CN116660851A - Method and system for distinguishing targets of birds and rotor unmanned aerial vehicle under low signal-to-noise ratio condition - Google Patents
Method and system for distinguishing targets of birds and rotor unmanned aerial vehicle under low signal-to-noise ratio condition Download PDFInfo
- Publication number
- CN116660851A CN116660851A CN202310638119.0A CN202310638119A CN116660851A CN 116660851 A CN116660851 A CN 116660851A CN 202310638119 A CN202310638119 A CN 202310638119A CN 116660851 A CN116660851 A CN 116660851A
- Authority
- CN
- China
- Prior art keywords
- unmanned aerial
- bird
- aerial vehicle
- rotor unmanned
- motion
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 96
- 230000033001 locomotion Effects 0.000 claims abstract description 130
- 238000001228 spectrum Methods 0.000 claims abstract description 72
- 230000009467 reduction Effects 0.000 claims abstract description 68
- 239000011159 matrix material Substances 0.000 claims abstract description 57
- 238000006243 chemical reaction Methods 0.000 claims abstract description 46
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 30
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 23
- 238000002592 echocardiography Methods 0.000 claims abstract description 16
- 238000004364 calculation method Methods 0.000 claims abstract description 13
- 239000013598 vector Substances 0.000 claims description 59
- 238000005457 optimization Methods 0.000 claims description 15
- 230000008569 process Effects 0.000 claims description 12
- 230000002452 interceptive effect Effects 0.000 claims description 11
- 238000000605 extraction Methods 0.000 claims description 9
- 230000002708 enhancing effect Effects 0.000 claims description 6
- 230000003190 augmentative effect Effects 0.000 claims description 5
- 230000008602 contraction Effects 0.000 claims description 4
- 230000003595 spectral effect Effects 0.000 claims description 4
- 238000012935 Averaging Methods 0.000 claims description 3
- 238000012545 processing Methods 0.000 abstract description 6
- 238000010586 diagram Methods 0.000 description 47
- 241000271566 Aves Species 0.000 description 43
- 238000004088 simulation Methods 0.000 description 22
- 230000006870 function Effects 0.000 description 14
- 230000004927 fusion Effects 0.000 description 11
- 238000004590 computer program Methods 0.000 description 10
- 238000007499 fusion processing Methods 0.000 description 6
- 238000005259 measurement Methods 0.000 description 6
- 230000000694 effects Effects 0.000 description 4
- 102100031262 Deleted in malignant brain tumors 1 protein Human genes 0.000 description 3
- 101000844721 Homo sapiens Deleted in malignant brain tumors 1 protein Proteins 0.000 description 3
- 230000001133 acceleration Effects 0.000 description 3
- 238000013527 convolutional neural network Methods 0.000 description 3
- ZINJLDJMHCUBIP-UHFFFAOYSA-N ethametsulfuron-methyl Chemical compound CCOC1=NC(NC)=NC(NC(=O)NS(=O)(=O)C=2C(=CC=CC=2)C(=O)OC)=N1 ZINJLDJMHCUBIP-UHFFFAOYSA-N 0.000 description 3
- 238000003860 storage Methods 0.000 description 3
- 230000009466 transformation Effects 0.000 description 3
- 238000010276 construction Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 230000001965 increasing effect Effects 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 230000008092 positive effect Effects 0.000 description 2
- 238000011946 reduction process Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000033764 rhythmic process Effects 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- NAWXUBYGYWOOIX-SFHVURJKSA-N (2s)-2-[[4-[2-(2,4-diaminoquinazolin-6-yl)ethyl]benzoyl]amino]-4-methylidenepentanedioic acid Chemical compound C1=CC2=NC(N)=NC(N)=C2C=C1CCC1=CC=C(C(=O)N[C@@H](CC(=C)C(O)=O)C(O)=O)C=C1 NAWXUBYGYWOOIX-SFHVURJKSA-N 0.000 description 1
- 208000031427 Foetal heart rate deceleration Diseases 0.000 description 1
- 230000001174 ascending effect Effects 0.000 description 1
- 238000010009 beating Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 210000000245 forearm Anatomy 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000009545 invasion Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
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
-
- 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
- G01S13/66—Radar-tracking systems; Analogous systems
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/14—Fourier, Walsh or analogous domain transformations, e.g. Laplace, Hilbert, Karhunen-Loeve, transforms
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/10—Pre-processing; Data cleansing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- 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
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Networks & Wireless Communication (AREA)
- Mathematical Physics (AREA)
- Mathematical Analysis (AREA)
- Computational Mathematics (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Probability & Statistics with Applications (AREA)
- Algebra (AREA)
- Databases & Information Systems (AREA)
- Software Systems (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
The invention belongs to the technical field of unmanned aerial vehicle information identification, and discloses a method and a system for distinguishing targets of birds and rotor unmanned aerial vehicles under the condition of low signal-to-noise ratio. The method is based on a tracking noise reduction method, radar echoes of targets of birds and rotor unmanned aerial vehicles are reconstructed, and noise-reduced echo signals are obtained; tracking two targets by using the obtained reconstructed echo signals and adopting an IMM algorithm, and extracting the motion model conversion frequency of the bird and rotor unmanned aerial vehicle targets as motion characteristics; using the obtained reconstructed echo signals, obtaining a time-frequency matrix of the bird and the rotor unmanned aerial vehicle by using STFT, performing singular decomposition on the time-frequency matrix, and constructing a characteristic spectrum energy entropy as a micro-motion characteristic; and constructing a fused characteristic quantity by utilizing the obtained motion model conversion frequency and characteristic spectrum energy entropy, and inputting the fused characteristic quantity into K-means to realize the identification of birds and rotor unmanned aerial vehicles. The invention has the advantages of simple processing steps and high calculation speed under the condition of low signal-to-noise ratio.
Description
Technical Field
The invention belongs to the technical field of unmanned aerial vehicle information identification, and particularly relates to a method and a system for identifying targets of birds and rotor unmanned aerial vehicles under the condition of low signal-to-noise ratio.
Background
With the rapid development of the unmanned aerial vehicle industry, the potential threat of unmanned aerial vehicles is also increasing. Meanwhile, with the continuous increase of the flight volume and the continuous improvement of the ecological environment, the bird strike prevention working pressure of the airport is increased. The airport takes different countermeasures after different invasion targets such as birds, rotor unmanned aerial vehicle are found, so that the identification of the birds and the rotor unmanned aerial vehicle targets has important significance for improving the monitoring level of the rotor unmanned aerial vehicle and the bird targets and guaranteeing the safety of civil aviation flight.
Due to the fact that radars have all-weather all-day-time working capacity, classification and identification of rotary-wing unmanned aerial vehicles and bird targets by using radars are increasingly paid attention to. Radar target micro-doppler features describe the motion characteristics of targets with rotation, vibration, etc., such as unmanned aerial vehicle blade rotation, wing flapping, and human hand or leg swing, etc., that have been widely used for classification and identification of targets. However, the technical defects of the prior art are that: unmanned aerial vehicle and bird often receive the influence of noise in actual flight environment, and its corresponding radar echo is weaker relatively, and then makes bird and unmanned aerial vehicle target discrimination performance reduce. The identification of rotorcraft and avian targets at low signal-to-noise ratios has become a research hotspot in radar target classification and detection.
The prior art has developed related research efforts in the identification of rotorcraft and avian targets. The method mainly comprises a distinguishing method based on feature extraction and a distinguishing method combining the feature extraction with machine learning. Features extracted in the bird and unmanned aerial vehicle target distinguishing method based on feature extraction comprise radar echo spectrum symmetrical pairs, motion model conversion frequency, a first left singular vector corresponding to singular value decomposition (Singular Value Decomposition, SVD) and a rhythm velocity spectrogram (Cadence Velocity Diagram, CVD), target speed, spectrum period and spectrum width, average value of singular values, variance of singular values and the like. The distinguishing method combining feature extraction and machine learning comprises the steps of combining feature vectors formed by the first two large feature values corresponding to SVD, combining a support vector machine (Support Vector Machine, SVM) to achieve target classification, combining a Micro Doppler bandwidth with a K-Nearest Neighbor (KNN) classifier to achieve target distinguishing, utilizing a combined Doppler image obtained by Micro-Doppler Signatures (MDS) and a rhythm speed map to conduct target distinguishing, analyzing Doppler information of a target in a time domain and a frequency domain, combining a convolutional neural network (Convolutional Neural Network, CNN) to achieve target distinguishing and the like.
