same inventive concept as the unmanned ship target identification method based on formation position intelligent transformation in the foregoing embodiments, the present application further provides an unmanned ship target identification device based on formation position intelligent transformation, which has a computer program stored thereon. The program is executed by a processor to implement the steps of any method of the foregoing unmanned ship target identification method based on formation position intelligent transformation.

[0151] wherein, in the method, Figure 3In the exemplary embodiment, a bus architecture (represented by bus 300) can include any number of interconnecting buses and bridges needed to link various circuitry, including a processor 302 represented by one or more processors together with memory, represented by memory 304. Bus 300 can also link various other circuitry, such as peripheral devices, voltage regulators, and power management circuitry, all of which are well known in the art, and therefore, not described further. Bus interface 306 provides an interface between bus 300 and receiver 301 and transmitter 303. Receiver 301 and transmitter 303 can be the same device, i.e., a transceiver, providing a unit for communicating with various other apparatus over the transmission medium.

[0152] Processor 302 is responsible for managing the bus 300 and general processing, while the memory 304 can be used for storing data used by the processor 302 in executing operational instructions.

[0153] Embodiment Four

[0154] Based on the same inventive concept as the target recognition method of the unmanned ship based on the intelligent transformation of the formation position in the foregoing embodiments, the application also provides a computer readable storage medium, which stores a computer program, and the program is executed by a processor to implement the following steps:

[0155] Obtaining the attitude angle and distance information of the target to be recognized through a radar sensor; obtaining the environment information of the target to be recognized; obtaining the image information of the target to be recognized through an optical sensor, wherein the image information is in one-to-one correspondence with the attitude angle, the distance information and the environment information; constructing an identification classifier model; obtaining an attitude angle, distance information, environment information and image information sample set; training the identification classifier model by using the attitude angle, distance information, environment information and image information sample set to obtain a trained identification classifier model; taking the attitude angle, distance information, environment information and image information of the target to be recognized as input items of the trained identification classifier model to obtain a classification result vector; obtaining the best attitude angle of the target to be recognized according to the classification result vector; calculating the formation position of the unmanned ship formation according to the best attitude angle; intelligently and dynamically transforming the formation position of the unmanned ship members according to the formation position; and obtaining the recognition result of the target to be recognized through the trained identification classifier model after the formation position of the unmanned ship members is transformed.

[0156] In the specific implementation process, the program executed by the processor can also implement any method step in embodiment one.

[0157] The one or more technical solutions described above in the embodiments of the application have at least one or more of the following technical effects:

[0158] The unmanned ship target recognition method and device based on formation position intelligent transformation provided by the embodiment of the application, the method comprises: obtaining attitude angle and distance information of a target to be recognized through a radar sensor; obtaining environment information of the target to be recognized; obtaining image information of the target to be recognized through an optical sensor, wherein the image information is in one-to-one correspondence with the attitude angle, the distance information and the environment information; constructing an identification classifier model; obtaining an attitude angle, distance information, environment information and image information sample set; training the identification classifier model by using the attitude angle, distance information, environment information and image information sample set to obtain a trained identification classifier model; taking the attitude angle, distance information, environment information and image information of the target to be recognized as input items of the trained identification classifier model to obtain a classification result vector; obtaining the best attitude angle of the target to be recognized according to the classification result vector; calculating the formation position of the unmanned ship formation according to the best attitude angle; intelligently and dynamically transforming the formation position of the unmanned ship members according to the formation position; and obtaining the recognition result of the target to be recognized through the trained identification classifier model after the formation position of the unmanned ship members is transformed. The technical problem that the recognition accuracy of the unmanned ship marine mobile target is low in the prior art is solved, because the photoelectric, motion and environment characteristics of the target to be recognized by the unmanned ship are not input as a whole event, and the comprehensive recognition performance of different formation positions of the unmanned ship is not considered, so that the formation position of the unmanned ship cannot be intelligently adjusted according to the multi-sensor perception information of the unmanned ship, and the technical effects of ensuring timeliness, optimizing the formation cooperative perception strategy and effectively improving the target recognition accuracy are achieved based on the comprehensive recognition performance of different formation positions of the unmanned ship, the best attitude angle of the target to be recognized is obtained by comprehensively utilizing the multi-sensor information of the unmanned ship formation, and the position is dynamically transformed in real time in the shortest path.

