CN113640764B - Radar one-dimensional range profile identification method and device based on multi-dimension one-dimensional convolution - Google Patents

Radar one-dimensional range profile identification method and device based on multi-dimension one-dimensional convolution Download PDF

Info

Publication number
CN113640764B
CN113640764B CN202110908110.8A CN202110908110A CN113640764B CN 113640764 B CN113640764 B CN 113640764B CN 202110908110 A CN202110908110 A CN 202110908110A CN 113640764 B CN113640764 B CN 113640764B
Authority
CN
China
Prior art keywords
radar
dimensional
range profile
dimensional range
obtaining
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202110908110.8A
Other languages
Chinese (zh)
Other versions
CN113640764A (en
Inventor
郭强
王海鹏
王中训
刘传辉
赵凌业
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
School Of Aeronautical Combat Service Naval Aeronautical University Of Pla
Original Assignee
School Of Aeronautical Combat Service Naval Aeronautical University Of Pla
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by School Of Aeronautical Combat Service Naval Aeronautical University Of Pla filed Critical School Of Aeronautical Combat Service Naval Aeronautical University Of Pla
Priority to CN202110908110.8A priority Critical patent/CN113640764B/en
Publication of CN113640764A publication Critical patent/CN113640764A/en
Application granted granted Critical
Publication of CN113640764B publication Critical patent/CN113640764B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
    • G01S7/02—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00
    • G01S7/41—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00 using analysis of echo signal for target characterisation; Target signature; Target cross-section
    • G01S7/417—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00 using analysis of echo signal for target characterisation; Target signature; Target cross-section involving the use of neural networks
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
    • G01S13/88—Radar or analogous systems specially adapted for specific applications
    • G01S13/89—Radar or analogous systems specially adapted for specific applications for mapping or imaging
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
    • G01S7/02—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00
    • G01S7/41—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00 using analysis of echo signal for target characterisation; Target signature; Target cross-section
    • G01S7/415—Identification of targets based on measurements of movement associated with the target
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
    • Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
    • Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation

Landscapes

  • Engineering & Computer Science (AREA)
  • Remote Sensing (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Physics & Mathematics (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • General Physics & Mathematics (AREA)
  • Electromagnetism (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Radar Systems Or Details Thereof (AREA)

Abstract

The invention provides a radar one-dimensional range profile identification method and device based on multi-dimension one-dimensional convolution, wherein the method comprises the following steps: calculating angle information of the maneuvering target; acquiring radar one-dimensional range profile information of a maneuvering target; obtaining an angle data set and a radar one-dimensional range profile data set; obtaining a one-dimensional convolution estimated size of a radar one-dimensional range profile; constructing a multi-dimension one-dimensional convolutional neural network model for automatically identifying radar one-dimensional range profile; training a multi-dimension one-dimensional convolutional neural network model; and inputting the angle information and the radar one-dimensional range profile information into the trained multi-dimension one-dimensional convolutional neural network model to obtain a radar one-dimensional range profile recognition result of the maneuvering target. The method achieves the technical effects of automatically extracting the length characteristics of different distance units of different appearance structures in the one-dimensional range profile of the maneuvering target radar, being closer to the actual appearance characteristics and being combined with the angle information, thereby realizing the distinguishing of the one-dimensional range profile data characteristics of the maneuvering target under different radar visual angles and effectively improving the recognition rate.