The bird and unmanned aerial vehicle target identification method does not consider the influence of actual environmental noise. The unmanned aerial vehicle automatic identification classification system based on the frequency modulation continuous wave radar (Frequency Modulation Continuous Wave, FMCW) is designed in the prior art, radar echoes of unmanned aerial vehicles, birds, people, vehicles and the like are decomposed and reconstructed by using empirical mode decomposition (Empirical Mode Decomposition, EMD), so that the radar echo noise reduction is realized, characteristics such as signal energy, logarithmic energy entropy, spectral energy density, signal energy ratio and the like are extracted, unmanned aerial vehicle, birds, pedestrians and vehicles are classified through the SVM, and the problem of modal aliasing and end effect existing in the EMD possibly causes the reduction of distinguishing performance. In the prior art, a matching tracking method is utilized to reconstruct the seismic signals, useful information is extracted, the purpose of noise reduction is achieved, and the method is complex in calculation and large in calculation amount. In the prior art, the speed normalization is carried out on echo signals of tracked vehicles and wheeled vehicle targets, then the signals are decomposed by utilizing wavelet transformation, so that the noise reduction treatment of the signals is realized, the characteristics of signal energy, energy ratio, peak amplitude ratio and the like are extracted, the classification of the wheeled vehicles and the tracked vehicles is realized by utilizing SVM, and the wavelet transformation self-adaptability is lower.
In the prior art, in the 7 th period of volume 36, pages 33-43 and 7 month 2022 of the journal of electronic measurement and instrumentation, a method for distinguishing a target of a bird from a rotor unmanned aerial vehicle is disclosed, aiming at the problem of reduced distinguishing performance of the rotor unmanned aerial vehicle with stronger maneuverability in the method for distinguishing the target of the bird from the rotor unmanned aerial vehicle based on the motion characteristics, considering that the frequency spectrum of a bird target in the process of vibrating wing echo is more complex relative to the rotor unmanned aerial vehicle, the two micro-motion characteristics are symmetrically distinguished by the energy entropy and the peak value of the characteristic spectrum corresponding to the frequency spectrum when the target echo is constructed, and the extracted motion characteristics and micro-motion characteristics are fused by using K-means, so that the target of the bird from the rotor unmanned aerial vehicle is distinguished.
Through the above analysis, the problems and defects existing in the prior art are as follows: in the prior art, the influence of low signal-to-noise ratio on the distinguishing performance is not considered, the signal-to-noise ratio of the corresponding radar echo is low due to the influence of noise on the radar echo in the actual observation environment, and the accuracy of distinguishing the target is reduced due to the fact that the radar echo signal is seriously influenced by the noise in the target distinguishing method under the condition of low signal-to-noise ratio.
Disclosure of Invention
In order to overcome the problems in the related art, the disclosed embodiments of the present invention provide a method and a system for identifying a bird and a rotor unmanned aerial vehicle target under a low signal-to-noise ratio condition. The invention aims to accurately distinguish two targets by carrying out noise reduction treatment on radar echoes of birds and rotor unmanned aerial vehicles and extracting corresponding features under the condition of low signal-to-noise ratio and carrying out feature fusion.
The technical scheme is as follows: a method of identifying bird and rotorcraft targets at low signal-to-noise ratios, the method comprising the steps of:
s1, reconstructing radar echoes of a bird and a rotor unmanned aerial vehicle target based on a base tracking noise reduction method, enhancing micro-motion characteristics of the bird and the rotor unmanned aerial vehicle target, and reducing noise to obtain a noise-reduced echo signal;
s2, performing target tracking on the birds and the rotor unmanned aerial vehicle by adopting an interactive multi-model algorithm, and extracting the motion model conversion frequency of the birds and the rotor unmanned aerial vehicle targets as motion characteristics;
s3, obtaining a time-frequency matrix of the bird and the rotor unmanned aerial vehicle by adopting STFT, performing singular decomposition on the time-frequency matrix, and constructing a characteristic spectrum energy entropy as a micro-motion characteristic;
and S4, utilizing the obtained motion model to convert frequency and characteristic spectrum energy entropy, constructing a fused characteristic quantity, inputting the fused characteristic quantity into K-means, and distinguishing the bird and the rotor unmanned aerial vehicle.
In step S1, the base tracking noise reduction method includes:
the radar echo signal y of the bird or unmanned aerial vehicle is:
y＝s+n
wherein y is a radar echo signal of a bird or unmanned aerial vehicle, s is a target signal, and n is noise;
the noise reduction problem of the radar echo signal is described as finding a sparse solution k, enabling a target signal to be sparsely represented in a frequency domain B, using an L1 norm as a measure of sparsity, and constructing an expression of an optimization problem as follows:
Wherein k is opt For the optimal solution of k, B is a frequency domain sparse representation matrix, k is a sparse representation coefficient vector of a target signal in a frequency domain, lambda is a balance vector, lambda.k is multiplication of the vector lambda and the coefficient vector k by elements, I I.I is L1 norm,is the L2 norm.
In step S1, the base tracking noise reduction method is to solve the problem by using a split augmented lagrangian contraction algorithm and construct an optimization problem, so as to obtain an optimal sparse solution and reconstruct an echo signal spectrum.
In step S2, the motion model conversion frequency of the bird and the rotorcraft target is extracted as a motion feature, specifically including:
after interactive multi-model algorithm target tracking is carried out on birds and rotor unmanned aerial vehicles, model probability mu of r models is calculated in continuous T moments n (k) The method comprises the steps of carrying out a first treatment on the surface of the n=1, 2 … r; k=1, 2 … T, the variance average is the motion model conversion frequency estimation value F, and the expression is:
wherein F is the variance average value as the motion model conversion frequency estimation value, r is the number of target motion models, T is the time interval, and var {. Cndot. Is the variance operation.
If one motion state exists in the interactive multi-model algorithm target, the obtained motion model conversion frequency estimated value is smaller than a target threshold value with multiple motion states, the threshold value is set to be Q, wherein Q is more than 0 and less than 1, and when the obtained motion model conversion frequency estimated value is higher than the threshold value Q, the target is judged to be birds; and if the value is lower than the threshold value Q, judging as a rotor unmanned aerial vehicle target.
In step S3, the characteristic spectrum energy entropy is constructed as a jog feature, specifically including:
short-time Fourier transform is carried out on radar echoes of birds and rotor unmanned aerial vehicles, a target signal is s (t), a window function is g (t), and then the short-time Fourier transform expression is:
wherein X is STFT (t, f) is the time spectrum of the radar echo signal, t is time, f is frequency, s (τ) is the target signal, g (τ -t) is the window function, e is the index, τ is the variable time, j is the imaginary unit, dτ is the small variation of τ, e -j2πfτ For exponential operation, infinity;
obtaining a time-frequency matrix X of the bird and the rotor unmanned aerial vehicle through short-time Fourier transform, and carrying out singular value decomposition on the time-frequency matrix X to obtain the following components:
X＝U∑V T
wherein, sigma is a diagonal matrix formed by corresponding singular values after singular value decomposition, p is the number of singular values; (. Cndot. T For transpose operation, < >>For the first singular value, +.>For the second singular value, +.>Is the P th singular value; u is an N x N standard orthogonal matrix, V is an M x M standard orthogonal matrix, U and V are respectively a left singular matrix formed by left singular vectors and a right singular matrix formed by right singular vectors, the left singular vectors represent the frequency information of the echo signals, and the right singular vectors represent the time information of the echo signals;
And the micro motion feature extraction of the unmanned aerial vehicle and birds is realized through the related feature quantity of the singular value decomposition structure.
In one embodiment, the expression for the characteristic spectrum energy entropy using left singular vectors is:
E n ＝|U n，1 | 2 +|U n，2 | 2
wherein eta is the energy entropy of the characteristic spectrum and q n The energy ratio of the left singular vector corresponding to the first two large singular values of the nth sample, E n For the sum of the energy of the first two left singular vectors of the nth sample, |U n，1 | 2 ，|U n，2 | 2 The energy of the first and second left singular vectors representing the nth sample, N being the number of samples, lnq, respectively n Is q n Is a natural logarithm of (a) to (b),is the mth singular value.