[0159] Those skilled in the art should understand that the embodiments of the application can be provided as a method, a system or a computer program product. Therefore, the application can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Moreover, the application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0160] The present application is described in reference to the accompanying drawings, which use flow diagrams and / or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each block of the flow diagrams and / or block diagrams, and combinations of blocks in the flow diagrams and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flow diagrams and / or block diagrams block or blocks. Figure 1 one or more flow or flow diagrams and / or block or blocks. Figure 1 one or more flow or flow diagrams and / or block or blocks.

[0161] These computer program instructions can also be stored in a computer- readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions which implement the function specified in the flow diagrams and / or block diagrams block or blocks. Figure 1 one or more flow or flow diagrams and / or block or blocks. Figure 1 one or more flow or flow diagrams and / or block or blocks.

[0162] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flow diagrams and / or block diagrams block or blocks. Figure 1 one or more flow or flow diagrams and / or block or blocks. ​ one or more flow or flow diagrams and / or block or blocks.

[0163] Obviously, numerous modifications and variations of the present application are possible in light of the above teachings. It is therefore to be understood that within the scope of the appended claims and their legal equivalents, the application can be practiced otherwise than as specifically described.

Claims

1. A method for identifying a target by an unmanned ship based on intelligent transformation of platoon position, characterized in that, The method comprises: obtaining the attitude angle and distance information of the target to be identified through a radar sensor; obtaining the environment information of the target to be identified; obtaining the image information of the target to be identified through an optical sensor, wherein the image information is in one-to-one correspondence with the attitude angle, the distance information and the environment information; constructing an identification classifier model; obtaining an attitude angle, distance information, environment information and image information sample set; training the identification classifier model using the attitude angle, distance information, environment information and image information sample set to obtain a trained identification classifier model; taking the attitude angle, distance information, environment information and image information of the target to be identified as input items of the trained identification classifier model to obtain a classification result vector; obtaining the optimal attitude angle of the target to be identified according to the classification result vector; calculating the formation position of the unmanned ship formation according to the optimal attitude angle; intelligently and dynamically changing the formation position of the unmanned ship members according to the formation position; obtaining the identification result of the target to be identified through the trained identification classifier model after the formation position of the unmanned ship members is changed; wherein the optimal attitude angle of the target to be identified is obtained according to the classification result vector, comprising: normalizing the classification result vector to obtain a plurality of normalized classification result vectors; obtaining the target type corresponding to the maximum value element in the vector according to each normalized classification result vector, i.e. obtaining the target type of the target to be identified; obtaining the target type optimal solution vector of the target to be identified according to the target type of the target to be identified, wherein the vector element corresponding to the target type is set to 1 and the vector element corresponding to other target types is set to 0; calculating the Euclidean distance between the target type optimal solution vector and each normalized classification result vector; calculating the minimum value of a plurality of Euclidean distances; obtaining the optimal attitude angle of the target to be identified according to the classification vector result of the minimum Euclidean distance, wherein the optimal attitude angle is in one-to-one correspondence with the classification vector result of the minimum Euclidean distance.

2. The method of claim 1, wherein, The attitude angle and distance information of the target to be identified are obtained through a radar sensor, comprising: obtaining the movement direction of the target to be identified through a radar sensor; obtaining the position information of the target to be identified and the first member of the unmanned ship formation; obtaining the line direction of the target to be identified and the first member of the unmanned ship formation according to the position information of the target to be identified and the first member of the unmanned ship formation; calculating the included angle between the movement direction and the line direction to obtain the attitude angle of the target to be identified.