Description

Radar one-dimensional range profile identification method and device based on multi-dimension one-dimensional convolution
Technical Field
The application relates to the technical field of automatic recognition of maneuvering targets, in particular to a radar one-dimensional range profile recognition method and device based on multi-dimension one-dimensional convolution.
Background
At present, the method for automatically identifying the maneuvering target based on the radar one-dimensional distance mainly comprises the steps of extracting the characteristics, and then obtaining the identification result of the maneuvering target through an identification classifier. The feature extraction method comprises the steps of extracting features through a multivariate statistical analysis method, automatically extracting multiple features of a radar one-dimensional range profile by using a deep neural network model, and comparing the features with the former, the latter has higher recognition accuracy.
However, in the process of implementing the technical solution in the embodiment of the present application, the present inventors have found that at least the following technical problems exist in the prior art:
in the existing automatic recognition method for the maneuvering target based on the radar one-dimensional distance image, the recognition rate is reduced due to different data characteristics of the target radar one-dimensional distance images of different angles under different radar vision lines, meanwhile, when the characteristics of the radar one-dimensional distance images are extracted, the convolution kernel size of the radar one-dimensional distance images needs to be manually preset, and only the same size can be selected, so that the multi-characteristic extraction of the radar one-dimensional distance images is incomplete and has errors, and the technical problem of low maneuvering target recognition rate is further caused.
Disclosure of Invention
The embodiment of the application provides a radar one-dimensional range profile identification method and device based on multi-dimension one-dimensional convolution, which solve the technical problems that in the existing radar one-dimensional range profile automatic identification method based on the radar one-dimensional range profile, due to different data characteristics of target radar one-dimensional range profiles of different angles under different radar sights, the identification rate is reduced, and meanwhile, when the characteristics of the radar one-dimensional range profiles are extracted, the convolution kernel size is required to be preset manually and only the same size can be selected, so that the multi-characteristic extraction of the radar one-dimensional range profiles is incomplete and has errors, and the identification rate of the maneuvering targets is low. Through the multi-dimension prediction of one-dimensional convolution, the method achieves the technical effects of automatically extracting the length characteristics of different distance units of different appearance structures in the one-dimensional range profile of the maneuvering target radar, being closer to the actual appearance characteristics of the maneuvering target, and increasing the detection dimension input of angle information, so that the one-dimensional range profile data characteristics of the maneuvering target under different radar visual angles are distinguished, and the recognition rate is effectively improved.
In view of the above problems, embodiments of the present application are provided to provide a method and apparatus for identifying a radar one-dimensional range profile based on multi-dimensional one-dimensional convolution.
In a first aspect, the present invention provides a radar one-dimensional range profile identification method based on multi-dimensional one-dimensional convolution, the method comprising: calculating angle information of a maneuvering target to be identified; acquiring radar one-dimensional distance image information of the maneuvering target to be identified through a broadband radar, wherein the radar one-dimensional distance image information and the angle information are in one-to-one correspondence; obtaining an angle data set and a radar one-dimensional range profile data set; obtaining a one-dimensional convolution estimated size of the radar one-dimensional range profile according to the radar one-dimensional range profile data set; constructing a multi-dimension one-dimensional convolutional neural network model for automatically identifying the radar one-dimensional range profile according to the one-dimensional convolutional estimated size; training the multi-dimension one-dimensional convolutional neural network model by utilizing the angle data set and the radar one-dimensional range profile data set; and inputting the angle information and the radar one-dimensional range profile information into the trained multi-dimension one-dimensional convolutional neural network model to obtain a radar one-dimensional range profile recognition result of the maneuvering target to be recognized.
Preferably, the calculating the angle information of the maneuvering target to be identified includes: calculating pitch angle information of the air platform to the maneuvering target to be identified on the sea/ground; calculating azimuth information of the motion track direction of the maneuvering target to be identified on the sea/ground; and obtaining the angle information of the maneuvering target to be identified according to the pitch angle information and the azimuth angle information.
Preferably, the calculating the pitch angle information of the aerial platform to the maneuvering target to be identified on the sea/ground comprises: obtaining pitching beams of an aerial platform through a laser radar; obtaining the maneuvering direction of the maneuvering target to be identified through a narrow-band radar; and calculating an included angle between the maneuvering direction and the pitching light beam according to the maneuvering direction of the pitching light beam of the aerial platform and the maneuvering target to be identified, and obtaining pitch angle information of the maneuvering target to be identified.
Preferably, the calculating the azimuth information of the motion trail direction of the maneuvering target to be identified on the sea/ground comprises: acquiring azimuth beams of an aerial platform through the laser radar; and calculating an included angle between the maneuvering direction and the azimuth light beam according to the azimuth light beam of the aerial platform and the maneuvering direction of the maneuvering target to be identified, and obtaining azimuth information of the maneuvering target to be identified.
Preferably, the constructing a multi-dimension one-dimensional convolutional neural network model for automatic identification of radar one-dimensional range profile includes: the multi-dimension one-dimensional convolution pooling layer is used for carrying out multi-dimension one-dimensional convolution pooling processing on the radar one-dimensional range profile and extracting multi-feature information of different range unit lengths of different appearance structures of targets of the radar one-dimensional range profile; the full-connection input layer node is formed by combining multi-characteristic information of different distance unit lengths of different outline structures of targets of the radar one-dimensional range profile extracted by the multi-dimension one-dimensional convolution pooling layer and the angle information; the full-connection middle layer is used for carrying out full connection on nodes of the full-connection input layer; and the full-connection output layer is in full connection with the nodes of the full-connection middle layer and is used for outputting the identification result of the multi-dimension one-dimensional convolutional neural network model, wherein the number of the nodes of the full-connection output layer is the number of the types of the target identification result.
Preferably, the obtaining the one-dimensional convolution estimated size of the radar one-dimensional range profile according to the radar one-dimensional range profile data set includes: extracting a historical radar one-dimensional range profile from the radar one-dimensional range profile data set, wherein the historical radar one-dimensional range profile comprises a plurality of scattered electromagnetic sub-echoes; according to the historical radar one-dimensional range profile, an ith electromagnetic sub-echo is obtained, wherein i=1, 2,3 and …; obtaining wave crests and wave troughs of the ith electromagnetic sub-echo; calculating the amplitude difference value of the wave crest and the wave trough of the ith electromagnetic sub-echo; obtaining an amplitude variation threshold; judging whether the amplitude difference value of the wave crest and the wave trough of the ith electromagnetic sub-echo is larger than the amplitude change threshold value or not; if the amplitude difference value of the wave crest and the wave trough of the ith electromagnetic sub-echo is larger than the amplitude change threshold value, obtaining an ith+n electromagnetic sub-echo, wherein n=1, 2,3 and … …; obtaining the wave crest and the wave trough of the (i+n) th electromagnetic sub-echo, and calculating the amplitude difference value of the wave crest and the wave trough of the (i+n) th electromagnetic sub-echo; judging whether the amplitude difference value of the wave crest and the wave trough of the (i+n) th electromagnetic sub-echo is larger than the amplitude change threshold value or not; if the amplitude difference value between the wave crest and the wave trough of the ith electromagnetic sub-echo is smaller than the amplitude change threshold value, obtaining the total width from the ith electromagnetic sub-echo to the ith electromagnetic sub-echo plus n-1, and storing the total width as a first width; if the difference value between the wave crest and the wave trough of the (i+n) th electromagnetic sub-echo is larger than the amplitude change threshold value, obtaining the total width from the (i) th electromagnetic sub-echo to the (i+n) th electromagnetic sub-echo, and storing the total width as a second width; sequentially extracting the remaining historical radar one-dimensional range profiles from the radar one-dimensional range profile data set until the extraction is completed, and obtaining a plurality of first widths and second widths; and storing the plurality of first widths and the plurality of second widths to obtain one-dimensional convolution estimated dimensions of the radar one-dimensional range profile.
Preferably, the obtaining the amplitude variation threshold includes: acquiring the amplitude variation mean value of a plurality of electromagnetic sub-echoes contained in the historical radar one-dimensional range profile; and obtaining the amplitude variation threshold according to the amplitude variation average value.
In a second aspect, the present invention provides a radar one-dimensional range profile identification device based on multi-dimensional one-dimensional convolution, the device comprising:
the first calculation unit is used for calculating angle information of the maneuvering target to be identified;
the first obtaining unit is used for obtaining radar one-dimensional distance image information of the maneuvering target to be identified through a broadband radar, wherein the radar one-dimensional distance image information and the angle information are in one-to-one correspondence;
the second obtaining unit is used for obtaining an angle data set and a radar one-dimensional range profile data set;
the third obtaining unit is used for obtaining the one-dimensional convolution estimated size of the radar one-dimensional range profile according to the radar one-dimensional range profile data set;
the first construction unit is used for constructing a multi-dimension one-dimensional convolutional neural network model for automatically identifying the radar one-dimensional range profile according to the one-dimensional convolutional estimated dimension;
The first training unit is used for training the multi-dimension one-dimensional convolutional neural network model by utilizing the angle data set and the radar one-dimensional range profile data set;
the fourth obtaining unit is used for inputting the angle information and the radar one-dimensional range profile information into the trained multi-dimension one-dimensional convolutional neural network model to obtain a radar one-dimensional range profile recognition result of the maneuvering target to be recognized.
Preferably, the first computing unit includes:
the second calculation unit is used for calculating pitch angle information of the air platform to the maneuvering target to be identified on the sea/ground;
the third calculation unit is used for calculating azimuth information of the movement track direction of the maneuvering target to be identified on the sea/ground;
and the fifth obtaining unit is used for obtaining the angle information of the maneuvering target to be identified according to the pitch angle information and the azimuth angle information.
