CN111273288B - Radar unknown target identification method based on long-term and short-term memory network - Google Patents

Radar unknown target identification method based on long-term and short-term memory network Download PDF

Info

Publication number
CN111273288B
CN111273288B CN202010151942.5A CN202010151942A CN111273288B CN 111273288 B CN111273288 B CN 111273288B CN 202010151942 A CN202010151942 A CN 202010151942A CN 111273288 B CN111273288 B CN 111273288B
Authority
CN
China
Prior art keywords
long
dimensional range
range profile
term memory
distance
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
CN202010151942.5A
Other languages
Chinese (zh)
Other versions
CN111273288A (en
Inventor
周代英
张同梦雪
胡晓龙
李粮余
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.)
University of Electronic Science and Technology of China
Original Assignee
University of Electronic Science and Technology of China
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 University of Electronic Science and Technology of China filed Critical University of Electronic Science and Technology of China
Priority to CN202010151942.5A priority Critical patent/CN111273288B/en
Publication of CN111273288A publication Critical patent/CN111273288A/en
Application granted granted Critical
Publication of CN111273288B publication Critical patent/CN111273288B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

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
    • 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
    • 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
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/044—Recurrent networks, e.g. Hopfield networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Remote Sensing (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Biophysics (AREA)
  • Mathematical Physics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Biomedical Technology (AREA)
  • Computational Linguistics (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Electromagnetism (AREA)
  • Health & Medical Sciences (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Image Analysis (AREA)
  • Radar Systems Or Details Thereof (AREA)

Abstract

The invention belongs to the technical field of unknown target identification, and particularly relates to a radar unknown target identification method based on a long-term and short-term memory network. The method comprises the steps of firstly preprocessing one-dimensional range profile data (HRRP) of a broadband radar; and secondly, performing Principal Component Analysis (PCA) on the feature vector extracted from the one-dimensional range profile by the long and short term memory network, reducing the dimension of the feature vector, finally processing the low-dimensional feature vector of the known target sample data by a nearest neighbor method (NN), acquiring a discrimination threshold, and discriminating the output vector of the long and short term memory network, thereby identifying the unknown target. For a conventional convolutional network, the long-term and short-term memory network can extract the related information between the adjacent one-dimensional range profile samples, and effectively describes the change characteristics between the input one-dimensional range profile sample sequences, so that the performance of identifying unknown targets is improved.