In step S4, the obtained motion model is utilized to convert frequency and energy entropy of a characteristic spectrum, and the fused characteristic quantity is constructed and input into K-means, so as to realize the identification of birds and rotor unmanned aerial vehicles, and the method specifically comprises the following steps:
the extracted motion model converts the frequency estimation value into a 1 The energy entropy of the characteristic spectrum is a 2 The method comprises the steps of carrying out a first treatment on the surface of the The time interval of the target movement process of the bird or rotor unmanned aerial vehicle is T, the time interval of the target movement is divided into m time periods, and each time period is T i The method comprises the steps of carrying out a first treatment on the surface of the Calculating motion model conversion frequency estimation values of a bird or rotor unmanned aerial vehicle target, and simultaneously, respectively carrying out time period T i Performing inching feature analysis to extract feature spectrum energy entropy a 2 ；
For the obtained i characteristic spectrum energy entropy a 2i Averaging, i=1, 2 … m, characteristic spectrum energy entropy a 2 The calculation formula is as follows:
wherein i represents the number of time periods;
converting the frequency estimate a using the extracted motion model 1 Characteristic spectrum energy entropy a 2 The fused characteristic quantity A is constructed, and the calculation formula is as follows:
A＝(a 1 a 2 )
and inputting the characteristic quantity A into the K-means to realize the identification of the bird and the rotor unmanned aerial vehicle.
Another object of the present invention is to provide a system for identifying objects of a bird and a rotorcraft in the presence of a low signal-to-noise ratio, implementing the method for identifying objects of a bird and a rotorcraft in the presence of a low signal-to-noise ratio, the system comprising:
the noise-reduced echo signal acquisition module is used for reconstructing radar echoes of the bird and rotor unmanned aerial vehicle targets based on a base tracking noise reduction method, enhancing micro-motion characteristics of the bird and rotor unmanned aerial vehicle targets and reducing noise to obtain noise-reduced echo signals;
the motion characteristic acquisition module is used for tracking targets of the birds and the rotor unmanned aerial vehicle by adopting an interactive multi-model algorithm, and extracting the motion model conversion frequency of the targets of the birds and the rotor unmanned aerial vehicle as motion characteristics;
the micro-motion feature acquisition module is used for obtaining a time-frequency matrix of the bird and the rotor unmanned aerial vehicle by adopting STFT, performing singular decomposition on the time-frequency matrix, and constructing a feature spectrum energy entropy as a micro-motion feature;
And the identification module of the bird and the rotor unmanned aerial vehicle utilizes the obtained motion model conversion frequency and the characteristic spectrum energy entropy to construct the fused characteristic quantity and input the fused characteristic quantity into the K-means so as to realize identification of the bird and the rotor unmanned aerial vehicle.
The recognition system of the bird and the rotor unmanned aerial vehicle is carried on the radar recognition network platform under the condition of low signal-to-noise ratio, and the unmanned aerial vehicle and the bird are classified and recognized.
By combining all the technical schemes, the invention has the advantages and positive effects that: the invention innovatively provides a recognition system for the bird and the unmanned aerial vehicle target, which aims at the recognition of the bird and the unmanned aerial vehicle target under the condition of low signal-to-noise ratio, carries out radar echo reconstruction on the received radar echo by using a BPD method, enhances the micro-motion characteristics of the target, realizes noise reduction, and increases the recognition system for the bird and the unmanned aerial vehicle target on the basis;
the technical scheme of the invention can improve the performances of bird expelling and unmanned aerial vehicle countering equipment and non-cooperative target monitoring equipment in an airport, and has commercial value and practical engineering application value for guaranteeing flight safety. The influence of low signal to noise ratio is not considered in the prior art for distinguishing the bird from the unmanned aerial vehicle target, and the bird and the unmanned aerial vehicle target can be distinguished under the condition of low signal to noise ratio by the technical scheme. The invention can combine the motion characteristics of the data level with the micro-motion characteristics of the signal level, combines the influence of low signal-to-noise ratio on the non-cooperative target distinguishing performance, and provides a new idea for non-cooperative target distinguishing.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the disclosure and together with the description, serve to explain the principles of the disclosure;
FIG. 1 is a flow chart of a method for identifying bird and rotorcraft targets at low signal-to-noise ratios provided by an embodiment of the present invention;
FIG. 2 is a schematic diagram of a method for identifying bird and rotorcraft targets at low signal-to-noise ratios provided by an embodiment of the present invention;
fig. 3 is a block diagram of a BPD noise reduction implementation provided by an embodiment of the present invention;
FIG. 4 is a time-frequency plot of bird targets in a time-frequency plot of bird and rotor unmanned aerial vehicle simulation data obtained prior to noise reduction provided by an embodiment of the present invention;
fig. 5 is a time-frequency diagram of the rotary wing unmanned aerial vehicle in a simulation data time-frequency diagram of the bird and the rotary wing unmanned aerial vehicle obtained before noise reduction provided by the embodiment of the invention;
fig. 6 is a time-frequency diagram of an avian target in a time-frequency diagram of simulation data of a bird and rotor unmanned aerial vehicle obtained after noise reduction of a BPD provided by an embodiment of the present invention;
fig. 7 is a time-frequency diagram of the rotary wing unmanned aerial vehicle in a simulation data time-frequency diagram of the bird and the rotary wing unmanned aerial vehicle obtained after noise reduction of the BPD provided by the embodiment of the invention;
fig. 8 is a time-frequency diagram of bird targets in a time-frequency diagram of birds and a rotor unmanned aerial vehicle with measured data before noise reduction provided by an embodiment of the present invention;
Fig. 9 is a time-frequency diagram of the rotary wing unmanned aerial vehicle in the time-frequency diagrams of the bird and the rotary wing unmanned aerial vehicle with measured data before noise reduction provided by the embodiment of the invention;
fig. 10 is a time-frequency diagram of bird targets in a time-frequency diagram of a bird and a rotor unmanned aerial vehicle based on measured data after noise reduction of a BPD provided by an embodiment of the present invention;
fig. 11 is a time-frequency diagram of a rotary wing unmanned aerial vehicle in time-frequency diagrams of a bird and a rotary wing unmanned aerial vehicle with measured data after noise reduction of a BPD provided by the embodiment of the invention;
fig. 12 is a diagram of a recognition result of motion characteristics in a recognition result of a bird and a rotor unmanned aerial vehicle before noise reduction of simulation data provided by an embodiment of the present invention;
FIG. 13 is a graph of micro-motion feature discrimination results among bird and rotorcraft discrimination results before noise reduction of simulation data provided by an embodiment of the present invention;
fig. 14 is a diagram of a feature fusion recognition result in a recognition result of a bird and a rotor unmanned aerial vehicle before noise reduction of simulation data provided by an embodiment of the present invention;
fig. 15 is a diagram of a recognition result of motion characteristics in recognition results of a bird and a rotor unmanned aerial vehicle after noise reduction of simulation data BPD provided by an embodiment of the present invention;
FIG. 16 is a graph of simulated data BPD denoised micro-motion features in bird and rotor unmanned aerial vehicle discrimination results;
fig. 17 is a diagram of a feature fusion recognition result in a recognition result of a bird and a rotor unmanned aerial vehicle after noise reduction of simulation data BPD provided by an embodiment of the present invention;
Fig. 18 is a diagram of a recognition result of motion characteristics in recognition results of birds and a rotorcraft before actual measurement data noise reduction provided by an embodiment of the present invention;
fig. 19 is a diagram of a micro-motion feature recognition result in a recognition result of a bird and a rotor unmanned aerial vehicle before actual measurement data noise reduction provided by an embodiment of the present invention;
fig. 20 is a diagram of a feature fusion recognition result among recognition results of birds and a rotor unmanned aerial vehicle before actual measurement data noise reduction provided by an embodiment of the present invention;
fig. 21 is a diagram of a recognition result of motion characteristics in recognition results of a bird and a rotor unmanned aerial vehicle after BPD noise reduction according to the embodiment of the present invention;
fig. 22 is a diagram of a micro-motion feature recognition result in a recognition result of a bird and a rotor unmanned aerial vehicle after BPD noise reduction according to the embodiment of the present invention;
fig. 23 is a diagram of a feature fusion recognition result in a recognition result of a bird and a rotor unmanned aerial vehicle after noise reduction of measured data BPD provided by an embodiment of the present invention;
FIG. 24 is a graph showing the performance of the method for identifying birds and rotorcraft after motion feature fusion processing alone according to the variation of accuracy with signal-to-noise ratio before and after noise reduction according to an embodiment of the present invention;
FIG. 25 is a graph showing the performance of the method for identifying birds and rotorcraft after independent micro-motion feature fusion processing according to the variation of accuracy with signal-to-noise ratio before and after noise reduction;
Fig. 26 is a graph showing the performance of the method for identifying a bird and a rotor unmanned aerial vehicle after the fusion processing of motion and micro motion features according to the embodiment of the present invention, wherein the accuracy of identification varies with the signal to noise ratio before and after noise reduction.