Preferably, the second calculating unit includes:
a sixth obtaining unit for obtaining a pitching beam of the aerial platform by the laser radar;
A seventh obtaining unit for obtaining the maneuvering direction of the maneuvering target to be identified through a narrowband radar;
the eighth obtaining unit is used for calculating an included angle between the maneuvering direction and the pitching light beam according to the maneuvering direction of the pitching light beam of the aerial platform and the maneuvering target to be identified, and obtaining pitch angle information of the maneuvering target to be identified.
Preferably, the third computing unit includes:
a ninth obtaining unit, configured to obtain an azimuth beam of an aerial platform through the lidar;
and the tenth obtaining unit is used for calculating the included angle between the maneuvering direction and the azimuth light beam according to the maneuvering direction of the azimuth light beam of the aerial platform and the maneuvering target to be identified and obtaining azimuth information of the maneuvering target to be identified.
Preferably, the first building unit includes:
the multi-dimension one-dimensional convolution pooling layer is used for carrying out multi-dimension one-dimensional convolution pooling processing on the radar one-dimensional range profile and extracting multi-feature information of different range unit lengths of different appearance structures of targets of the radar one-dimensional range profile;
The full-connection input layer node is formed by combining multi-characteristic information of different distance unit lengths of different outline structures of targets of the radar one-dimensional range profile extracted by the multi-dimension one-dimensional convolution pooling layer and the angle information;
the full-connection middle layer is used for carrying out full connection on nodes of the full-connection input layer;
and the full-connection output layer is in full connection with the nodes of the full-connection middle layer and is used for outputting the identification result of the multi-dimension one-dimensional convolutional neural network model, wherein the number of the nodes of the full-connection output layer is the number of the types of the target identification result.
Preferably, the third obtaining unit includes:
the first extraction unit is used for extracting a historical radar one-dimensional range profile from the radar one-dimensional range profile data set, wherein the historical radar one-dimensional range profile comprises a plurality of scattered electromagnetic sub-echoes;
an eleventh obtaining unit, configured to obtain an ith electromagnetic sub-echo according to the historical radar one-dimensional range profile, where i=1, 2,3, … …;
a twelfth obtaining unit for obtaining a peak and a trough of the ith electromagnetic sub-echo;
The fourth calculation unit is used for calculating the amplitude difference value of the wave crest and the wave trough of the ith electromagnetic sub-echo;
a thirteenth obtaining unit configured to obtain an amplitude variation threshold value;
the first judging unit is used for judging whether the amplitude difference value of the wave crest and the wave trough of the ith electromagnetic sub-echo is larger than the amplitude change threshold value or not;
a fourteenth obtaining unit configured to obtain an i+n electromagnetic sub-echo, where n=1, 2,3, … …, if a difference in amplitude of a peak and a trough of the i electromagnetic sub-echo is greater than the amplitude variation threshold;
a fifteenth obtaining unit, configured to obtain a peak and a trough of the i+n electromagnetic sub-echo, and calculate an amplitude difference value of the peak and the trough of the i+n electromagnetic sub-echo;
the second judging unit is used for judging whether the amplitude difference value between the wave crest and the wave trough of the (i+n) th electromagnetic sub-echo is larger than the amplitude change threshold value;
the first storage unit is used for obtaining the total width from the ith electromagnetic sub-echo to the (i+n-1) th electromagnetic sub-echo and storing the total width as a first width if the amplitude difference value between the wave crest and the wave trough of the (i+n) th electromagnetic sub-echo is smaller than the amplitude change threshold;
The second storage unit is used for obtaining the total width from the ith electromagnetic sub-echo to the (i+n) th electromagnetic sub-echo and storing the total width as a second width if the difference value between the wave crest and the wave trough of the (i+n) th electromagnetic sub-echo is larger than the amplitude change threshold;
a sixteenth obtaining unit, configured to sequentially extract remaining historical radar one-dimensional range profiles from the radar one-dimensional range profile dataset, until extraction is completed, to obtain a plurality of first widths and second widths;
and the seventeenth obtaining unit is used for storing the plurality of first widths and the plurality of second widths to obtain one-dimensional convolution estimated sizes of the radar one-dimensional range profile.
Preferably, the thirteenth obtaining unit includes:
an eighteenth obtaining unit, configured to obtain an amplitude variation average value of a plurality of electromagnetic sub-echoes included in the one-dimensional range profile of the historical radar;
a nineteenth obtaining unit, configured to obtain the amplitude variation threshold according to the amplitude variation average value.
In a third aspect, the present invention provides a radar one-dimensional range profile recognition device based on multi-dimensional one-dimensional convolution, comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the following steps when executing the program:
Calculating angle information of a maneuvering target to be identified; acquiring radar one-dimensional distance image information of the maneuvering target to be identified through a broadband radar, wherein the radar one-dimensional distance image information and the angle information are in one-to-one correspondence; obtaining an angle data set and a radar one-dimensional range profile data set; obtaining a one-dimensional convolution estimated size of the radar one-dimensional range profile according to the radar one-dimensional range profile data set; constructing a multi-dimension one-dimensional convolutional neural network model for automatically identifying the radar one-dimensional range profile according to the one-dimensional convolutional estimated size; training the multi-dimension one-dimensional convolutional neural network model by utilizing the angle data set and the radar one-dimensional range profile data set; and inputting the angle information and the radar one-dimensional range profile information into the trained multi-dimension one-dimensional convolutional neural network model to obtain a radar one-dimensional range profile recognition result of the maneuvering target to be recognized.
In a fourth aspect, the present invention provides a computer readable storage medium having stored thereon a computer program which when executed by a processor performs the steps of:
calculating angle information of a maneuvering target to be identified; acquiring radar one-dimensional distance image information of the maneuvering target to be identified through a broadband radar, wherein the radar one-dimensional distance image information and the angle information are in one-to-one correspondence; obtaining an angle data set and a radar one-dimensional range profile data set; obtaining a one-dimensional convolution estimated size of the radar one-dimensional range profile according to the radar one-dimensional range profile data set; constructing a multi-dimension one-dimensional convolutional neural network model for automatically identifying the radar one-dimensional range profile according to the one-dimensional convolutional estimated size; training the multi-dimension one-dimensional convolutional neural network model by utilizing the angle data set and the radar one-dimensional range profile data set; and inputting the angle information and the radar one-dimensional range profile information into the trained multi-dimension one-dimensional convolutional neural network model to obtain a radar one-dimensional range profile recognition result of the maneuvering target to be recognized.
The above technical solutions in the embodiments of the present application at least have one or more of the following technical effects:
the embodiment of the application provides a radar one-dimensional range profile identification method and device based on multi-dimension one-dimensional convolution, wherein the method comprises the following steps: calculating angle information of a maneuvering target to be identified; acquiring radar one-dimensional distance image information of the maneuvering target to be identified through a broadband radar, wherein the radar one-dimensional distance image information and the angle information are in one-to-one correspondence; obtaining an angle data set and a radar one-dimensional range profile data set; obtaining a one-dimensional convolution estimated size of the radar one-dimensional range profile according to the radar one-dimensional range profile data set; constructing a multi-dimension one-dimensional convolutional neural network model for automatically identifying the radar one-dimensional range profile according to the one-dimensional convolutional estimated size; training the multi-dimension one-dimensional convolutional neural network model by utilizing the angle data set and the radar one-dimensional range profile data set; and inputting the angle information and the radar one-dimensional range profile information into the trained multi-dimension one-dimensional convolutional neural network model to obtain a radar one-dimensional range profile recognition result of the maneuvering target to be recognized. The method solves the technical problems that in the existing automatic recognition method based on the radar one-dimensional distance image maneuvering target, the recognition rate is reduced due to different data characteristics of target radar one-dimensional distance images of different angles under different radar vision lines, and meanwhile, when the characteristics of the radar one-dimensional distance images are extracted, the convolution kernel size of the radar one-dimensional distance images needs to be manually preset and can only be selected to be the same size, so that the multi-characteristic extraction of the radar one-dimensional distance images is incomplete and has errors, and the maneuvering target recognition rate is low. Through the multi-dimension prediction of one-dimensional convolution, the method achieves the technical effects of automatically extracting the length characteristics of different distance units of different appearance structures in the one-dimensional range profile of the maneuvering target radar, being closer to the actual appearance characteristics of the maneuvering target, and increasing the detection dimension input of angle information, so that the one-dimensional range profile data characteristics of the maneuvering target under different radar visual angles are distinguished, and the recognition rate is effectively improved.
The foregoing description is only an overview of the present invention, and is intended to be implemented in accordance with the teachings of the present invention in order that the same may be more clearly understood and to make the same and other objects, features and advantages of the present invention more readily apparent.
Drawings
FIG. 1 is a schematic flow chart of a radar one-dimensional range profile recognition method based on multi-dimensional one-dimensional convolution in an embodiment of the invention;
FIG. 2 is a schematic structural diagram of a radar one-dimensional range profile recognition device based on multi-dimensional one-dimensional convolution in an embodiment of the invention;
FIG. 3 is a schematic diagram of another radar one-dimensional range profile recognition device based on multi-dimensional one-dimensional convolution according to an embodiment of the present invention;
FIG. 4 is a schematic diagram of a radar one-dimensional range profile in an embodiment of the present invention;
FIG. 5 is a schematic diagram of a multi-dimensional convolutional neural network model in accordance with an embodiment of the present invention.
Reference numerals illustrate: the device comprises a first computing unit 11, a first obtaining unit 12, a second obtaining unit 13, a third obtaining unit 14, a first constructing unit 15, a first training unit 16, a fourth obtaining unit 17, a bus 300, a receiver 301, a processor 302, a transmitter 303, a memory 304, and a bus interface 306.
Detailed Description