Description

Radar unknown target identification method based on long-term and short-term memory network
Technical Field
The invention belongs to the technical field of unknown target identification, and particularly relates to a radar unknown target identification method based on a long-term and short-term memory network.
Background
The one-dimensional distance image (HRRP) reflects the structural characteristics of the size, the scattering center distribution and the like of the target, has the advantages of strong real-time performance, easiness in acquisition and storage and the like, and is a main means for identifying the air targets such as airplanes and the like at present.
With the successful application of deep learning theory in the fields of voice, image, natural language processing and the like, some learners begin to introduce a deep learning method into the aspect of radar target one-dimensional range profile recognition. For example, a convolutional neural network with translation invariance characteristics has a good recognition rate on a trained radar target one-dimensional range profile, but cannot correctly recognize an unknown target (i.e., a target which does not participate in training), and for this reason, the recognition of the unknown target can be realized by using a deep convolutional neural network + a threshold.
Disclosure of Invention
The invention provides a radar unknown target identification method based on a deep convolutional neural network, aiming at the problems. The method adopts the long-short term memory network to extract the correlation characteristics between adjacent samples in the input sequence, and effectively describes the characteristic information related to the target structure, thereby improving the recognition rate of unknown targets.
The technical scheme of the invention is as follows: a radar unknown target identification method based on a long-short term memory network comprises the following steps:
s1, based on the target scattering center model, setting the single target one-dimensional range profile sample obtained by the broadband radar as x ═ x1,x2,...,xi,…,xn]Wherein n is the number of distance units, xiRepresenting the amplitude of the ith distance unit, highlighting the contrast effect of the strong scattering point and other scattering points in order to reduce the influence of the amplitude sensitivity of the one-dimensional distance image on the identification performance, and carrying out beta-mean value standardization treatment on the one-dimensional distance image:
Figure BDA0002402756910000011
wherein
Figure BDA0002402756910000021
Represents the normalized amplitude of the ith distance element, beta is a constant, ExThe mean value of the single range profile is shown, and the single one-dimensional range profile after beta-mean value standardization processing is
Figure BDA0002402756910000022
S2, constructing a long and short term memory neural network model, and modifying the long and short term memory neural network (LSTM) on the basis of a recurrent neural network, wherein the problem of gradient disappearance can be solved to a certain extent, as shown in figure 1, the long and short term memory neural network of the invention has 11 layers in total, namely an LSTM layer, a Dropout layer 1, a Flatten layer, a full connection layer 1, a Dropout layer 2, a full connection layer 3, a Dropout layer 3, a full connection layer 4, a full connection layer 5 and a classifier in sequence; besides a single LSTM layer and a classification output layer with softmax as an activation function, a plurality of full-connection layers and dropouts are contained to enable part of neurons to be inactivated randomly so as to reduce training parameters and reduce the risk of model overfitting; preprocessing a plurality of one-dimensional range profiles and inputting the preprocessed one-dimensional range profiles into a network in time sequence, considering radar minimum resolution unit and HRRP data characteristics, setting a time step length to be 1, and inputting long-short term memory neural network as
Figure BDA0002402756910000023
Outputting identification label given for classifier
Figure BDA0002402756910000024
Wherein
Figure BDA0002402756910000025
Is a one-dimensional range profile preprocessed at the mth moment, m is more than or equal to 1 and less than or equal to N, N is the number of the one-dimensional range profiles,
Figure BDA0002402756910000026
the identification label corresponding to the one-dimensional range profile preprocessed at the mth moment; the basic units of the LSTM layer are memory units, and each memory unit comprises an input gate, a forgetting gate and an output gate III as shown in FIG. 2An element, the m-th time memory unit has inputs of
Figure BDA0002402756910000027
The memory cell has an output hmAnd a memory value cm，
Figure BDA0002402756910000028
The output value h of the memory cell at the last moment is memorized through an input gate, a forgetting gate and an output gatem-1And the memory value cm-1By the combined action ofmThe memory cells are connected in series as shown in fig. 3, specifically:
Figure BDA0002402756910000029
Figure BDA00024027569100000210
Figure BDA00024027569100000211
Figure BDA00024027569100000212
hm＝om⊙R(cm)
wherein imIs an input gate, fmTo forget the door or omIs an output gate, WxiAnd VhiTo input the gate weight parameters, biFor input of gate bias terms, WxfAnd VhfTo forget the weight parameter of the door, bfTo forget the gate bias term, WxcAnd VhcTo memorize the value weight parameters, bcFor memorizing a value offset term, WxoAnd VhoTo output the gate weight parameters, boBiasing an item for an output gate, indicating the product of corresponding elements in the vector, σ (-) as a sigmoid activation function, R (-) as a relu activation function, passing through the LSTM layerObtaining 1024-dimensional feature vectors, repeatedly extracting the feature vectors from a plurality of full-connected layers with the activation function relu, sequentially reducing the dimensions in the order of 800,500,100,50 and 20, and finally outputting identification tags through a softmax function
Figure BDA0002402756910000031
S3, determining an identification threshold: for a conventional neural network object classification task, an unknown class object will be forcibly identified as a known class in the classification. In order to identify unknown targets, the invention adopts a Principal Component Analysis (PCA) method and a nearest neighbor method (NN) to combine to obtain an identification threshold, which specifically comprises the following steps:
a feature vector Z ═ Z with dimension 20 can be obtained in the last fully-connected layer of the LSTM network1,z2,…,z20]. In order to reduce subsequent calculation amount, the Principal Component Analysis (PCA) is used for reducing the dimension of the feature vector, and Z is reduced to two dimensions to obtain
Figure BDA0002402756910000032
Handle
Figure BDA0002402756910000033
As a coordinate on the horizontal axis,
Figure BDA0002402756910000034
performing scattered point visualization analysis after the coordinate of the longitudinal axis is taken as a coordinate, and obtaining a clustering image of each known target class;
dividing a training one-dimensional range profile data set of a known target into two parts, training a long-short term memory neural network by using first part data, and obtaining a characteristic vector Z with the dimension of 20 through a full connection layer 5 of the long-short term memory neural network for each one-dimensional range profile1,z2,…,z20]Reducing the dimension of the feature vector by using a principal component analysis method, and reducing Z to two dimensions to obtain the feature vector
Figure BDA0002402756910000035
Reducing the dimension of the feature vector obtained after the first part of data is trained to form a feature vector modelA plate warehouse; extracting feature vectors corresponding to the second part of one-dimensional distance image data by using the long and short term memory neural network trained by the first part of data, reducing dimensions, calculating the shortest distance between the corresponding feature vectors and a library feature vector template to form a shortest distance vector d ═ d [ [ d ] ] 1,d2,...,dk,…,dM]TWherein d iskK is more than or equal to 1 and less than or equal to M, and M is the number of the one-dimensional distance images of the second part;
calculating a closest distance histogram by using elements in the closest distance vector d, and selecting a distance value from the closest distance histogram as an identification threshold tau according to the correct discrimination rate of a predetermined known target;
s4, unknown target identification:
one-dimensional range profile data x of target to be recognizedtInputting the normalized beta-mean value into a trained long and short term memory neural network model, extracting and reducing dimensions to obtain a feature vector
Figure BDA0002402756910000036
Recalculation
Figure BDA0002402756910000037
Closest distance d to library feature vector templatet(ii) a If the nearest distance is greater than or equal to the threshold, dtIf the distance is more than or equal to tau, the one-dimensional distance image x to be recognizedtIdentified as an unknown target; if the nearest distance is less than the threshold, dt<Tau, then one-dimensional range profile x to be recognizedtAre identified as known targets.
The method has the advantages that the discrimination threshold obtained by adopting a difference probability method is introduced, so that the statistical distribution area boundary of the known target and the unknown target data set is effectively described, and the problem that the conventional convolutional neural network cannot identify the unknown target is solved.
Drawings
FIG. 1 is a diagram of a long-term and short-term memory neural network model;
FIG. 2 is a diagram of an LSTM memory cell;
FIG. 3 is a series diagram of LSTM memory cells.
Detailed Description
The effectiveness of the invention is demonstrated below in connection with the simulation example.
Experiments are carried out by utilizing simulated one-dimensional distance images of five different types of military aircrafts including AH64, AN26, F15, B1B and B52 obtained by a special electromagnetic simulation characteristic scene. The experimental simulation radar parameters comprise: the radar carrier frequency is 6GHz, and the radar bandwidth is 400 MHz. In the simulation scene, a simulation target collects a one-dimensional range profile at an interval of 0.1 degrees in the range of 0-180 degrees of azimuth angle at an elevation angle of 3 degrees, each type of airplane collects 1801 one-dimensional range profiles, each one-dimensional range profile contains 320 range units, namely, input data of each type of airplane is a 1801 × 320 one-dimensional range profile matrix.
In the process of training the updated parameters, the weight W is initialized randomlyxi、Vhi、Wxf、Vhf、Wxc、Vhc、Wxo、VhoAnd offset bi、bf、bc、boAnd selecting a cross entropy loss function as a loss function and optimization parameters of an Adam optimizer with an adaptive learning rate, wherein the learning rate is initialized to 0.0001.
The identification results of the above 5 types of simulated radar target data using the conventional convolutional neural network method and the method herein are shown in table 1:
TABLE 1 recognition results of two methods on unknown targets
Figure BDA0002402756910000041
Figure BDA0002402756910000051
From experimental results, under the condition that three types of airplanes are randomly extracted as known targets and the other two types of airplanes are unknown targets, the average unknown target recognition rate is about 81% by using a deep convolutional neural network + threshold method, the correlation characteristics between adjacent one-dimensional distance images are extracted by using a long-short term memory network, the variation characteristics between sample sequences are effectively described, the recognition capability of the unknown targets is further improved, the average correct recognition rate of the unknown targets is over 95%, and therefore the method is verified to be effective.

Claims (1)

1. A radar unknown target identification method based on a long-short term memory network is characterized by comprising the following steps:
s1, setting a single target one-dimensional range profile sample obtained by the broadband radar as x ═ x based on the target scattering center model1,x2,…,xi,…,xn]Where n is the number of distance units, xiRepresenting the amplitude of the ith range bin, and carrying out beta-mean normalization processing on the one-dimensional range profile:
Figure FDA0002402756900000011
wherein
Figure FDA0002402756900000012
Represents the normalized amplitude of the ith distance element, beta is a constant, ExThe mean value of the single range profile is shown, and the single one-dimensional range profile after beta-mean value standardization processing is
Figure FDA0002402756900000013
S2, constructing a long-short term memory neural network model, wherein the long-short term memory neural network has 11 layers in total, and the long-short term memory neural network model sequentially comprises an LSTM layer, a Dropout layer 1, a Flatten layer, a full connection layer 1, a Dropout layer 2, a full connection layer 3, a Dropout layer 3, a full connection layer 4, a full connection layer 5 and a classifier; the input of the long-short term memory neural network is
Figure FDA0002402756900000014
Output as classificationIdentification tag given by device
Figure FDA0002402756900000015
Wherein
Figure FDA0002402756900000016
Is a one-dimensional range profile preprocessed at the mth moment, m is more than or equal to 1 and less than or equal to N, N is the number of the one-dimensional range profiles,
Figure FDA0002402756900000017
the identification label corresponding to the one-dimensional range profile preprocessed at the mth moment; the basic unit of the LSTM layer is a memory unit, each memory unit comprises an input gate, a forgetting gate and an output gate, and the input of the memory unit at the mth moment is
Figure FDA0002402756900000018
The memory cell has an output hmAnd a memory value cm，
Figure FDA0002402756900000019
The output value h of the memory cell at the last moment is memorized through an input gate, a forgetting gate and an output gatem-1And the memory value cm-1By the combined action ofmThe method specifically comprises the following steps:
Figure FDA00024027569000000110
Figure FDA00024027569000000111
Figure FDA00024027569000000112
Figure FDA00024027569000000113
hm＝om⊙R(cm)
wherein imIs an input gate, fmTo forget the door or omIs an output gate, WxiAnd VhiTo input the gate weight parameters, biFor input of gate bias terms, WxfAnd VhfTo forget the weight parameter of the door, bfTo forget the gate bias term, W xcAnd VhcTo memorize the value weight parameter, bcFor memorizing a value offset term, WxoAnd VhoTo output the gate weight parameters, boFor output gate bias items, indicating the product of corresponding elements in a vector, sigma (-) being a sigmoid activation function, R (-) being a relu activation function, obtaining 1024-dimensional feature vectors through an LSTM layer, repeatedly extracting feature vectors from a fully-connected layer with multiple activation functions as relu, sequentially reducing dimensions in the order of 800,500,100,50,20, and finally outputting identification tags through a softmax function
Figure FDA0002402756900000021
S3, determining an identification threshold: dividing a training one-dimensional range profile data set of a known target into two parts, training a long-short term memory neural network by using first part data, and obtaining a characteristic vector Z with the dimension of 20 through a full connection layer 5 of the long-short term memory neural network for each one-dimensional range profile1,z2,…,z20]Reducing the dimension of the feature vector by using a principal component analysis method, and reducing Z to two dimensions to obtain the feature vector
Figure FDA0002402756900000022
Reducing the dimension of the feature vector obtained after the first part of data is trained to form a feature vector template library; extracting feature vectors corresponding to the second part of one-dimensional distance image data by using the long and short term memory neural network trained by the first part of data, reducing dimensions, calculating the shortest distance between the corresponding feature vectors and a library feature vector template to form a shortest distance vector d ═ d [ [ d ] ] 1,d2,...,dk,...,dM]TWherein d iskK is more than or equal to 1 and less than or equal to M, and M is the number of the one-dimensional range profiles of the second part;
calculating a closest distance histogram by using elements in the closest distance vector d, and selecting a distance value from the closest distance histogram as an identification threshold tau according to the correct discrimination rate of a predetermined known target;
s4, unknown target identification:
one-dimensional range profile data x of target to be recognizedtInputting the normalized beta-mean value into a trained long and short term memory neural network model, extracting and reducing dimensions to obtain a feature vector
Figure FDA0002402756900000023
Recalculation
Figure FDA0002402756900000024
Closest distance d to library feature vector templatet(ii) a If the nearest distance is greater than or equal to the threshold, dtIf the distance is more than or equal to tau, the one-dimensional distance image x to be recognizedtIdentified as an unknown target; if the nearest distance is less than the threshold, dt<Tau, then one-dimensional range profile x to be recognizedtAre identified as known targets.
CN202010151942.5A 2020-03-06 2020-03-06 Radar unknown target identification method based on long-term and short-term memory network Active CN111273288B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010151942.5A CN111273288B (en) 2020-03-06 2020-03-06 Radar unknown target identification method based on long-term and short-term memory network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010151942.5A CN111273288B (en) 2020-03-06 2020-03-06 Radar unknown target identification method based on long-term and short-term memory network

Publications (2)

Publication Number Publication Date
CN111273288A CN111273288A (en) 2020-06-12
CN111273288B true CN111273288B (en) 2022-05-24

Family

ID=70999428

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010151942.5A Active CN111273288B (en) 2020-03-06 2020-03-06 Radar unknown target identification method based on long-term and short-term memory network

Country Status (1)

Country Link
CN (1) CN111273288B (en)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112014821B (en) * 2020-08-27 2022-05-17 电子科技大学 Unknown vehicle target identification method based on radar broadband characteristics
CN112163636B (en) * 2020-10-15 2023-09-26 电子科技大学 Unknown mode identification method of electromagnetic signal radiation source based on twin neural network
CN114821335B (en) * 2022-05-20 2023-04-25 电子科技大学 Unknown target discrimination method based on fusion of depth features and linear discrimination features
CN114943286B (en) * 2022-05-20 2023-04-07 电子科技大学 Unknown target discrimination method based on fusion of time domain features and space domain features

Citations (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH09152480A (en) * 1995-11-30 1997-06-10 Mitsubishi Electric Corp Automatic target recognition apparatus
CN107463966A (en) * 2017-08-17 2017-12-12 电子科技大学 Radar range profile's target identification method based on dual-depth neutral net
CN107545279A (en) * 2017-08-30 2018-01-05 电子科技大学 Image-recognizing method based on convolutional neural networks Yu Weighted Kernel signature analysis
CN107609601A (en) * 2017-09-28 2018-01-19 北京计算机技术及应用研究所 A kind of ship seakeeping method based on multilayer convolutional neural networks
CN107862329A (en) * 2017-10-31 2018-03-30 电子科技大学 A kind of true and false target identification method of Radar range profile's based on depth confidence network
CN108267724A (en) * 2018-01-19 2018-07-10 中国人民解放军陆军装甲兵学院 A kind of unknown object recognition methods of radar target recognition
CN108776779A (en) * 2018-05-25 2018-11-09 西安电子科技大学 SAR Target Recognition of Sequential Images methods based on convolution loop network
CN109086700A (en) * 2018-07-20 2018-12-25 杭州电子科技大学 Radar range profile's target identification method based on depth convolutional neural networks
CN109376574A (en) * 2018-08-14 2019-02-22 西安电子科技大学 Refuse to sentence radar HRRP target identification method based on CNN
EP3525131A1 (en) * 2018-02-09 2019-08-14 Bayerische Motoren Werke Aktiengesellschaft Methods and apparatuses for object detection in a scene represented by depth data of a range detection sensor and image data of a camera
CN110334741A (en) * 2019-06-06 2019-10-15 西安电子科技大学 Radar range profile's recognition methods based on Recognition with Recurrent Neural Network
CN110363151A (en) * 2019-07-16 2019-10-22 中国人民解放军海军航空大学 Based on the controllable radar target detection method of binary channels convolutional neural networks false-alarm
CN110738984A (en) * 2019-05-13 2020-01-31 苏州闪驰数控***集成有限公司 Artificial intelligence CNN, LSTM neural network speech recognition system

Patent Citations (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH09152480A (en) * 1995-11-30 1997-06-10 Mitsubishi Electric Corp Automatic target recognition apparatus
CN107463966A (en) * 2017-08-17 2017-12-12 电子科技大学 Radar range profile's target identification method based on dual-depth neutral net
CN107545279A (en) * 2017-08-30 2018-01-05 电子科技大学 Image-recognizing method based on convolutional neural networks Yu Weighted Kernel signature analysis
CN107609601A (en) * 2017-09-28 2018-01-19 北京计算机技术及应用研究所 A kind of ship seakeeping method based on multilayer convolutional neural networks
CN107862329A (en) * 2017-10-31 2018-03-30 电子科技大学 A kind of true and false target identification method of Radar range profile's based on depth confidence network
CN108267724A (en) * 2018-01-19 2018-07-10 中国人民解放军陆军装甲兵学院 A kind of unknown object recognition methods of radar target recognition
EP3525131A1 (en) * 2018-02-09 2019-08-14 Bayerische Motoren Werke Aktiengesellschaft Methods and apparatuses for object detection in a scene represented by depth data of a range detection sensor and image data of a camera
CN108776779A (en) * 2018-05-25 2018-11-09 西安电子科技大学 SAR Target Recognition of Sequential Images methods based on convolution loop network
CN109086700A (en) * 2018-07-20 2018-12-25 杭州电子科技大学 Radar range profile's target identification method based on depth convolutional neural networks
CN109376574A (en) * 2018-08-14 2019-02-22 西安电子科技大学 Refuse to sentence radar HRRP target identification method based on CNN
CN110738984A (en) * 2019-05-13 2020-01-31 苏州闪驰数控***集成有限公司 Artificial intelligence CNN, LSTM neural network speech recognition system
CN110334741A (en) * 2019-06-06 2019-10-15 西安电子科技大学 Radar range profile's recognition methods based on Recognition with Recurrent Neural Network
CN110363151A (en) * 2019-07-16 2019-10-22 中国人民解放军海军航空大学 Based on the controllable radar target detection method of binary channels convolutional neural networks false-alarm

Non-Patent Citations (8)

* Cited by examiner, † Cited by third party
Title
Nasrabadi, NM.Automatic target recognition using deep convolutional neural networks.《AUTOMATIC TARGET RECOGNITION XXVIII》.2019, *
Wang, XQ.Multi-Object Detection in Traffic Scenes Based on Improved SSD.《ELECTRONICS》.2018, *
Zhou D.Radar target HRRP recognition based on reconstructive and discriminative dictionary learning.《Signal Processing》.2016, *
周代英.基于二阶模糊聚类算法的雷达目标距离像识别.《计算机应用》.2011, *
周毅.基于高分辨一维距离像的目标特征提取及融合识别研究.《中国优秀硕士学位论文全文数据库 信息科技辑》.2018,(第9期), *
李文辉.基于深度学习的SAR图像目标识别.《万方数据库》.2018, *
王泓霖.基于PSO-BP神经网络的HRRP目标识别.《火力与指挥控制》.2019, *
盖晴晴.极化雷达目标识别与多分类器融合研究.《万方数据库》.2018, *

Also Published As

Publication number Publication date
CN111273288A (en) 2020-06-12

Similar Documents

Publication Publication Date Title
CN111273288B (en) Radar unknown target identification method based on long-term and short-term memory network
CN110334741B (en) Radar one-dimensional range profile identification method based on cyclic neural network
US10713563B2 (en) Object recognition using a convolutional neural network trained by principal component analysis and repeated spectral clustering
CN106599797B (en) A kind of infrared face recognition method based on local parallel neural network
Chen et al. SAR target recognition based on deep learning
CN103489005B (en) A kind of Classification of High Resolution Satellite Images method based on multiple Classifiers Combination
CN111352086B (en) Unknown target identification method based on deep convolutional neural network
CN110569793A (en) Target tracking method for unsupervised similarity discrimination learning
US11150327B1 (en) System and method for synthetic aperture radar target recognition using multi-layer, recurrent spiking neuromorphic networks
CN111753874A (en) Image scene classification method and system combined with semi-supervised clustering
CN109766934B (en) Image target identification method based on depth Gabor network
CN110018453A (en) Intelligent type recognition methods based on aircraft track feature
CN114898151A (en) Image classification method based on deep learning and support vector machine fusion
Arzar et al. Butterfly species identification using convolutional neural network (CNN)
Xiao et al. Specific emitter identification of radar based on one dimensional convolution neural network
Shang et al. Image spam classification based on convolutional neural network
CN115393631A (en) Hyperspectral image classification method based on Bayesian layer graph convolution neural network
CN114973019A (en) Deep learning-based geospatial information change detection classification method and system
Astuti et al. Multilevel thresholding hyperspectral image segmentation based on independent component analysis and swarm optimization methods
CN112990279B (en) Radar high-resolution range profile library outside target rejection method based on automatic encoder
CN114624715A (en) Radar echo extrapolation method based on self-attention space-time neural network model
CN113128387A (en) Drug addict drug addiction attack identification method based on facial expression feature analysis
Qu et al. Frequency-Hopping Signal Sorting Based on Deep Leaming
Blomerus et al. Improved explainability through uncertainty estimation in automatic target recognition of SAR images
Sivachandiran et al. Automated deep learning based age and gender classification model using facial features for video surveillance

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