Detailed Description
In order that the above objects, features and advantages of the invention will be readily understood, a more particular description of the invention will be rendered by reference to the appended drawings. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention. The invention may be embodied in many other forms than described herein and similarly modified by those skilled in the art without departing from the spirit or scope of the invention, which is therefore not limited to the specific embodiments disclosed below.
The embodiment of the invention provides a method for distinguishing targets of a bird and a rotor unmanned aerial vehicle under the condition of low signal-to-noise ratio, which comprises the steps of firstly reconstructing radar echo signals of the bird and the rotor unmanned aerial vehicle by using base tracking noise reduction (Basis Pursuit Denoising, BPD) so as to achieve the purpose of reducing noise; performing target tracking on the bird and the rotor unmanned aerial vehicle through an IMM algorithm, and extracting the conversion frequency of the motion model as a motion characteristic; performing short-time Fourier transform (STFT) on radar echoes of the birds and the rotor unmanned aerial vehicle to obtain a time-frequency matrix, performing feature extraction on the time-frequency matrix through SVD, and constructing a feature spectrum energy entropy as a micro-motion feature; and (3) carrying out fusion processing on the extracted motion model conversion frequency and characteristic spectrum energy entropy, and inputting the fusion processing to a K-means (K-means) method to realize the identification of the bird and the rotor unmanned aerial vehicle. The method and the device have the advantages of simple processing steps, high calculation speed and the like under the condition of low signal-to-noise ratio, and utilize the BPD to carry out noise reduction processing on the target radar echo signals so as to realize target identification.
In embodiment 1, as shown in fig. 1, the method for identifying the targets of the bird and the rotor unmanned aerial vehicle under the condition of low signal-to-noise ratio provided by the embodiment of the invention comprises the following steps:
s1, reconstructing radar echoes of targets of birds and rotor unmanned aerial vehicles based on a base tracking noise reduction method (Basis Pursuit Denoising, BPD), enhancing corresponding micro-motion characteristics of the radar echoes, and obtaining noise-reduced echo signals;
it can be understood that the radar echo of the bird and the rotor unmanned aerial vehicle target is reconstructed based on the tracking noise reduction method, the micro-motion characteristics of the bird and the rotor unmanned aerial vehicle target are enhanced, noise is reduced, and a noise reduced echo signal is obtained;
s2, tracking two targets by using the reconstructed echo signals obtained in the step S1 and adopting an Interactive Mu|time Model (IMM) algorithm, and extracting the motion Model conversion frequency of the bird and rotor unmanned aerial vehicle targets as motion characteristics;
s3, utilizing the reconstructed echo signals obtained in the step S1, adopting short-time Fourier transform (Short Time Fourier Transform, STFT) to obtain time-frequency matrixes of the bird and the rotor unmanned aerial vehicle, carrying out singular decomposition on the time-frequency matrixes, and constructing a characteristic spectrum energy entropy as a micro-motion characteristic;
and S4, constructing a fused characteristic quantity by utilizing the motion model conversion frequency and the characteristic spectrum energy entropy obtained in the step S2 and the step S3, and inputting the fused characteristic quantity into the K-means to realize the identification of the bird and the rotor unmanned aerial vehicle.
Further, in step S1, the method uses BPD to reconstruct radar echo signals of the bird and the rotor unmanned plane to obtain echo signals after noise reduction, and the specific formula is as follows:
suppose that the bird or drone radar echo signal y is:
y＝s+n
wherein y is a radar echo signal of a bird or unmanned aerial vehicle, s is a target signal, and n is noise;
the noise reduction problem of the radar echo signal is described as finding a sparse solution k, enabling a target signal to be sparsely represented in a frequency domain B, using an L1 norm as a measure of sparsity, and constructing an expression of an optimization problem as follows:
wherein k is opt For the optimal solution of k, B is a frequency domain sparse representation matrix, k is a sparse representation coefficient vector of a target signal in a frequency domain, lambda is a balance vector, lambda.k is multiplication of the vector lambda and the coefficient vector k by elements, I I.I is L1 norm,is the L2 norm. The basic idea of the BPD method is to construct a cost function based on signal sparseness characteristics, using a split augmented lagrangian contraction algorithm (SplitAugmented Lagrangian Shrinkage A|gorithm, SALSA) solves the optimization problem described above.
In step S2, the reconstructed echo signals obtained in step S1 are used to track two targets by using an interactive multi-model algorithm IMM, and the motion model conversion frequency of the bird and rotor unmanned aerial vehicle targets is extracted as a calculation formula of motion characteristics:
After interactive multi-model algorithm IMM target tracking is carried out on birds and rotor unmanned aerial vehicles, model probability mu of r models is calculated in continuous T moments n (k) The method comprises the steps of carrying out a first treatment on the surface of the n=1, 2 … r; k=1, 2 … T, the variance average is the motion model conversion frequency estimation value F, and the expression is:
wherein F is the variance average value as the motion model conversion frequency estimation value, r is the number of target motion models, T is the time interval, and var {. Cndot. Is the variance operation.
Further, using the reconstructed echo signal obtained in the step S1, obtaining a time-frequency matrix of the bird and the rotor unmanned aerial vehicle by using STFT, and performing SVD construction on the time-frequency matrix to obtain a calculation formula of taking a characteristic spectrum energy entropy as a inching characteristic, wherein the calculation formula is as follows:
short-time Fourier transform (STFT) is performed on radar echoes of birds and rotor unmanned aerial vehicles, a target signal is s (t), a window function is g (t), and the short-time Fourier transform expression is:
wherein X is STFT (t, f) is the time spectrum of the radar echo signal, t is time, f is frequency, s (τ) is the target signal, g (τ -t) is the window function, e is the index, τ is the variable time, j is the imaginary unit, dτ is the small variation of τ, e -j2πfτ For exponential operation, infinity;
Obtaining a time-frequency matrix X of the bird and the rotor unmanned aerial vehicle through short-time Fourier transform, and carrying out singular value decomposition on the time-frequency matrix X to obtain the following components:
X＝U∑V T
wherein, sigma is a diagonal matrix formed by corresponding singular values after singular value decomposition, p is the number of singular values; (. Cndot. T For transpose operation, < >>For the first singular value, +.>For the second singular value, +.>Is the P th singular value; u is an N x N standard orthogonal matrix, V is an M x M standard orthogonal matrix, U and V are respectively a left singular matrix formed by left singular vectors and a right singular matrix formed by right singular vectors, the left singular vectors represent the frequency information of the echo signals, and the right singular vectors represent the time information of the echo signals; and the micro motion feature extraction of the unmanned aerial vehicle and birds is realized through the related feature quantity of the singular value decomposition structure.
The expression for calculating the energy entropy of the characteristic spectrum by using the left singular vector (frequency information) is as follows:
E n ＝|U n，1 | 2 +|U n，2 | 2
wherein eta isCharacteristic spectrum energy entropy, q n The energy ratio of the left singular vector corresponding to the first two large singular values of the nth sample, E n For the sum of the energy of the first two left singular vectors of the nth sample, |U n，1 | 2 ，|U n，2 | 2 The energy of the first and second left singular vectors representing the nth sample, N being the number of samples, lnq, respectively n Is q n Is a natural logarithm of (a) to (b),is the mth singular value.
Further, the process of converting the frequency and the characteristic spectrum energy entropy by utilizing the motion model obtained in the step S2 and the step S3, constructing the fused characteristic quantity and inputting the fused characteristic quantity into the K-means, and realizing the identification of the bird and the rotor unmanned aerial vehicle is as follows:
let the extracted motion model conversion frequency estimation value be a 1 The energy entropy of the characteristic spectrum is a 2 The method comprises the steps of carrying out a first treatment on the surface of the The time interval of the target movement process of the bird or rotor unmanned aerial vehicle is T, the time interval of the target movement is divided into m time periods, and each time period is T i The method comprises the steps of carrying out a first treatment on the surface of the Calculating motion model conversion frequency estimation values of a bird or rotor unmanned aerial vehicle target, and simultaneously, respectively carrying out time period T i Performing inching feature analysis to extract feature spectrum energy entropy a 2 The method comprises the steps of carrying out a first treatment on the surface of the For the obtained i characteristic spectrum energy entropy a 2i Averaging, i=1, 2 … m, characteristic spectrum energy entropy a 2 The calculation formula is as follows:
wherein i represents the number of time periods;
converting the frequency estimate a using the extracted motion model 1 Characteristic spectrum energy entropy a 2 The fused characteristic quantity A is constructed, and the calculation formula is as follows:
A＝(a 1 a 2 )
and inputting the characteristic quantity A into the K-means to realize the identification of the bird and the rotor unmanned aerial vehicle.