The embodiment of the invention provides a radar one-dimensional range profile identification method and device based on multi-dimension one-dimensional convolution, which solve the technical problems that in the existing radar one-dimensional range profile automatic identification method based on the radar one-dimensional range profile, due to different data characteristics of target radar one-dimensional range profiles of different angles under different radar sights, the identification rate is reduced, and meanwhile, when the characteristics of the radar one-dimensional range profiles are extracted, the convolution kernel size is required to be preset manually and only the same size can be selected, so that the multi-characteristic extraction of the radar one-dimensional range profiles is incomplete and has errors, and the identification rate of the maneuvering targets is low.
The technical scheme provided by the invention has the following overall thought: calculating angle information of a maneuvering target to be identified; acquiring radar one-dimensional distance image information of the maneuvering target to be identified through a broadband radar, wherein the radar one-dimensional distance image information and the angle information are in one-to-one correspondence; obtaining an angle data set and a radar one-dimensional range profile data set; obtaining a one-dimensional convolution estimated size of the radar one-dimensional range profile according to the radar one-dimensional range profile data set; constructing a multi-dimension one-dimensional convolutional neural network model for automatically identifying the radar one-dimensional range profile according to the one-dimensional convolutional estimated size; training the multi-dimension one-dimensional convolutional neural network model by utilizing the angle data set and the radar one-dimensional range profile data set; and inputting the angle information and the radar one-dimensional range profile information into the trained multi-dimension one-dimensional convolutional neural network model to obtain a radar one-dimensional range profile recognition result of the maneuvering target to be recognized. Through the multi-dimension prediction of one-dimensional convolution, the method achieves the technical effects of automatically extracting the length characteristics of different distance units of different appearance structures in the one-dimensional range profile of the maneuvering target radar, being closer to the actual appearance characteristics of the maneuvering target, and increasing the detection dimension input of angle information, so that the one-dimensional range profile data characteristics of the maneuvering target under different radar visual angles are distinguished, and the recognition rate is effectively improved.
The following detailed description of the technical solutions of the present application will be given by way of the accompanying drawings and specific embodiments, and it should be understood that the specific features of the embodiments and embodiments of the present application are detailed descriptions of the technical solutions of the present application, and not limiting the technical solutions of the present application, and that the embodiments and technical features of the embodiments of the present application may be combined with each other without conflict.
The term "and/or" is herein merely an association relationship describing an associated object, meaning that there may be three relationships, e.g., a and/or B, may represent: a exists alone, A and B exist together, and B exists alone. In addition, the character "/" herein generally indicates that the front and rear associated objects are an "or" relationship.
Example 1
Fig. 1 is a schematic flow chart of a radar one-dimensional range profile recognition method based on multi-dimensional one-dimensional convolution in an embodiment of the application. As shown in fig. 1, an embodiment of the present application provides a radar one-dimensional range profile identification method based on multi-dimensional one-dimensional convolution, where the method includes:
step 110: calculating angle information of a maneuvering target to be identified;
specifically, in the field of radar target detection, the radar one-dimensional range profile is a radar high-resolution range profile, and the wavelength of high-frequency electromagnetic waves emitted by the high-resolution radar is far smaller than the size of the maneuvering target to be identified. In the detection process of the broadband radar on the target, the maneuvering target to be identified can be regarded as a set of mutually independent scattering points, firstly, the broadband radar transmits electromagnetic waves to the maneuvering target to be identified, and then the group of scattering points are subjected to back scattering so as to form vector sums of electromagnetic wave sub-echoes, namely a radar one-dimensional range profile. In addition, the radar one-dimensional range profile has the problem of angle sensitivity, namely, under the radar sight lines of different angles, the data difference of the radar one-dimensional range profile is huge, and particularly for a maneuvering target, the automatic recognition rate is not high if the angle change is not considered. According to the embodiment of the application, the angle information of the maneuvering target to be identified relative to the air platform on the sea/ground is taken as an important factor influencing the identification result, and the angle information of the maneuvering target to be identified is obtained through calculation while the one-dimensional range profile of the maneuvering radar to be identified is detected, so that the dimension of the detection information is increased, and the identification precision of the maneuvering target on the sea/ground is further effectively improved.
In step 110, the angle information of the maneuvering target to be identified on the sea level or the ground relative to the aerial platform comprises pitch angle information and azimuth angle information, so calculating the angle information of the maneuvering target to be identified comprises: and respectively calculating pitch angle information and azimuth angle information of the air platform to the maneuvering target to be identified on the sea/ground, and then calculating the angle information of the maneuvering target to be identified according to the pitch angle information and the azimuth angle information.
Further, the calculating the pitch angle information of the air platform to the maneuvering target to be identified on the sea/ground comprises: firstly, obtaining pitching beams of an air platform relative to the maneuvering target to be identified on the sea/ground through a laser radar; obtaining the maneuvering direction of the maneuvering target to be identified on the sea/ground through a narrow-band radar; according to the pitching light beam of the aerial platform and the maneuvering direction of the maneuvering target to be identified, calculating to obtain the included angle between the maneuvering direction and the pitching light beam, thereby realizing the technical effect of obtaining the pitching angle information of the maneuvering target.
Further, the calculating the azimuth information of the movement track direction of the maneuvering target to be identified on the sea/ground comprises: firstly, obtaining azimuth beams of the air platform relative to the maneuvering target to be identified on the sea/ground through the laser radar; and then calculating and obtaining the included angle between the maneuvering direction and the azimuth beam according to the azimuth beam and the maneuvering direction, thereby realizing the technical effect of obtaining maneuvering target azimuth information.
Step 120: acquiring radar one-dimensional distance image information of the maneuvering target to be identified through a broadband radar, wherein the radar one-dimensional distance image information and the angle information are in one-to-one correspondence;
specifically, the embodiment of the invention acquires the radar one-dimensional range profile information of the maneuvering target to be identified on the sea level or the ground in real time by using the broadband radar, and is remarkable in that the radar one-dimensional range profile information and the angle information are matched with each other in a one-to-one correspondence relationship, in other words, the angle information acquired and calculated by the laser radar or the narrow-band radar has a unique serial number, the acquired radar one-dimensional range profile information also has a unique serial number, and the unique serial number of the radar one-dimensional range profile information acquired at the same moment is the same as the unique serial number of the angle information, namely, the radar one-dimensional range profile information and the angle information of the maneuvering target to be identified are acquired at the same moment, so that the technical effect of the association and pairing of the radar one-dimensional range profile information and the angle information is realized.
Step 130: obtaining an angle data set and a radar one-dimensional range profile data set;
Specifically, the air platform detects historical maneuvering targets on the sea level or the ground, wherein the historical maneuvering targets are a plurality of monitored existing maneuvering targets, angle information and radar one-dimensional range profile information of the historical maneuvering targets are obtained, the angle information and the radar one-dimensional range profile information of the historical maneuvering targets are in one-to-one correspondence, and a plurality of related angle information and radar one-dimensional range profile information are combined together to form an angle data set and a radar one-dimensional range profile data set, and are used for training the multi-dimension one-dimensional convolutional neural network model.
Step 140: obtaining a one-dimensional convolution estimated size of the radar one-dimensional range profile according to the radar one-dimensional range profile data set;
further, the obtaining, according to the radar one-dimensional range profile data set, a one-dimensional convolution estimated size of the radar one-dimensional range profile includes:
extracting a historical radar one-dimensional range profile from the radar one-dimensional range profile data set, wherein the historical radar one-dimensional range profile comprises a plurality of scattered electromagnetic sub-echoes;
according to the historical radar one-dimensional range profile, an ith electromagnetic sub-echo is obtained, wherein i=1, 2,3 and …;
Obtaining wave crests and wave troughs of the ith electromagnetic sub-echo;
calculating the amplitude difference value of the wave crest and the wave trough of the ith electromagnetic sub-echo;
obtaining an amplitude variation threshold;
judging whether the amplitude difference value of the wave crest and the wave trough of the ith electromagnetic sub-echo is larger than the amplitude change threshold value or not;
if the amplitude difference value of the wave crest and the wave trough of the ith electromagnetic sub-echo is larger than the amplitude change threshold value, obtaining an ith+n electromagnetic sub-echo, wherein n=1, 2,3 and …;
obtaining the wave crest and the wave trough of the (i+n) th electromagnetic sub-echo, and calculating the amplitude difference value of the wave crest and the wave trough of the (i+n) th electromagnetic sub-echo;
judging whether the amplitude difference value of the wave crest and the wave trough of the (i+n) th electromagnetic sub-echo is larger than the amplitude change threshold value or not;
if the amplitude difference value between the wave crest and the wave trough of the ith electromagnetic sub-echo is smaller than the amplitude change threshold value, obtaining the total width from the ith electromagnetic sub-echo to the ith electromagnetic sub-echo plus n-1, and storing the total width as a first width;
if the difference value between the wave crest and the wave trough of the (i+n) th electromagnetic sub-echo is larger than the amplitude change threshold value, obtaining the total width from the (i) th electromagnetic sub-echo to the (i+n) th electromagnetic sub-echo, and storing the total width as a second width;
Sequentially extracting the remaining historical radar one-dimensional range profiles from the radar one-dimensional range profile data set until the extraction is completed, and obtaining a plurality of first widths and second widths;
and storing the plurality of first widths and the plurality of second widths to obtain one-dimensional convolution estimated dimensions of the radar one-dimensional range profile.
Further, the obtaining the amplitude variation threshold includes: acquiring the amplitude variation mean value of a plurality of electromagnetic sub-echoes contained in the historical radar one-dimensional range profile; and obtaining the amplitude variation threshold according to the amplitude variation average value.