3. The method of claim 2, wherein, The movement direction of the target to be identified is obtained, comprising: obtaining the first position information of the target to be identified through the radar sensor; obtaining the second position information of the target to be identified through the radar sensor; obtaining the movement direction of the target to be identified according to the first position information and the second position information.

4. The method of claim 1, wherein, The environment information of the target to be identified is obtained, comprising: Obtain intelligence information through an intelligence information receiver; Obtain environment information of the target to be identified according to the intelligence information, wherein the environment information includes illumination, ocean wind, and sea state information.

5. The method of claim 1, wherein, The constructed identification classifier model comprises: A convolution pooling layer, configured to perform convolution pooling processing on a plurality of parallel target-to-be-identified image information to form a first one-dimensional vector, wherein the number of the target-to-be-identified image information is equal to the number of members of the unmanned ship formation; A binary coding layer, configured to convert the attitude angle, distance information, and environment information of the target to be identified into a binary array to form a second one-dimensional vector; A full connection layer, wherein the input nodes of the full connection layer are composed of the first one-dimensional vector and the second one-dimensional vector in series; An output node layer, wherein the number of nodes of the output node layer is equal to the number of target types of the target to be identified.

6. The method of claim 1, wherein, The calculation of the formation position of the unmanned ship formation according to the optimal attitude angle comprises: Obtaining the current attitude angle of the first member of the unmanned ship formation and the target to be identified; Calculating the angle difference between the current attitude angle and the optimal attitude angle; According to the angle difference and the distance information, calculating the perpendicular point direction and the perpendicular point distance of the first member of the unmanned ship formation to the direction of the optimal attitude angle; According to the perpendicular point direction and the perpendicular point distance, performing position transformation on the first member of the unmanned ship formation; Calculating the perpendicular point direction and the perpendicular point distance of all members of the unmanned ship formation; According to the perpendicular point direction and the perpendicular point distance of all members of the unmanned ship formation, obtaining the formation position of the unmanned ship formation.

7. An unmanned ship target recognition device based on intelligent transformation of formation position, characterized in that, The device comprises: A first obtaining unit, configured to obtain the attitude angle and distance information of a target to be identified through a radar sensor; A second obtaining unit, configured to obtain environment information of the target to be identified; A third obtaining unit, configured to obtain image information of the target to be identified through an optical sensor, wherein the image information, the attitude angle, the distance information, and the environment information are in one-to-one correspondence; A first constructing unit, configured to construct an identification classifier model; A fourth obtaining unit, configured to obtain a sample set of attitude angle, distance information, environment information, and image information; A first training unit, configured to train the identification classifier model by using the sample set of attitude angle, distance information, environment information, and image information, to obtain a trained identification classifier model; A fifth obtaining unit, configured to take the attitude angle, distance information, environment information, and image information of the target to be identified as input items of the trained identification classifier model, to obtain a classification result vector; A sixth obtaining unit, configured to obtain an optimal attitude angle of the target to be identified according to the classification result vector; A first calculating unit, configured to calculate the formation position of the unmanned ship formation according to the optimal attitude angle. The first transformation unit is configured to intelligently and dynamically transform the formation position of the unmanned ship member according to the formation position. The seventh obtaining unit is configured to obtain, by the trained identification classifier model, an identification result of the to-be-identified target after the formation position of the unmanned ship member is transformed. The sixth obtaining unit includes: The sixteenth obtaining unit is configured to normalize the classification result vector to obtain a plurality of normalized classification result vectors. The seventeenth obtaining unit is configured to obtain, according to each normalized classification result vector, a target type corresponding to the maximum value element in the vector, that is, a target type of the to-be-identified target. The eighteenth obtaining unit is configured to obtain, according to the target type of the to-be-identified target, a target type optimal solution vector of the to-be-identified target, wherein the vector element corresponding to the target type is set to 1, and the vector elements corresponding to other target types are set to 0. The third calculation unit is configured to calculate the Euclidean distance between the target type optimal solution vector and each normalized classification result vector. The fourth calculation unit is configured to calculate the minimum value of a plurality of Euclidean distances. The nineteenth obtaining unit is configured to obtain, according to the classification vector result of the minimum Euclidean distance, the best attitude angle of the to-be-identified target, wherein the best attitude angle and the classification vector result of the minimum Euclidean distance are in a one-to-one correspondence.