Embodiment 2, as another implementation manner of the present invention, as shown in fig. 2, the method for identifying a bird and a rotor unmanned aerial vehicle under the condition of low signal-to-noise ratio provided by the embodiment of the present invention includes the following steps:
Firstly, reconstructing radar echo signals of birds and rotor unmanned aerial vehicles by using a BPD (Business process kit) to obtain echo signals after noise reduction;
the basic idea of the base tracking noise reduction (BPD) method is to construct a cost function based on signal sparseness characteristics, solve an optimization problem by utilizing a split augmented Lagrangian contraction algorithm (SALSA), and further obtain an optimal sparse solution reconstructed echo signal spectrum to achieve the purpose of noise reduction.
Suppose that the bird or drone radar echo signal y is:
y＝s+n(1)
wherein y is a radar echo signal of a bird or unmanned aerial vehicle, s is a target signal, and n is noise;
the noise reduction problem of the radar echo signal is described as finding a sparse solution k, enabling a target signal to be sparsely represented in a frequency domain B, using an L1 norm as a measure of sparsity, and constructing an expression of an optimization problem as follows:
wherein k is opt For the optimal solution of k, B is a frequency domain sparse representation matrix, k is a sparse representation coefficient vector of a target signal in a frequency domain, lambda is a balance vector, lambda.k is multiplication of the vector lambda and the coefficient vector k by elements, I I.I is L1 norm,is the L2 norm.
And solving the optimization problem by adopting an SALSA algorithm. The optimization problem of equation (2) can be described as:
in the formula, s.t: representation constrained And (3) in the process;representing the value of the corresponding independent variable when the cost function takes the minimum value;
then, the optimization of the formula (3) can be solved by using an augmented Lagrangian algorithm, a vector g is introduced, the vector g needs to be initialized before optimization iteration, the vector g is usually initialized to be a zero vector, and the step length is zeta, so that the algorithm 1 is obtained.
Illustratively, algorithm 1 comprises:
initializing:
ζ＞0，g
optimization iteration:
g←g-(u-k) (6)
and (5) ending.
Where lambda.u represents the multiplication of the vector lambda with u by the same position of the element,as an operation, the value of the corresponding independent variable u when the cost function is the minimum value is represented, and u-k-g represents vector difference operation;
equation (4) represents updating the variable u with the corresponding u value after the cost function is minimized,
equation (5) represents updating the variable k with the corresponding k value after the cost function is minimized;
the formula (6) represents reassigning the value updated to the variable g, and u-k represents a difference operation;
in the embodiment of the invention, the soft threshold rule is adopted to perform the minimization treatment on the formula (4), the minimization problem of the formula (5) is a constrained least square problem, and the constraint least square problem can be treated by using the Robiida rule and matrix operation. Thus, algorithm 1 may be further simplified, resulting in algorithm 2.
Illustratively, algorithm 2 includes:
Initializing:
ζ＞0，g
optimization iteration:
k←(B H B+ζI) -1 (B H y+ζ(u-g)) (8)
g←g-(u-k) (9)
and (5) ending.
Wherein, the formulas (7), (8) and (9) respectively represent iterative formulas for solving the sparse representation coefficient vector,representing division, soft (·) representing soft thresholding, B H Representing the transpose of the matrix B, I representing the identity matrix;
b in formula (8) H Is the conjugate transpose of B, transform B generally satisfies the pasawal form, representing the characteristic formula of the frequency domain representation matrix B:
BB H ＝pI (10)
where p is a Pascal constant, generally set to 1, and I is an identity matrix.
In order to further reduce the amount of computation, let the variable v=u-g, the formula (8) is simplified by matrix operation, and the inversion operation on the left side of the equal sign is simplified to obtain:
in order to avoid inversion operation and reduce the operation amount of the algorithm 2, the algorithm 3 can be obtained by simplifying the algorithm 2 according to the formula (10) and the formula (11). A corresponding implementation block diagram is shown in fig. 3.
Illustratively, algorithm 3 includes:
initializing:
ζ＞0，g
and (3) solving optimization iteration of the optimal sparse representation coefficient:
k←g+v (14)
and (5) ending.
In the method, in the process of the invention,the two variables are represented and added to obtain the reciprocal, and Bv represents the multiplication of the matrix B and the vector v;
tracking two targets by using the reconstructed echo signals obtained in the first step and adopting an IMM algorithm, and extracting the motion model conversion frequency of the targets of the bird and the rotor unmanned aerial vehicle as motion characteristics;
Tracking the target through IMM, extracting model probability, and calculating model probability mu of r models n (k) The method comprises the steps of carrying out a first treatment on the surface of the n=1, 2 … r; the variance average value of k=1, 2 … T is the motion model conversion frequency estimated value F, and the expression is:
/>
wherein F is the variance average value as the motion model conversion frequency estimation value, r is the number of target motion models, T is the time interval, and var {. Cndot. Is the variance operation.
If the motion state of the target is single, only one motion state exists, and if one motion state exists, the obtained motion model conversion frequency estimated value is smaller than the target threshold value with multiple motion states. In general, since the maneuverability of the unmanned rotorcraft is weaker than that of birds, a threshold Q may be set, when the obtained motion model conversion frequency estimation value is higher than the threshold Q, the target may be determined to be birds, otherwise, if the motion model conversion frequency estimation value is lower than the threshold Q, the target is determined to be the unmanned rotorcraft, and if the motion model conversion frequency estimation value is 0 < Q < 1, the target may be determined to be the unmanned rotorcraft, so that the identification of the two can be achieved by extracting the target motion model conversion frequency estimation value.
Thirdly, utilizing the reconstructed echo signals obtained in the first step, obtaining a time-frequency matrix of the bird and the rotor unmanned aerial vehicle by adopting STFT, and carrying out SVD construction characteristic spectrum energy entropy on the time-frequency matrix as a micro-motion characteristic;
Short-time Fourier transform (STFT) is performed on radar echoes of birds and rotor unmanned aerial vehicles, and a target signal is s (t), a window function is g (t), and the STFT is defined as:
wherein X is STFT (t, f) is the time spectrum of the radar echo signal, t is time, f is frequency, s (τ) is the target signal, g (τ -t) is the window function, e is the index, which is a mathematical operation; τ is the variable time, j is the imaginary unit, dτ is the small variation of τ, e -j2πfτ For exponential operation, infinity;
fig. 4 is a time-frequency diagram of a bird target in a time-frequency diagram of simulation data of a bird and a rotor unmanned aerial vehicle obtained before noise reduction, and fig. 5 is a time-frequency diagram of a rotor unmanned aerial vehicle in a time-frequency diagram of simulation data of a bird and a rotor unmanned aerial vehicle obtained before noise reduction. Fig. 6 is a time-frequency diagram of a bird target in a time-frequency diagram of simulation data of a bird and a rotor unmanned aerial vehicle obtained after noise reduction of a BPD, and fig. 7 is a time-frequency diagram of a rotor unmanned aerial vehicle in a time-frequency diagram of simulation data of a bird and a rotor unmanned aerial vehicle obtained after noise reduction of a BPD. As can be seen from fig. 4-7, the jogging features of the bird and rotorcraft targets are almost buried by noise. After the BPD noise reduction treatment, the time-frequency diagram of the rotor unmanned aerial vehicle has obvious periodicity, the time-frequency spectrum periodicity of the bird target is weaker, and the characteristics can be used for distinguishing the rotor unmanned aerial vehicle from the bird target.
Fig. 8 is a time-frequency diagram of an avian target in a time-frequency diagram of a bird and a rotary-wing unmanned aerial vehicle with measured data before noise reduction, and fig. 9 is a time-frequency diagram of a rotary-wing unmanned aerial vehicle in a time-frequency diagram of a bird and a rotary-wing unmanned aerial vehicle with measured data before noise reduction. Fig. 10 is a time-frequency diagram of an avian target in a time-frequency diagram of a bird and a rotary-wing unmanned aerial vehicle with measured data after noise reduction of a BPD, and fig. 11 is a time-frequency diagram of a rotary-wing unmanned aerial vehicle in a time-frequency diagram of a bird and a rotary-wing unmanned aerial vehicle with measured data after noise reduction of a BPD. As can be seen from fig. 8-11, before noise reduction, the micro-motion characteristics of the two targets are submerged in noise as the actual measured birds and the rotorcraft are affected by various environmental noise. After the BPD noise reduction process, the noise is suppressed.