Specifically, the radar one-dimensional range profile data set comprises a plurality of historical radar one-dimensional range profiles, each historical radar one-dimensional range profile comprises a plurality of scattered electromagnetic sub-echoes, wherein the wave crest and the wave trough of the electromagnetic sub-echoes reflect the distribution condition of scattering points of the maneuvering target to be identified under the radar sight of a certain angle. Because the capability of scattering electromagnetic waves of different outline structures of the maneuvering target to be identified is different, such as a nose, a wing and the like of an airplane, typical outline structures generally having the characteristics of non-streamline design, large radar reflection area and the like have stronger capability of scattering electromagnetic waves, while atypical outline structures having the characteristics of streamline design, small radar reflection area and the like have weaker capability of scattering electromagnetic waves, and outline structures with strong capability of scattering can form electromagnetic sub-echoes with larger amplitude change in the radar one-dimensional range profile, so that abundant outline structure information is contained in the radar one-dimensional range profile, and the method is important information for identifying the target.
If the radar bandwidth is B, the radar range resolution is Δr=c/2B, where c is the light speed, along the line-of-sight direction detected by the broadband radar, the maneuvering target to be identified may be approximately expressed as a number of range units, and the width is Δr, the radar one-dimensional range profile is a vector superposition of electromagnetic sub-echoes of all scattering points in each range unit, as shown in fig. 4, the abscissa of the radar one-dimensional range profile is the range unit of the maneuvering target to be identified under the radar line-of-sight of a certain angle, and the ordinate is the amplitude of the electromagnetic echo of each range unit of the maneuvering target to be identified, where the amplitude of the radar one-dimensional range profile is affected by factors such as radar, target, distance and environment.
In order to estimate the one-dimensional convolution size of the radar one-dimensional range profile, the embodiment of the invention extracts a historical radar one-dimensional range profile from the radar one-dimensional range profile data set, sequentially obtains an ith electromagnetic sub-echo along an abscissa, calculates the amplitude difference value of the peak and the trough of the ith electromagnetic sub-echo, and if the amplitude difference value of the peak and the trough of the ith electromagnetic sub-echo is larger than the amplitude variation threshold, the amplitude variation threshold is the amplitude variation average value of a plurality of electromagnetic sub-echoes contained in the historical radar one-dimensional range profile, which indicates that the ith electromagnetic sub-echo is a typical appearance structure on the maneuvering target to be identified, and then continues to obtain the next electromagnetic sub-echo (i.e. the ith+n electromagnetic sub-echo) immediately after the ith electromagnetic sub-echo, if the amplitude difference value of the peak and the trough of the ith electromagnetic sub-echo is smaller than the amplitude variation threshold, the ith+n electromagnetic sub-echo does not have a typical appearance structure, and only needs to store the width of the ith electromagnetic sub-echo at this moment as a neural network size; if the amplitude difference value between the wave crest and the wave trough of the ith electromagnetic sub-echo is larger than the amplitude change threshold value, the ith electromagnetic sub-echo is indicated to have a typical appearance structure, and the total width of the ith electromagnetic sub-echo to the ith electromagnetic sub-echo is required to be stored at the moment and is used as the other dimension of the neural network model one-dimensional convolution kernel; sequentially extracting the remaining historical radar one-dimensional range images from the radar one-dimensional range image data set until the extraction is completed, obtaining all different sizes of the one-dimensional convolution kernels, and further obtaining the one-dimensional convolution estimated size of the radar one-dimensional range images.
Step 150: constructing a multi-dimension one-dimensional convolutional neural network model for automatically identifying the radar one-dimensional range profile according to the one-dimensional convolutional estimated size;
further, the constructing a multi-dimension one-dimensional convolutional neural network model for automatic identification of radar one-dimensional range profile includes:
the multi-dimension one-dimensional convolution pooling layer is used for carrying out multi-dimension one-dimensional convolution pooling processing on the radar one-dimensional range profile and extracting multi-feature information of different range unit lengths of different appearance structures of targets of the radar one-dimensional range profile;
the full-connection input layer node is formed by combining multi-characteristic information of different distance unit lengths of different outline structures of targets of the radar one-dimensional range profile extracted by the multi-dimension one-dimensional convolution pooling layer and the angle information;
the full-connection middle layer is used for carrying out full connection on nodes of the full-connection input layer;
and the full-connection output layer is in full connection with the nodes of the full-connection middle layer and is used for outputting the identification result of the multi-dimension one-dimensional convolutional neural network model, wherein the number of the nodes of the full-connection output layer is the number of the types of the target identification result.
Specifically, as shown in fig. 5, since the multi-dimensional one-dimensional convolutional neural network model for automatically identifying the radar one-dimensional range profile has two input items of radar one-dimensional range profile information and angle information, in the multi-dimensional one-dimensional convolutional neural network model, the radar one-dimensional range profile information is firstly processed by a multi-dimensional one-dimensional convolutional pooling layer, multi-feature information (namely multi-dimensional multi-feature information of the radar one-dimensional range profile) of different range cell lengths of targets of the radar one-dimensional range profile is extracted, namely the radar one-dimensional range profile information is subjected to dimension reduction to form a one-dimensional vector, specifically, x in fig. 5 represents elements of different dimension convolutional kernels, the number of the elements of the different convolutional kernels is different, namely the dimension is different, the one-dimensional vector is formed by rolling and pooling processing of a plurality of one-dimensional convolutional kernels of different dimensions, simultaneously, the angle information is directly input into a one-dimensional vector, the multi-dimension multi-feature information of the radar one-dimensional range profile and the angle information are connected in parallel to the fully connected input layer, wherein the node number of the fully connected input layer is equal to the sum of the multi-dimension multi-feature information of the radar one-dimensional range profile and the number of one-dimensional vector elements formed by combining the angle information), the radar one-dimensional range profile information and the angle information are subjected to deep training, the internal relation between the radar one-dimensional range profile information and the angle information is mined, so that two input items are mutually related, and the output result which is closer to the real situation of a maneuvering target can be obtained through the fully connected intermediate layer and the fully connected output layer, thereby achieving the purpose of automatically extracting the length features of different distance units of different appearance structures in the radar one-dimensional range profile of the maneuvering target and being closer to the actual appearance features of the maneuvering target, meanwhile, the detection dimension of the angle information is increased, so that the one-dimensional range profile data characteristics of maneuvering targets under different radar view angles are distinguished, and the technical effect of the recognition rate is effectively improved.
Step 160: training the multi-dimension one-dimensional convolutional neural network model by utilizing the angle data set and the radar one-dimensional range profile data set;
specifically, the angle data set and the radar one-dimensional range profile data set contain angle information and radar one-dimensional range profile information of a plurality of correlation paired maneuvering targets, so that the multi-dimension one-dimensional convolutional neural network model performs infinite deep learning and training in the angle information and the radar one-dimensional range profile information until the multi-dimension one-dimensional convolutional neural network model converges, and the technical effect of training the neural network model is achieved.
Step 170: and inputting the angle information and the radar one-dimensional range profile information into the trained multi-dimension one-dimensional convolutional neural network model to obtain a radar one-dimensional range profile recognition result of the maneuvering target to be recognized.
Specifically, when an unknown maneuvering target exists on the sea level or the ground, namely, the maneuvering target to be identified in the embodiment of the invention is firstly detected by the narrow-band radar to be in the position and maneuvering direction of the maneuvering target to be identified, then the system sends a close detection instruction to the narrow-band radar, meanwhile, the wide-band radar acquires radar one-dimensional range profile information of the maneuvering target to be identified from different angles in real time, the laser radar is responsible for acquiring azimuth angle information and pitch angle information of the maneuvering target to be identified, angle information of the maneuvering target to be identified relative to an aerial platform is obtained through calculation, then the radar one-dimensional range profile information and the angle information are associated and paired to be used as input items, and meanwhile, the radar one-dimensional range profile identification result of the maneuvering target to be identified is obtained by being connected into the trained multi-dimensional convolutional neural network model, wherein the identification result is continuously updated in real time along with the change of the angle information and the radar one-dimensional range profile information.
Example two
Based on the same inventive concept as the radar one-dimensional range profile recognition method based on the multi-dimensional one-dimensional convolution in the foregoing embodiment, the present invention further provides a radar one-dimensional range profile recognition device based on the multi-dimensional one-dimensional convolution, as shown in fig. 2, where the device includes:
a first calculation unit 11, wherein the first calculation unit 11 is used for calculating angle information of a maneuvering target to be identified;
a first obtaining unit 12, where the first obtaining unit 12 is configured to obtain radar one-dimensional range profile information of the maneuvering target to be identified through a wideband radar, where the radar one-dimensional range profile information and the angle information are in a one-to-one correspondence;
a second obtaining unit 13, where the second obtaining unit 13 is configured to obtain an angle data set and a radar one-dimensional range profile data set;
a third obtaining unit 14, where the third obtaining unit 14 is configured to obtain a one-dimensional convolution estimated size of the radar one-dimensional range profile according to the radar one-dimensional range profile data set;
the first construction unit 15 is used for constructing a multi-dimension one-dimensional convolutional neural network model for automatically identifying the radar one-dimensional range profile according to the one-dimensional convolutional estimated dimension;
A first training unit 16, where the first training unit 16 is configured to train the multi-size one-dimensional convolutional neural network model using the angle data set and the radar one-dimensional range profile data set;
the fourth obtaining unit 17 is configured to input the angle information and the radar one-dimensional range profile information into the trained multi-dimensional one-dimensional convolutional neural network model, and obtain a radar one-dimensional range profile recognition result of the maneuvering target to be recognized.
Further, the first computing unit 11 includes:
the second calculation unit is used for calculating pitch angle information of the air platform to the maneuvering target to be identified on the sea/ground;
the third calculation unit is used for calculating azimuth information of the movement track direction of the maneuvering target to be identified on the sea/ground;