8. An unmanned ship target recognition device based on intelligent transformation of formation position, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor implements the following steps when executing the program: Obtain the attitude angle and distance information of the to-be-identified target through a radar sensor; Obtain the environment information of the to-be-identified target; Obtain the image information of the to-be-identified target through an optical sensor, wherein the image information, the attitude angle, the distance information, and the environment information are in a one-to-one correspondence; Construct an identification classifier model; Obtain an attitude angle, distance information, environment information, and image information sample set; Train the identification classifier model using the attitude angle, distance information, environment information, and image information sample set to obtain a trained identification classifier model; Use the attitude angle, distance information, environment information, and image information of the to-be-identified target as input items of the trained identification classifier model to obtain a classification result vector; Obtain the best attitude angle of the to-be-identified target according to the classification result vector; Calculate the formation position of the unmanned ship formation according to the best attitude angle; Intelligently and dynamically transform the formation position of the unmanned ship member according to the formation position. Obtain the identification result of the to-be-identified target through the trained identification classifier model after the formation position of the unmanned ship member is transformed. The method further includes: Normalize the classification result vector to obtain a plurality of normalized classification result vectors. According to each of the normalized classification result vectors, a target type corresponding to a maximum value element in the vector is obtained, that is, a target type of the target to be identified is obtained; According to the target type of the target to be identified, a target type optimal solution vector of the target to be identified is obtained, wherein a vector element corresponding to the target type is set to 1, and vector elements corresponding to other target types are set to 0; Euclidean distances between the target type optimal solution vector and each of the normalized classification result vectors are calculated; A minimum value of the Euclidean distances is calculated; According to a classification vector result of the minimum Euclidean distance, a best attitude angle of the target to be identified is obtained, wherein the best attitude angle and the classification vector result of the minimum Euclidean distance are in a one-to-one correspondence.

9. A computer-readable storage medium having stored thereon a computer program, characterized in that, When the program is executed by the processor, the following steps are implemented: Through a radar sensor, attitude angle and distance information of a target to be identified are obtained; Environment information of the target to be identified is obtained; Through an optical sensor, image information of the target to be identified is obtained, wherein the image information, the attitude angle, the distance information, and the environment information are in a one-to-one correspondence; An identification classifier model is constructed; An attitude angle, distance information, environment information, and image information sample set are obtained; The identification classifier model is trained by using the attitude angle, distance information, environment information, and image information sample set, to obtain a trained identification classifier model; The attitude angle, distance information, environment information, and image information of the target to be identified are taken as input items of the trained identification classifier model, to obtain a classification result vector; According to the classification result vector, a best attitude angle of the target to be identified is obtained; According to the best attitude angle, a formation position of an unmanned ship formation is calculated; According to the formation position, an intelligent dynamic transformation of formation positions of unmanned ship members is performed; After the formation positions of the unmanned ship members are transformed, an identification result of the target to be identified is obtained by using the trained identification classifier model; The method according to the classification result vector, to obtain the best attitude angle of the target to be identified, includes: The classification result vector is normalized to obtain a plurality of normalized classification result vectors; According to each of the normalized classification result vectors, a target type corresponding to a maximum value element in the vector is obtained, that is, a target type of the target to be identified is obtained; According to the target type of the target to be identified, a target type optimal solution vector of the target to be identified is obtained, wherein a vector element corresponding to the target type is set to 1, and vector elements corresponding to other target types are set to 0; Euclidean distances between the target type optimal solution vector and each of the normalized classification result vectors are calculated; A minimum value of the Euclidean distances is calculated; According to a classification vector result of the minimum Euclidean distance, a best attitude angle of the target to be identified is obtained, wherein the best attitude angle and the classification vector result of the minimum Euclidean distance are in a one-to-one correspondence.

Images