Obtaining a time-frequency matrix X of the bird and the rotor unmanned aerial vehicle through STFT, and carrying out singular value decomposition on the time-frequency matrix X:
X＝U∑V T (17)
wherein, sigma is a diagonal matrix formed by corresponding singular values after singular value decomposition, p is the number of singular values; (. Cndot. T For transpose operation, < >>For the first singular value, +.>For the second singular value, +.>Is the P th singular value; u is an N x N standard orthogonal matrix, V is an M x M standard orthogonal matrix, U and V are respectively a left singular matrix formed by left singular vectors and a right singular matrix formed by right singular vectors, the left singular vectors represent the frequency information of the echo signals, and the right singular vectors represent the time information of the echo signals; thus, the characteristics of the spectral components of the target echo signal can be reflected by the left singular vectors.
And the micro motion feature extraction of the unmanned aerial vehicle and birds is realized through the related feature quantity of the singular value decomposition structure. Solving characteristic spectrum energy entropy aiming at left singular vectors (frequency information):
E n ＝|U n，1 | 2 +|U n，2 | 2
wherein eta is the energy entropy of the characteristic spectrum and q n The energy ratio of the left singular vector corresponding to the first two large singular values of the nth sample, E n For the sum of the energy of the first two left singular vectors of the nth sample, |U n，1 | 2 ，|U n，2 | 2 The energy of the first and second left singular vectors representing the nth sample, N being the number of samples, lnq, respectively n Is q n Is a natural logarithm of (a) to (b),is the mth singular value.
The larger the energy entropy value is, the stronger the disorder of the time-frequency diagram is represented, the time-frequency diagram of the rotor unmanned aerial vehicle is higher in order than birds, according to the microscopic physical meaning of entropy, if the obtained characteristic spectrum energy entropy value is larger, the target can be considered to be a bird target, otherwise, the obtained characteristic spectrum energy entropy value is smaller, the target can be considered to be a rotor unmanned aerial vehicle target, and therefore the birds and the rotor unmanned aerial vehicle can be distinguished through the size of the entropy value.
And step four, utilizing the motion model conversion frequency and the characteristic spectrum energy entropy obtained in the step two and the step three to construct the fused characteristic quantity and inputting the characteristic quantity into K-measns to realize the identification of the bird and the rotor unmanned aerial vehicle.
Let the extracted motion model conversion frequency estimation value be a 1 The energy entropy of the characteristic spectrum is a 2 The method comprises the steps of carrying out a first treatment on the surface of the The time interval of the target movement process of the bird or rotor unmanned plane is T, and the target is movedThe time interval is divided into m time periods, each time period is T i The method comprises the steps of carrying out a first treatment on the surface of the Calculating motion model conversion frequency estimation values of a bird or rotor unmanned aerial vehicle target, and simultaneously, respectively carrying out time period T i Performing inching feature analysis to extract feature spectrum energy entropy a 2 The method comprises the steps of carrying out a first treatment on the surface of the For the obtained i characteristic spectrum energy entropy a 2i (i=1, 2 … m), i.e.:
wherein i represents the number of time slots, a 2 Representing the energy entropy of a characteristic spectrum, m represents the number of time periods divided by the movement time of a target, and i represents the ith time period;
converting the frequency estimate a using the extracted motion model 1 Characteristic spectrum energy entropy a 2 Constructing a fused characteristic quantity A, namely:
A＝(a 1 a 2 ) (20)
and inputting the characteristic quantity A into the K-means to realize the identification of the bird and the rotor unmanned aerial vehicle.
Embodiment 3 the system for identifying objects of a bird and a rotorcraft under a low signal-to-noise condition provided by the embodiment of the present invention includes:
the noise-reduced echo signal acquisition module is used for reconstructing radar echoes of the targets of the bird and the rotor unmanned aerial vehicle based on a base tracking noise reduction method (BPD), enhancing the corresponding micro-motion characteristics of the radar echoes, and obtaining noise-reduced echo signals;
The motion characteristic acquisition module is used for tracking two targets by using the acquired echo signals after reconstruction and adopting an IMM algorithm to extract the motion model conversion frequency of the bird and rotor unmanned aerial vehicle targets as motion characteristics;
the micro-motion feature acquisition module is used for obtaining a time-frequency matrix of the bird and the rotor unmanned aerial vehicle by using the obtained reconstructed echo signals and STFT, performing singular decomposition on the time-frequency matrix, and constructing a feature spectrum energy entropy as a micro-motion feature;
and the identification module of the bird and the rotor unmanned aerial vehicle is used for utilizing the obtained motion model to convert frequency and characteristic spectrum energy entropy, constructing the fused characteristic quantity and inputting the fused characteristic quantity into the K-means to realize identification of the bird and the rotor unmanned aerial vehicle.
In the foregoing embodiments, the descriptions of the embodiments are emphasized, and in part, not described or illustrated in any particular embodiment, reference is made to the related descriptions of other embodiments.
The content of the information interaction and the execution process between the devices/units and the like is based on the same conception as the method embodiment of the present invention, and specific functions and technical effects brought by the content can be referred to in the method embodiment section, and will not be described herein.
It will be apparent to those skilled in the art that, for convenience and brevity of description, only the above-described division of the functional units and modules is illustrated, and in practical application, the above-described functional distribution may be performed by different functional units and modules according to needs, i.e. the internal structure of the apparatus is divided into different functional units or modules to perform all or part of the above-described functions. The functional units and modules in the embodiment may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit, where the integrated units may be implemented in a form of hardware or a form of a software functional unit. In addition, the specific names of the functional units and modules are only for distinguishing from each other, and are not used for limiting the protection scope of the present invention. For specific working processes of the units and modules in the system, reference may be made to corresponding processes in the foregoing method embodiments.
Based on the technical solutions described in the embodiments of the present application, the following application examples may be further proposed.
According to an embodiment of the present application, there is also provided a computer apparatus including: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, which when executed by the processor performs the steps of any of the various method embodiments described above.
Embodiments of the present application also provide a computer readable storage medium storing a computer program which, when executed by a processor, performs the steps of the respective method embodiments described above.
The embodiment of the application also provides an information data processing terminal, which is used for providing a user input interface to implement the steps in the method embodiments when being implemented on an electronic device, and the information data processing terminal is not limited to a mobile phone, a computer and a switch.
The embodiment of the application also provides a server, which is used for realizing the steps in the method embodiments when being executed on the electronic device and providing a user input interface.
Embodiments of the present application also provide a computer program product which, when run on an electronic device, causes the electronic device to perform the steps of the method embodiments described above.
The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the present application may implement all or part of the flow of the method of the above embodiments, and may be implemented by a computer program to instruct related hardware, where the computer program may be stored in a computer readable storage medium, and when the computer program is executed by a processor, the computer program may implement the steps of each of the method embodiments described above. Wherein the computer program comprises computer program code which may be in source code form, object code form, executable file or some intermediate form etc. The computer readable medium may include at least: any entity or device capable of carrying computer program code to a photographing device/terminal apparatus, recording medium, computer Memory, read-Only Memory (ROM), random access Memory (Random Access Memory, RAM), electrical carrier signals, telecommunications signals, and software distribution media. Such as a U-disk, removable hard disk, magnetic or optical disk, etc.
To further demonstrate the positive effects of the above embodiments, the present invention was based on the above technical solutions to perform the following experiments.
Fig. 12 is a motion feature recognition result in a simulation data noise reduction front bird and rotor unmanned aerial vehicle recognition result, fig. 13 is a jog feature recognition result in a simulation data noise reduction front bird and rotor unmanned aerial vehicle recognition result, and fig. 14 is a feature fusion recognition result in a simulation data noise reduction front bird and rotor unmanned aerial vehicle recognition result.
Fig. 15 is a motion feature recognition result in a recognition result of the bird and the rotor unmanned aerial vehicle after the simulation data BPD is noise reduced, fig. 16 is a micro motion feature recognition result in a recognition result of the bird and the rotor unmanned aerial vehicle after the simulation data BPD is noise reduced, and fig. 17 is a feature fusion recognition result in a recognition result of the bird and the rotor unmanned aerial vehicle after the simulation data BPD is noise reduced. The motion states of the rotary wing unmanned aerial vehicle and the bird target in the simulation data are uniform acceleration motion combined with uniform motion, the motion model conversion frequency is extracted aiming at the simulation data, fusion processing is carried out on two features of the feature spectrum energy entropy value, and the identification of the bird and the rotary wing unmanned aerial vehicle, which are fused with each other, is realized through K-means.
As can be seen from fig. 15 to fig. 17, when the rotor unmanned aerial vehicle is in a variable speed motion state, the conversion frequency of the motion model of the rotor unmanned aerial vehicle is between 0.04 and 0.06 close to that of the bird target, and the rotor unmanned aerial vehicle cannot distinguish the bird target from the bird target. For jog features, the presence of individual sample rotor drones and birds in this case have similar spectral energy entropy, which may lead to discrimination inaccuracies. After the noise reduction process, the difference between the motion characteristics is not large, since the extracted motion model conversion frequency is related to the motion state of the target. The identification effect of the micro-motion features is obviously improved, and the identification performance of the rotor unmanned aerial vehicle and the bird targets is improved after the fusion treatment of the motion features and the micro-motion features.
As can be seen from fig. 18 to fig. 23, in the actual measurement data, the motion state of the bird target is variable deceleration, variable acceleration, and uniform motion, and the motion state of the rotary-wing unmanned aerial vehicle is uniform acceleration, uniform velocity, and uniform deceleration motion. And carrying out feature fusion on the two features of the motion model conversion frequency extracted from the measured data and the feature spectrum energy entropy value, and distinguishing the bird and the rotor unmanned aerial vehicle with the motion features and the inching features fused through K-means.
As can be seen from fig. 21-23, the flight state of the rotor unmanned aerial vehicle is maneuvering, the obtained transformation frequency of the movement model is similar to that of the bird target and is between 0.02 and 0.04, and the identification of the two targets cannot be realized by virtue of the movement characteristics. For inching characteristics, samples exist, the characteristic spectrum energy entropy of the rotor unmanned aerial vehicle is similar to that of the bird, and individual conditions can be inaccurately judged. After noise reduction, the rotor unmanned aerial vehicle and the birds are in a maneuvering state, the influence on the motion characteristics is small after noise reduction, the micro-motion characteristics are obviously improved greatly, but the characteristic spectrum energy entropy of the rotor unmanned aerial vehicle and the characteristic spectrum energy entropy of the birds are similar in a certain sample, and misjudgment is possible. The distinguishing result after the feature fusion can be seen that the distinguishing effect of the rotor unmanned aerial vehicle and the bird target is better after the BPD is used for reducing noise.
As can be seen from fig. 24 to fig. 26, after noise reduction of the BPD, the discrimination accuracy using the motion feature, using the jog feature, and fusing the motion and jog features is improved relative to the case of no noise reduction, and the discrimination accuracy of the three methods shows an ascending state with the improvement of the signal-to-noise ratio. The unmanned aerial vehicle is in the non-uniform condition, because unmanned aerial vehicle's motion model conversion frequency is close with the bird, utilizes the bird of motion characteristic and rotor unmanned aerial vehicle to distinguish the rate of accuracy when reaching about 60% this moment the biggest. After feature fusion, the identification accuracy of the bird and the rotor unmanned aerial vehicle is higher than that of the unmanned aerial vehicle which considers the motion and the inching features independently. The problem of identification of birds and rotor unmanned aerial vehicle in the actual measurement environment is solved.
In this example, simulation data radar parameters, bird parameters, and drone parameters are shown in tables 1, 2, and 3.
TABLE 1
| Parameters (parameters) | Numerical value |
| Operating frequency | 5.5GHz |
| PRI | 60μs |
| Fast time sampling rate | 80MHz |
| Bandwidth of a communication device | 20MHz |
| Observation time | 192ms |
TABLE 2
| Parameters (parameters) | Numerical value |
| Rotor number | 4 |
| Number of blades | 2 |
| Blade length | 120mm |
| Rotational speed | 50r/s，64r/s |
TABLE 3 Table 3
| Parameters (parameters) | Numerical value |
| Length of upper arm | 0.5m |
| Forearm length | 0.5m |
| Length of bird body | 1m |
| Beating frequency | 4Hz |
While the invention has been described with respect to what is presently considered to be the most practical and preferred embodiments, it is to be understood that the invention is not limited to the disclosed embodiments, but on the contrary, is intended to cover various modifications, equivalents, and alternatives falling within the spirit and scope of the invention.
Claims (10)
1. A method for identifying bird and rotorcraft targets at low signal-to-noise ratios, the method comprising the steps of:
s1, reconstructing radar echoes of a bird and a rotor unmanned aerial vehicle target based on a base tracking noise reduction method, enhancing micro-motion characteristics of the bird and the rotor unmanned aerial vehicle target, and reducing noise to obtain a noise-reduced echo signal;
s2, performing target tracking on the birds and the rotor unmanned aerial vehicle by adopting an interactive multi-model algorithm, and extracting the motion model conversion frequency of the birds and the rotor unmanned aerial vehicle targets as motion characteristics;
s3, obtaining a time-frequency matrix of the bird and the rotor unmanned aerial vehicle by adopting STFT, performing singular decomposition on the time-frequency matrix, and constructing a characteristic spectrum energy entropy as a micro-motion characteristic;
and S4, utilizing the obtained motion model to convert frequency and characteristic spectrum energy entropy, constructing a fused characteristic quantity, inputting the fused characteristic quantity into K-means, and distinguishing the bird and the rotor unmanned aerial vehicle.
2. The method of claim 1, wherein in step S1, the method of base tracking noise reduction comprises:
the radar echo signal y of the bird or unmanned aerial vehicle is:
y＝s+n
wherein y is a radar echo signal of a bird or unmanned aerial vehicle, s is a target signal, and n is noise;
The noise reduction problem of the radar echo signal is described as finding a sparse solution k, enabling a target signal to be sparsely represented in a frequency domain B, using an L1 norm as a measure of sparsity, and constructing an expression of an optimization problem as follows:
wherein k is opt For the optimal solution of k, B is a frequency domain sparse representation matrix, k is a sparse representation coefficient vector of a target signal in a frequency domain, lambda is a balance vector, lambda.k is multiplication of the vector lambda and the coefficient vector k by elements, I I.I is L1 norm,is the L2 norm.
3. The method for distinguishing a target of a bird from a rotor unmanned aerial vehicle under the condition of low signal-to-noise ratio according to claim 2, wherein in step S1, the base tracking noise reduction method is solved by using a split augmented lagrangian contraction algorithm, and the constructed optimization problem is solved to obtain an optimal sparse solution, and an echo signal spectrum is reconstructed.
4. The method for distinguishing between a bird and a rotorcraft object at a low signal-to-noise ratio according to claim 1, wherein in step S2, extracting a motion model conversion frequency of the bird and the rotorcraft object as a motion feature, specifically comprises:
after interactive multi-model algorithm target tracking is carried out on birds and rotor unmanned aerial vehicles, model probability mu of r models is calculated in continuous T moments n (k) The method comprises the steps of carrying out a first treatment on the surface of the n=1, 2 … r; k=1, 2 … T, the variance average is the motion model conversion frequency estimation value F, and the expression is:
wherein F is the variance average value as the motion model conversion frequency estimation value, r is the number of target motion models, T is the time interval, and var {. Cndot. Is the variance operation.
5. The method for distinguishing between bird and rotorcraft objects with low signal-to-noise ratio according to claim 4, wherein the motion state of the interactive multi-model algorithm object, if any, is smaller than the object threshold value with multiple motion states, the threshold value is set to be Q, wherein 0< Q <1, and when the motion model conversion frequency estimated value is higher than the threshold value Q, the object is determined to be bird; and if the value is lower than the threshold value Q, judging as a rotor unmanned aerial vehicle target.
6. Method for discriminating between bird and rotorcraft objects at low signal-to-noise ratio according to claim 1 characterized in that in step S3 the characteristic spectral energy entropy is constructed as micro-motion characteristic, comprising in particular:
short-time Fourier transform is carried out on radar echoes of birds and rotor unmanned aerial vehicles, a target signal is s (t), a window function is g (t), and then the short-time Fourier transform expression is:
Wherein X is STFT (t, f) is the time spectrum of the radar echo signal, t is time, f is frequency, s (τ) is the target signal, g (τ -t) is the window function, e is the index, τ is the variable time, j is the imaginary unit, dτ is the small variation of τ, e -j2πfτ For exponential operation, infinity;
obtaining a time-frequency matrix X of the bird and the rotor unmanned aerial vehicle through short-time Fourier transform, and carrying out singular value decomposition on the time-frequency matrix X to obtain the following components:
X＝U∑V T
wherein, sigma is a diagonal matrix formed by corresponding singular values after singular value decomposition, p is the number of singular values; (. Cndot. T For transpose operation, < >>For the first singular value, +.>For the second singular value, +.>Is the P th singular value; u is an N x N standard orthogonal matrix, V is an M x M standard orthogonal matrix, U and V are respectively a left singular matrix formed by left singular vectors and a right singular matrix formed by right singular vectors, and the left singular vectors representFrequency information of the echo signal, and right singular vectors represent time information of the echo signal;
and the micro motion feature extraction of the unmanned aerial vehicle and birds is realized through the related feature quantity of the singular value decomposition structure.
7. The method for distinguishing between bird and rotorcraft objects at low signal-to-noise ratio of claim 6, wherein the expression for the characteristic spectrum energy entropy using left singular vectors is:
E n ＝|U n,1 | 2 +|U n,2 | 2
Wherein eta is the energy entropy of the characteristic spectrum and q n The energy ratio of the left singular vector corresponding to the first two large singular values of the nth sample, E n For the sum of the energy of the first two left singular vectors of the nth sample, |U n,1 | 2 ,|U n,2 | 2 The energy of the first and second left singular vectors representing the nth sample, N being the number of samples, lnq, respectively n Is q n Is a natural logarithm of (a) to (b),is the mth singular value.
8. The method for distinguishing the targets of the bird and the rotary-wing unmanned aerial vehicle under the condition of low signal-to-noise ratio according to claim 1, wherein in the step S4, the obtained motion model is utilized to convert frequency and characteristic spectrum energy entropy, the fused characteristic quantity is constructed and is input into K-means, and the distinguishing of the bird and the rotary-wing unmanned aerial vehicle is realized, and the method specifically comprises the following steps:
the extracted motion model converts the frequency estimation value into a 1 The energy entropy of the characteristic spectrum is a 2 The method comprises the steps of carrying out a first treatment on the surface of the The time interval of the target movement process of the bird or rotor unmanned aerial vehicle is T, the time interval of the target movement is divided into m time periods, and each time period is T i The method comprises the steps of carrying out a first treatment on the surface of the Calculating motion model conversion frequency estimation values of a bird or rotor unmanned aerial vehicle target, and simultaneously, respectively carrying out time period T i Performing inching feature analysis to extract feature spectrum energy entropy a 2 ；
For the obtained i characteristic spectrum energy entropy a 2i Averaging, i=1, 2 … m, characteristic spectrum energy entropy a 2 The calculation formula is as follows:
wherein i represents the number of time periods;
converting the frequency estimate a using the extracted motion model 1 Characteristic spectrum energy entropy a 2 The fused characteristic quantity A is constructed, and the calculation formula is as follows:
A＝(a 1 a 2 )
and inputting the characteristic quantity A into the K-means to realize the identification of the bird and the rotor unmanned aerial vehicle.
9. A system for distinguishing between bird and rotorcraft objects at low signal to noise ratio, wherein a method for distinguishing between bird and rotorcraft objects at low signal to noise ratio as claimed in any one of claims 1 to 8 is implemented, the system comprising:
the noise-reduced echo signal acquisition module is used for reconstructing radar echoes of the bird and rotor unmanned aerial vehicle targets based on a base tracking noise reduction method, enhancing micro-motion characteristics of the bird and rotor unmanned aerial vehicle targets and reducing noise to obtain noise-reduced echo signals;
the motion characteristic acquisition module is used for tracking targets of the birds and the rotor unmanned aerial vehicle by adopting an interactive multi-model algorithm, and extracting the motion model conversion frequency of the targets of the birds and the rotor unmanned aerial vehicle as motion characteristics;
the micro-motion feature acquisition module is used for obtaining a time-frequency matrix of the bird and the rotor unmanned aerial vehicle by adopting STFT, performing singular decomposition on the time-frequency matrix, and constructing a feature spectrum energy entropy as a micro-motion feature;
And the identification module of the bird and the rotor unmanned aerial vehicle utilizes the obtained motion model conversion frequency and the characteristic spectrum energy entropy to construct the fused characteristic quantity and input the fused characteristic quantity into the K-means so as to realize identification of the bird and the rotor unmanned aerial vehicle.
10. The system for distinguishing the targets of the bird and the rotary wing unmanned aerial vehicle under the condition of low signal to noise ratio according to claim 9, wherein the system for distinguishing the targets of the bird and the rotary wing unmanned aerial vehicle under the condition of low signal to noise ratio is carried on a radar recognition network platform for classifying and recognizing the unmanned aerial vehicle and the bird.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310638119.0A CN116660851A (en) | 2023-06-01 | 2023-06-01 | Method and system for distinguishing targets of birds and rotor unmanned aerial vehicle under low signal-to-noise ratio condition |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310638119.0A CN116660851A (en) | 2023-06-01 | 2023-06-01 | Method and system for distinguishing targets of birds and rotor unmanned aerial vehicle under low signal-to-noise ratio condition |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116660851A true CN116660851A (en) | 2023-08-29 |
Family
ID=87716701
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310638119.0A Pending CN116660851A (en) | 2023-06-01 | 2023-06-01 | Method and system for distinguishing targets of birds and rotor unmanned aerial vehicle under low signal-to-noise ratio condition |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116660851A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117949917A (en) * | 2024-03-26 | 2024-04-30 | 中国民航大学 | Method for identifying bird targets in apron clutter environment and storage medium |
-
2023
- 2023-06-01 CN CN202310638119.0A patent/CN116660851A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117949917A (en) * | 2024-03-26 | 2024-04-30 | 中国民航大学 | Method for identifying bird targets in apron clutter environment and storage medium |
| CN117949917B (en) * | 2024-03-26 | 2024-05-28 | 中国民航大学 | Method for identifying bird targets in apron clutter environment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Huizing et al. | Deep learning for classification of mini-UAVs using micro-Doppler spectrograms in cognitive radar | |
| CN107728142B (en) | Radar high-resolution range profile target identification method based on two-dimensional convolutional network | |
| Molchanov et al. | Classification of aircraft using micro-Doppler bicoherence-based features | |
| CN109829509B (en) | Radar gesture recognition method based on fusion neural network | |
| Liu et al. | Deep learning and recognition of radar jamming based on CNN | |
| CN103885043B (en) | The miscellaneous Robust classification method of making an uproar of Aircraft Targets based on broad match filtering | |
| CN109901130B (en) | Rotor unmanned aerial vehicle detection and identification method based on Radon transformation and improved 2DPCA | |
| CN107728143A (en) | Radar High Range Resolution target identification method based on one-dimensional convolutional neural networks | |
| CN113534065B (en) | Radar target micro-motion feature extraction and intelligent classification method and system | |
| CN111401168B (en) | Multilayer radar feature extraction and selection method for unmanned aerial vehicle | |
| CN116660851A (en) | Method and system for distinguishing targets of birds and rotor unmanned aerial vehicle under low signal-to-noise ratio condition | |
| CN111693953B (en) | micro-Doppler-based target classification recognition model, method, system and device | |
| CN112882012B (en) | Radar target noise robust identification method based on signal-to-noise ratio matching and echo enhancement | |
| CN115343704A (en) | Gesture recognition method of FMCW millimeter wave radar based on multi-task learning | |
| CN109521411A (en) | A kind of detection method of range extension target | |
| CN117192491A (en) | Intermittent sampling forwarding interference resisting method and device for coded radar waveform | |
| Zhang et al. | Removal of micro-Doppler effect of ISAR image based on Laplacian regularized nonconvex low-rank representation | |
| Orduyilmaz et al. | Machine learning-based radar waveform classification for cognitive EW | |
| CN116699526A (en) | Vehicle millimeter wave radar interference suppression method based on sparse and low-rank model | |
| Zhu et al. | Radar HRRP group-target recognition based on combined methods in the backgroud of sea clutter | |
| CN112327286A (en) | Low-complexity daily activity classification method, device, equipment and storage medium | |
| Yuan | A time-frequency feature fusion algorithm based on neural network for HRRP | |
| CN109766899B (en) | Physical feature extraction and SVM SAR image vehicle target recognition method | |
| Dai et al. | UAVs and birds classification using robust coordinate attention synergy residual split-attention network based on micro-Doppler signature measurement by using L-band staring radar | |
| Shreyamsha Kumar et al. | Target identification using harmonic wavelet based ISAR imaging |
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 |