and the fifth obtaining unit is used for obtaining the angle information of the maneuvering target to be identified according to the pitch angle information and the azimuth angle information.
Further, the second calculation unit includes:
a sixth obtaining unit for obtaining a pitching beam of the aerial platform by the laser radar;
A seventh obtaining unit for obtaining the maneuvering direction of the maneuvering target to be identified through a narrowband radar;
the eighth obtaining unit is used for calculating an included angle between the maneuvering direction and the pitching light beam according to the maneuvering direction of the pitching light beam of the aerial platform and the maneuvering target to be identified, and obtaining pitch angle information of the maneuvering target to be identified.
Further, the third computing unit includes:
a ninth obtaining unit, configured to obtain an azimuth beam of an aerial platform through the lidar;
and the tenth obtaining unit is used for calculating the included angle between the maneuvering direction and the azimuth light beam according to the maneuvering direction of the azimuth light beam of the aerial platform and the maneuvering target to be identified and obtaining azimuth information of the maneuvering target to be identified.
Further, the first construction unit 15 includes:
the multi-dimension one-dimensional convolution pooling layer is used for carrying out multi-dimension one-dimensional convolution pooling processing on the radar one-dimensional range profile and extracting multi-feature information of different dimensions of the radar one-dimensional range profile;
The full-connection input layer node is formed by combining the radar one-dimensional range profile multi-size multi-feature information extracted by the multi-size one-dimensional convolution pooling layer and the angle information;
the full-connection middle layer is used for carrying out full connection on nodes of the full-connection input layer;
and the full-connection output layer is in full connection with the nodes of the full-connection middle layer and is used for outputting the identification result of the multi-dimension one-dimensional convolutional neural network model, wherein the number of the nodes of the full-connection output layer is the number of the types of the target identification result.
Further, the third obtaining unit 14 includes:
the first extraction unit is used for extracting a historical radar one-dimensional range profile from the radar one-dimensional range profile data set, wherein the historical radar one-dimensional range profile comprises a plurality of scattered electromagnetic sub-echoes;
an eleventh obtaining unit, configured to obtain an ith electromagnetic sub-echo according to the historical radar one-dimensional range profile, where i=1, 2,3, …;
a twelfth obtaining unit for obtaining a peak and a trough of the ith electromagnetic sub-echo;
The fourth calculation unit is used for calculating the amplitude difference value of the wave crest and the wave trough of the ith electromagnetic sub-echo;
a thirteenth obtaining unit configured to obtain an amplitude variation threshold value;
the first judging unit is used for judging whether the amplitude difference value of the wave crest and the wave trough of the ith electromagnetic sub-echo is larger than the amplitude change threshold value or not;
a fourteenth obtaining unit for obtaining an i+n electromagnetic sub-echo, if a difference in amplitude of a peak and a trough of the i electromagnetic sub-echo is greater than the amplitude variation threshold, wherein n=1, 2,3.
A fifteenth obtaining unit, configured to obtain a peak and a trough of the i+n electromagnetic sub-echo, and calculate an amplitude difference value of the peak and the trough of the i+n electromagnetic sub-echo;
the second judging unit is used for judging whether the amplitude difference value between the wave crest and the wave trough of the (i+n) th electromagnetic sub-echo is larger than the amplitude change threshold value;
the first storage unit is used for obtaining the total width from the ith electromagnetic sub-echo to the (i+n-1) th electromagnetic sub-echo and storing the total width as a first width if the amplitude difference value between the wave crest and the wave trough of the (i+n) th electromagnetic sub-echo is smaller than the amplitude change threshold;
The second storage unit is used for obtaining the total width from the ith electromagnetic sub-echo to the (i+n) th electromagnetic sub-echo and storing the total width as a second width if the difference value between the wave crest and the wave trough of the (i+n) th electromagnetic sub-echo is larger than the amplitude change threshold;
a sixteenth obtaining unit, configured to sequentially extract remaining historical radar one-dimensional range profiles from the radar one-dimensional range profile dataset, until extraction is completed, to obtain a plurality of first widths and second widths;
and the seventeenth obtaining unit is used for storing the plurality of first widths and the plurality of second widths to obtain one-dimensional convolution estimated sizes of the radar one-dimensional range profile.
Further, the thirteenth obtaining unit includes:
an eighteenth obtaining unit, configured to obtain an amplitude variation average value of a plurality of electromagnetic sub-echoes included in the one-dimensional range profile of the historical radar;
a nineteenth obtaining unit, configured to obtain the amplitude variation threshold according to the amplitude variation average value.
The various variations and specific examples of the radar one-dimensional range profile recognition method based on the multi-dimensional one-dimensional convolution in the first embodiment of fig. 1 are also applicable to the radar one-dimensional range profile recognition device based on the multi-dimensional one-dimensional convolution in this embodiment, and by the foregoing detailed description of the radar one-dimensional range profile recognition method based on the multi-dimensional one-dimensional convolution, those skilled in the art can clearly know the implementation method of the radar one-dimensional range profile recognition device based on the multi-dimensional one-dimensional convolution in this embodiment, so that the description is omitted herein for brevity.
Example III
Based on the same inventive concept as the radar one-dimensional range profile recognition method based on the multi-dimensional one-dimensional convolution in the foregoing embodiment, the present invention further provides a radar one-dimensional range profile recognition device based on the multi-dimensional one-dimensional convolution, on which a computer program is stored, which when executed by a processor, implements the steps of any one of the foregoing methods of the radar one-dimensional range profile recognition method based on the multi-dimensional one-dimensional convolution.
Where in FIG. 3 a bus architecture (represented by bus 300), bus 300 may comprise any number of interconnected buses and bridges, with bus 300 linking together various circuits, including one or more processors, represented by processor 302, and memory, represented by memory 304. Bus 300 may also link together various other circuits such as peripheral devices, voltage regulators, power management circuits, etc., as are well known in the art and, therefore, will not be described further herein. Bus interface 306 provides an interface between bus 300 and receiver 301 and transmitter 303. The receiver 301 and the transmitter 303 may be the same element, i.e. a transceiver, providing a means for communicating with various other apparatus over a transmission medium.
The processor 302 is responsible for managing the bus 300 and general processing, while the memory 304 may be used to store data used by the processor 302 in performing operations.
Example IV
Based on the same inventive concept as the radar one-dimensional range profile recognition method based on multi-dimensional one-dimensional convolution in the foregoing embodiments, the present invention further provides a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, performs the steps of:
calculating angle information of a maneuvering target to be identified; acquiring radar one-dimensional distance image information of the maneuvering target to be identified through a broadband radar, wherein the radar one-dimensional distance image information and the angle information are in one-to-one correspondence; obtaining an angle data set and a radar one-dimensional range profile data set; obtaining a one-dimensional convolution estimated size of the radar one-dimensional range profile according to the radar one-dimensional range profile data set; constructing a multi-dimension one-dimensional convolutional neural network model for automatically identifying the radar one-dimensional range profile according to the one-dimensional convolutional estimated size; training the multi-dimension one-dimensional convolutional neural network model by utilizing the angle data set and the radar one-dimensional range profile data set; and inputting the angle information and the radar one-dimensional range profile information into the trained multi-dimension one-dimensional convolutional neural network model to obtain a radar one-dimensional range profile recognition result of the maneuvering target to be recognized.
In the implementation process, when the program is executed by the processor, any method step in the first embodiment may also be implemented.
The above technical solutions in the embodiments of the present application at least have one or more of the following technical effects:
the embodiment of the application provides a radar one-dimensional range profile identification method and device based on multi-dimension one-dimensional convolution, wherein the method comprises the following steps: calculating angle information of a maneuvering target to be identified; acquiring radar one-dimensional distance image information of the maneuvering target to be identified through a broadband radar, wherein the radar one-dimensional distance image information and the angle information are in one-to-one correspondence; obtaining an angle data set and a radar one-dimensional range profile data set; obtaining a one-dimensional convolution estimated size of the radar one-dimensional range profile according to the radar one-dimensional range profile data set; constructing a multi-dimension one-dimensional convolutional neural network model for automatically identifying the radar one-dimensional range profile according to the one-dimensional convolutional estimated size; training the multi-dimension one-dimensional convolutional neural network model by utilizing the angle data set and the radar one-dimensional range profile data set; and inputting the angle information and the radar one-dimensional range profile information into the trained multi-dimension one-dimensional convolutional neural network model to obtain a radar one-dimensional range profile recognition result of the maneuvering target to be recognized. The method solves the technical problems that in the existing automatic recognition method based on the radar one-dimensional distance image maneuvering target, the recognition rate is reduced due to different data characteristics of target radar one-dimensional distance images of different angles under different radar vision lines, and meanwhile, when the characteristics of the radar one-dimensional distance images are extracted, the convolution kernel size of the radar one-dimensional distance images needs to be manually preset and can only be selected to be the same size, so that the multi-characteristic extraction of the radar one-dimensional distance images is incomplete and has errors, and the maneuvering target recognition rate is low. Through the multi-dimension prediction of one-dimensional convolution, the length characteristics of different distance units of different appearance structures in the one-dimensional range profile of the maneuvering target radar are automatically extracted, the length characteristics are closer to the actual appearance characteristics of the maneuvering target, and meanwhile, the angle information dimension is increased, so that the one-dimensional range profile data characteristics of the maneuvering target under different radar view angles are distinguished, and the technical effect of effectively improving the recognition rate is achieved.
It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may 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 flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may 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 instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may 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 flowchart flow or flows and/or block diagram block or blocks.
It will be apparent to those skilled in the art that various modifications and variations can be made to the present invention without departing from the spirit or scope of the invention. Thus, it is intended that the present invention also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.

Claims (10)

1. A radar one-dimensional range profile identification method based on multi-dimensional one-dimensional convolution, the method comprising:
calculating angle information of a maneuvering target to be identified;
acquiring radar one-dimensional distance image information of the maneuvering target to be identified through a broadband radar, wherein the radar one-dimensional distance image information and the angle information are in one-to-one correspondence;
obtaining an angle data set and a radar one-dimensional range profile data set;
obtaining a one-dimensional convolution estimated size of the radar one-dimensional range profile according to the radar one-dimensional range profile data set;
constructing a multi-dimension one-dimensional convolutional neural network model for automatically identifying the radar one-dimensional range profile according to the one-dimensional convolutional estimated size;
training the multi-dimension one-dimensional convolutional neural network model by utilizing the angle data set and the radar one-dimensional range profile data set;
and inputting the angle information and the radar one-dimensional range profile information into the trained multi-dimension one-dimensional convolutional neural network model to obtain a radar one-dimensional range profile recognition result of the maneuvering target to be recognized.
2. The method of claim 1, wherein the calculating angle information of the maneuver object to be identified comprises:
Calculating pitch angle information of the air platform to the maneuvering target to be identified on the sea/ground;
calculating azimuth information of the motion track direction of the maneuvering target to be identified on the sea/ground;
and obtaining the angle information of the maneuvering target to be identified according to the pitch angle information and the azimuth angle information.
3. The method of claim 2, wherein said calculating pitch angle information of the aerial platform to the maneuvering target to be identified on the sea/ground comprises:
obtaining pitching beams of an aerial platform through a laser radar;
obtaining the maneuvering direction of the maneuvering target to be identified through a narrow-band radar;
and calculating an included angle between the maneuvering direction and the pitching light beam according to the maneuvering direction of the pitching light beam of the aerial platform and the maneuvering target to be identified, and obtaining pitch angle information of the maneuvering target to be identified.
4. A method according to claim 3, wherein said calculating azimuth information of the trajectory direction of said maneuvering target to be identified on the sea/ground comprises:
acquiring azimuth beams of an aerial platform through the laser radar;
and calculating an included angle between the maneuvering direction and the azimuth light beam according to the azimuth light beam of the aerial platform and the maneuvering direction of the maneuvering target to be identified, and obtaining azimuth information of the maneuvering target to be identified.
5. The method of claim 1, wherein constructing the multi-dimensional one-dimensional convolutional neural network model for radar one-dimensional range profile automatic recognition comprises:
the multi-dimension one-dimensional convolution pooling layer is used for carrying out multi-dimension one-dimensional convolution pooling processing on the radar one-dimensional range profile and extracting multi-feature information of different range unit lengths of different appearance structures of targets of the radar one-dimensional range profile;
the full-connection input layer node is formed by combining multi-characteristic information of different distance unit lengths of different outline structures of targets of the radar one-dimensional range profile extracted by the multi-dimension one-dimensional convolution pooling layer and the angle information;
the full-connection middle layer is used for carrying out full connection on nodes of the full-connection input layer;
and the full-connection output layer is in full connection with the nodes of the full-connection middle layer and is used for outputting the identification result of the multi-dimension one-dimensional convolutional neural network model, wherein the number of the nodes of the full-connection output layer is the number of the types of the target identification result.
6. The method of claim 1, wherein the obtaining a one-dimensional convolution estimate of the radar one-dimensional range profile from the radar one-dimensional range profile dataset comprises:
Extracting a historical radar one-dimensional range profile from the radar one-dimensional range profile data set, wherein the historical radar one-dimensional range profile comprises a plurality of scattered electromagnetic sub-echoes;
according to the historical radar one-dimensional range profile, an ith electromagnetic sub-echo is obtained, wherein i=1, 2,3 and …;
obtaining wave crests and wave troughs of the ith electromagnetic sub-echo;
calculating the amplitude difference value of the wave crest and the wave trough of the ith electromagnetic sub-echo;
obtaining an amplitude variation threshold;
judging whether the amplitude difference value of the wave crest and the wave trough of the ith electromagnetic sub-echo is larger than the amplitude change threshold value or not;
if the amplitude difference value of the wave crest and the wave trough of the ith electromagnetic sub-echo is larger than the amplitude change threshold value, obtaining an ith+n electromagnetic sub-echo, wherein n=1, 2,3 and …;
obtaining the wave crest and the wave trough of the (i+n) th electromagnetic sub-echo, and calculating the amplitude difference value of the wave crest and the wave trough of the (i+n) th electromagnetic sub-echo;
judging whether the amplitude difference value of the wave crest and the wave trough of the (i+n) th electromagnetic sub-echo is larger than the amplitude change threshold value or not;
if the amplitude difference value between the wave crest and the wave trough of the ith electromagnetic sub-echo is smaller than the amplitude change threshold value, obtaining the total width from the ith electromagnetic sub-echo to the ith electromagnetic sub-echo plus n-1, and storing the total width as a first width;
If the difference value between the wave crest and the wave trough of the (i+n) th electromagnetic sub-echo is larger than the amplitude change threshold value, obtaining the total width from the (i) th electromagnetic sub-echo to the (i+n) th electromagnetic sub-echo, and storing the total width as a second width;
sequentially extracting the remaining historical radar one-dimensional range profiles from the radar one-dimensional range profile data set until the extraction is completed, and obtaining a plurality of first widths and second widths;
and storing the plurality of first widths and the plurality of second widths to obtain one-dimensional convolution estimated dimensions of the radar one-dimensional range profile.
7. The method of claim 6, wherein the obtaining the amplitude variation threshold comprises:
acquiring the amplitude variation mean value of a plurality of electromagnetic sub-echoes contained in the historical radar one-dimensional range profile;
and obtaining the amplitude variation threshold according to the amplitude variation average value.
8. A radar one-dimensional range profile recognition device based on deep learning, the device comprising:
the first calculation unit is used for calculating angle information of the maneuvering target to be identified;
the first obtaining unit is used for obtaining radar one-dimensional distance image information of the maneuvering target to be identified through a broadband radar, wherein the radar one-dimensional distance image information and the angle information are in one-to-one correspondence;
The second obtaining unit is used for obtaining an angle data set and a radar one-dimensional range profile data set;
the third obtaining unit is used for obtaining the one-dimensional convolution estimated size of the radar one-dimensional range profile according to the radar one-dimensional range profile data set;
the first construction unit is used for constructing a multi-dimension one-dimensional convolutional neural network model for automatically identifying the radar one-dimensional range profile according to the one-dimensional convolutional estimated dimension;
the first training unit is used for training the multi-dimension one-dimensional convolutional neural network model by utilizing the angle data set and the radar one-dimensional range profile data set;
the fourth obtaining unit is used for inputting the angle information and the radar one-dimensional range profile information into the trained multi-dimension one-dimensional convolutional neural network model to obtain a radar one-dimensional range profile recognition result of the maneuvering target to be recognized.
9. A radar one-dimensional range profile recognition device based on deep learning, comprising a memory, a processor and a computer program stored on the memory and operable on the processor, wherein the processor implements the following steps when executing the program:
Calculating angle information of a maneuvering target to be identified;
acquiring radar one-dimensional distance image information of the maneuvering target to be identified through a broadband radar, wherein the radar one-dimensional distance image information and the angle information are in one-to-one correspondence;
obtaining an angle data set and a radar one-dimensional range profile data set;
obtaining a one-dimensional convolution estimated size of the radar one-dimensional range profile according to the radar one-dimensional range profile data set;
constructing a multi-dimension one-dimensional convolutional neural network model for automatically identifying the radar one-dimensional range profile according to the one-dimensional convolutional estimated size;
training the multi-dimension one-dimensional convolutional neural network model by utilizing the angle data set and the radar one-dimensional range profile data set;
and inputting the angle information and the radar one-dimensional range profile information into the trained multi-dimension one-dimensional convolutional neural network model to obtain a radar one-dimensional range profile recognition result of the maneuvering target to be recognized.
10. A computer readable storage medium having stored thereon a computer program, characterized in that the program when executed by a processor performs the steps of:
calculating angle information of a maneuvering target to be identified;
acquiring radar one-dimensional distance image information of the maneuvering target to be identified through a broadband radar, wherein the radar one-dimensional distance image information and the angle information are in one-to-one correspondence;
Obtaining an angle data set and a radar one-dimensional range profile data set;
obtaining a one-dimensional convolution estimated size of the radar one-dimensional range profile according to the radar one-dimensional range profile data set;
constructing a multi-dimension one-dimensional convolutional neural network model for automatically identifying the radar one-dimensional range profile according to the one-dimensional convolutional estimated size;
training the multi-dimension one-dimensional convolutional neural network model by utilizing the angle data set and the radar one-dimensional range profile data set;
and inputting the angle information and the radar one-dimensional range profile information into the trained multi-dimension one-dimensional convolutional neural network model to obtain a radar one-dimensional range profile recognition result of the maneuvering target to be recognized.
CN202110908110.8A 2021-08-09 2021-08-09 Radar one-dimensional range profile identification method and device based on multi-dimension one-dimensional convolution Active CN113640764B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110908110.8A CN113640764B (en) 2021-08-09 2021-08-09 Radar one-dimensional range profile identification method and device based on multi-dimension one-dimensional convolution

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110908110.8A CN113640764B (en) 2021-08-09 2021-08-09 Radar one-dimensional range profile identification method and device based on multi-dimension one-dimensional convolution

Publications (2)

Publication Number Publication Date
CN113640764A CN113640764A (en) 2021-11-12
CN113640764B true CN113640764B (en) 2023-08-11

Family

ID=78420174

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110908110.8A Active CN113640764B (en) 2021-08-09 2021-08-09 Radar one-dimensional range profile identification method and device based on multi-dimension one-dimensional convolution

Country Status (1)

Country Link
CN (1) CN113640764B (en)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109828251A (en) * 2019-03-07 2019-05-31 中国人民解放军海军航空大学 Radar target identification method based on feature pyramid light weight convolutional neural networks
CN109902589A (en) * 2019-01-30 2019-06-18 南京航空航天大学 A kind of target identification method and system based on Radar range profile's
CN111352086A (en) * 2020-03-06 2020-06-30 电子科技大学 Unknown target identification method based on deep convolutional neural network
AU2020104006A4 (en) * 2020-12-10 2021-02-18 Naval Aviation University Radar target recognition method based on feature pyramid lightweight convolutional neural network

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109902589A (en) * 2019-01-30 2019-06-18 南京航空航天大学 A kind of target identification method and system based on Radar range profile's
CN109828251A (en) * 2019-03-07 2019-05-31 中国人民解放军海军航空大学 Radar target identification method based on feature pyramid light weight convolutional neural networks
CN111352086A (en) * 2020-03-06 2020-06-30 电子科技大学 Unknown target identification method based on deep convolutional neural network
AU2020104006A4 (en) * 2020-12-10 2021-02-18 Naval Aviation University Radar target recognition method based on feature pyramid lightweight convolutional neural network

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
王容川.高分辨一维距离像引信技术.《中国优秀硕士学位论文全文数据库 工程科技II辑》.2022,第2022卷(第1期),C032-96. *

Also Published As

Publication number Publication date
CN113640764A (en) 2021-11-12

Similar Documents

Publication Publication Date Title
CN113902897B (en) Training of target detection model, target detection method, device, equipment and medium
CN107972662B (en) Vehicle forward collision early warning method based on deep learning
CN111326023B (en) Unmanned aerial vehicle route early warning method, device, equipment and storage medium
CN109087510B (en) Traffic monitoring method and device
Wheeler et al. Deep stochastic radar models
US20130004017A1 (en) Context-Based Target Recognition
CN113264066A (en) Obstacle trajectory prediction method and device, automatic driving vehicle and road side equipment
EP3575827A1 (en) Method for robust estimation of the velocity of a target using a host vehicle
CN113378760A (en) Training target detection model and method and device for detecting target
CN102928822A (en) Radar target length calculation method based on high-resolution one-dimensional range profiles
Patil et al. A survey on joint object detection and pose estimation using monocular vision
CN113093128A (en) Method and device for calibrating millimeter wave radar, electronic equipment and road side equipment
CN114137526A (en) Label-based vehicle-mounted millimeter wave radar multi-target detection method and system
US20220245835A1 (en) Geo-motion and appearance aware data association
CN113408504B (en) Lane line identification method and device based on radar, electronic equipment and storage medium
CN113640764B (en) Radar one-dimensional range profile identification method and device based on multi-dimension one-dimensional convolution
CN115586506B (en) Anti-interference target classification method and device
CN114842305A (en) Depth prediction model training method, depth prediction method and related device
CN113514825A (en) Road edge obtaining method and device and terminal equipment
CN113642448B (en) Image recognition method and device for sea/ground maneuvering target by air platform
CN115128597B (en) Maneuvering target tracking method under non-Gaussian noise based on IMM-STEKF
CN118134856A (en) Data evaluation method, device, equipment and medium
CN114140699B (en) SAR image target identification method and system based on online local Fisher vector machine
CN115728735A (en) Aircraft push-out collision avoidance method and system based on laser point cloud
Subash Unscented particle filter in road extraction from high resoltuion satellite images

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant