CN111126226A - Radiation source individual identification method based on small sample learning and feature enhancement - Google Patents
Radiation source individual identification method based on small sample learning and feature enhancement Download PDFInfo
- Publication number
- CN111126226A CN111126226A CN201911302131.4A CN201911302131A CN111126226A CN 111126226 A CN111126226 A CN 111126226A CN 201911302131 A CN201911302131 A CN 201911302131A CN 111126226 A CN111126226 A CN 111126226A
- Authority
- CN
- China
- Prior art keywords
- feature
- matrix
- enhanced
- enhancement
- radiation source
- 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.)
- Granted
Links
Images
Classifications
-
- 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
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a radiation source individual identification method based on small sample learning and feature enhancement, which comprises the following steps: s11, carrying out short-time Fourier transform and gray processing on the radar pulse signals of different individuals to obtain a gray matrix of the radar pulse signals; s12, inputting the obtained gray matrix into a sparse self-encoder for feature extraction to obtain a feature matrix; s13, performing feature enhancement on the extracted feature matrix in an image enhancement mode to obtain an enhanced feature matrix; s14, judging whether the number of the samples of the radar pulse signals is smaller than a threshold value, if so, inputting the enhanced feature matrix into an enhanced conditional countermeasure generation network to expand the number of the samples, and returning to the step S14; if not, go to step S15; and S15, inputting the enhanced feature matrix into a convolutional neural network for training, and inputting the features into a classifier for classification and identification.
Description
Technical Field
The invention relates to the technical field of radar individual identification, in particular to a radiation source individual identification method based on small sample learning and feature enhancement.
Background
With the continuous development of modern electronic information technology, electronic countermeasure among countries is more and more intense, the types, modulation modes and frequency ranges of signals emitted by radar radiation source individuals are various, and an effective radiation source individual identification algorithm is urgently found. The identification of individuals with radar radiation sources is an important component in radar electronic countermeasure, and plays an important role in electronic support and threat warning systems. Radar source individual identification is one of the important functions of electronic support measures esm (electronic support measures) and intelligence reconnaissance, and intercepts, locates, analyzes and identifies radar signals.
A large number of scholars discover through studying the transmission mechanism of the radar radiation source transmitter that although the radar transmitter can transmit various radar signals, no matter how factors such as waveform, modulation mode and environment change, signal nonlinearity caused by differences of internal components of the radar transmitter can affect the signals to a certain extent. Therefore, the radiation source individual identification technology is provided, the radiation source individual identification technology does not care which type of signals are emitted by the radar transmitter any more, but only care that when the radiation source individual emitting radar signals emits the same type of signals, the individual characteristics of the radiation source carried by the radar signals generate a slight change rule for radar waveforms, and the identification of the radar radiation source individual and even a carrying platform can be realized by analyzing and extracting the slight change rule of the signals. However, the traditional radar radiation source identification method is limited in capacity and low in reliability, and effective fine features cannot be extracted. Therefore, finding a more effective method to extract the fine features in the radar signal and obtain an effective identification result has important research significance.
In recent years, a deep learning algorithm is researched by a large number of scholars along with the development of artificial intelligence, and a large number of experiments show that the deep learning algorithm can process data with complex rules and distribution and the effectiveness is greatly improved. The deep learning algorithm inputs data into a network of one layer and another layer, potential feature information in the data is automatically extracted through a plurality of iterations and function mapping modes, and finally the feature information is classified through a deep learning classifier. In the field of radar radiation source individual identification, potential individual subtle features in radar radiation source signals can be automatically extracted by using a deep learning algorithm, and radar radiation source individual identification is better realized.
Disclosure of Invention
The invention aims to provide a radiation source individual identification method based on small sample learning and feature enhancement, aiming at the defects of the prior art, so that the identification probability of radar individuals of the same type can be improved, and the identification range of radar signals can be enlarged.
In order to achieve the purpose, the invention adopts the following technical scheme:
a radiation source individual identification method based on small sample learning and feature enhancement comprises the following steps:
s1, carrying out short-time Fourier transform and gray processing on radar pulse signals of different individuals to obtain a gray matrix of the radar pulse signals;
s2, inputting the obtained gray matrix into a sparse self-encoder for feature extraction to obtain a feature matrix;
s3, performing feature enhancement on the extracted feature matrix in an image enhancement mode to obtain an enhanced feature matrix;
s4, judging whether the number of the samples of the radar pulse signals is smaller than a threshold value, if so, inputting the enhanced feature matrix into an enhanced conditional countermeasure generation network to expand the number of the samples, and returning to the step S4; if not, go to step S5;
and S5, inputting the enhanced feature matrix into a convolutional neural network for training, and inputting the features into a classifier for classification and identification.
Further, step S1 is specifically to perform short-time fourier transform on the radar signals of different individuals to obtain a time-frequency diagram matrix Z (t, f); and carrying out gray processing on the obtained time-frequency diagram matrix Z (t, f) to obtain a gray matrix G.
Further, the step S1 includes performing vectorization operation on the obtained gray-scale matrix G to obtain a gray-scale matrix V of the radar pulse signal.
Further, the step S2 is specifically to input the obtained grayscale matrix V into a sparse self-encoder for feature extraction, and obtain a feature matrix H by adjusting network parameters.
Further, the image enhancement method in step S3 includes one or more of histogram modification, gray scale transformation, local statistics, image filtering, and color enhancement.
Further, in step S3, the feature matrix H is output to be feature enhanced in an image enhancement mode, and the enhanced feature matrix is output to be feature enhanced K.
Further, in step S4, if the number of samples of the radar pulse signal is smaller than the threshold, the enhanced feature matrix K is input into the enhanced conditional robust generation network to expand the number of samples, a generated network model in the enhanced conditional robust generation network is extracted after training is completed, a plurality of generated feature samples are obtained through the generated network model, and the generated feature samples and the enhanced feature samples are mixed to obtain a new enhanced feature matrix.
Further, the step S5 of inputting the features into the classifier for classification and identification is performed by using a SoftMax classifier for identification
Compared with the prior art, the method comprehensively considers the difference caused by unintentional modulation between radar individual signals, extracts the corresponding characteristics of the radar individual signals aiming at the difference in the aspects, enhances the characteristics, realizes small sample learning through a generative countermeasure network, and finally uses a convolutional neural network to train and identify to ensure the effectiveness and reliability of classification and identification of different radar radiation source pulse signals.
Drawings
FIG. 1 is a flow chart of a method for identifying individuals of a radiation source based on small sample learning and feature enhancement provided by an embodiment;
fig. 2 is a schematic diagram of an implementation of a radiation source individual identification algorithm based on feature enhancement under a small sample condition provided in the first embodiment;
FIG. 3 is a schematic structural diagram of a sparse self-encoder according to a first embodiment;
fig. 4 is a schematic diagram of an enhanced condition generating network according to an embodiment.
Detailed Description
The embodiments of the present invention are described below with reference to specific embodiments, and other advantages and effects of the present invention will be easily understood by those skilled in the art from the disclosure of the present specification. The invention is capable of other and different embodiments and of being practiced or of being carried out in various ways, and its several details are capable of modification in various respects, all without departing from the spirit and scope of the present invention. It is to be noted that the features in the following embodiments and examples may be combined with each other without conflict.
The invention aims to overcome the defects of the prior art and provides a radiation source individual identification method based on small sample learning and feature enhancement.
Aiming at the problem of low signal identification performance of the traditional radar radiation source, the invention can improve the identification probability of the radar individuals of the same type by adding the deep learning algorithm and enlarge the identification range of the radar signals.
Example one
The embodiment provides a radiation source individual identification method based on small sample learning and feature enhancement, as shown in fig. 1-2, comprising the steps of:
s11, carrying out short-time Fourier transform and gray processing on the radar pulse signals of different individuals to obtain a gray matrix of the radar pulse signals;
s12, inputting the obtained gray matrix into a sparse self-encoder for feature extraction to obtain a feature matrix;
s13, performing feature enhancement on the extracted feature matrix in an image enhancement mode to obtain an enhanced feature matrix;
s14, judging whether the number of the samples of the radar pulse signals is smaller than a threshold value, if so, inputting the enhanced feature matrix into an enhanced conditional countermeasure generation network to expand the number of the samples, and returning to the step S14; if not, go to step S15;
and S15, inputting the enhanced feature matrix into a convolutional neural network for training, and inputting the features into a classifier for classification and identification.
In step S11, a grayscale matrix of the radar pulse signal is obtained by performing short-time fourier transform and grayscale processing on the radar pulse signals of different individuals.
In this embodiment, the data is preprocessed.
Firstly, carrying out short-time Fourier transform (STFT) on a received radar pulse signal s (t) to obtain a time-frequency graph matrix Z (t, f) of the radar pulse signal s (t);
the time-frequency diagram matrix Z (t, f) of the signal s (t) can be expressed as:
where denotes the complex conjugate, h (t) is a window function.
Then carrying out gray processing on the time-frequency diagram matrix Z (t, f) to obtain a gray matrix G; and finally, performing vectorization operation on the gray matrix G to obtain an N-dimensional vector v as an input sample. And performing the operation on the acquired M pulse signals under the plurality of radar radiation sources to obtain an M multiplied by N gray matrix V.
The hypothesis time-frequency graph becomes a vector after vectorization operationThe grayscale matrix V can be expressed as:
in step S12, the obtained grayscale matrix is input to a sparse self-encoder to perform feature extraction, thereby obtaining a feature matrix.
In an embodiment, feature extraction is performed.
Inputting the gray matrix V into a sparse self-encoder (SAE) for feature extraction, and adjusting network parameters to enable a network to output a feature matrix H with dimension of M multiplied by nM×NHere N ∈ N.
Sparse Autoencoder (SAE) is an unsupervised machine learning algorithm whose basic structure is shown in FIG. 3, and whose training goal is to make its output equal to the input, then get a compressed representation of the input through the hidden layer, where the resulting hidden layer output is denoted by h, and in sparse autoencoding neural networks using hjRepresenting the output of the jth cell of the hidden layer. After training the neural network, there is a new sample viAfter the coefficient is input into the trained coefficient self-encoder, the signal feature vector h ═ h composed of the activation values of the units in the hidden layer1,h2,...,hn]Is the sample viAnd (5) reducing the vector after dimension reduction.
In step S13, feature enhancement is performed on the extracted feature matrix by an image enhancement method, and an enhanced feature matrix is obtained.
In the present embodiment, feature enhancement is performed.
And performing feature enhancement on the output feature matrix H by using an image enhancement method (such as a histogram correction method, a gray level transformation method, a local statistical method, an image filtering method, a color enhancement method and the like), and finally outputting to obtain an enhanced feature matrix K.
The image enhancement method is specifically described by taking histogram equalization and specification as an example:
histogram equalization and stipulation are carried out on the output feature matrix H, so that the original feature matrix H is projected into a newly constructed homogenization space, the purpose of feature enhancement is achieved, and finally, an enhanced feature matrix K is outputM×n。
The histogram in image processing mainly refers to a gray level histogram, and the expression is as follows:
P(rk)=nk/n k=0,1,...,L-1
wherein n is the characteristic length extracted after SAE, rkThe k-th gray level is represented,nkrepresenting grey levels r in an imagekThe number of pixels present, P (r)k) Representing a grey level rkThe probability of occurrence.
Therefore, the gray level histogram can reflect the probability statistical characteristics of the gray level values in the characteristic image, and the distribution of the gray level can be changed through probability transformation, so that the required characteristic quantity in the image is highlighted.
Histogram equalization is to modify a histogram in an original image into a histogram with uniformly distributed gray levels by using a gray level transformation function, and the image looks clear due to the increase of the dynamic range of the gray levels.
Let r, s denote the gray levels of the original image and the enhanced image, respectively, and have been normalized. When r ═ s ═ 0, it represents black; when r ═ s ═ 1, white is indicated. The gray scale transformation function is such that the following two conditions are satisfied: (1) r is more than or equal to 0 and less than or equal to 1, and T (r) is monotonically increased; (2) r is more than or equal to 0 and less than or equal to 1, and T is more than or equal to 0 and less than or equal to 1 (r).
Cumulative distribution function s of gradation of imagekComprises the following steps:
the histogram specification is mainly divided into the following three steps:
(1) raw histogram equalization
(2) The desired histogram is determined while homogenizing.
(3) And projecting the original space into the newly constructed homogenized space, thereby achieving the purpose of feature enhancement.
In step S14, it is determined whether the number of samples of the radar pulse signal is less than a threshold, and if so, the enhanced feature matrix is input to the enhanced conditional countermeasure generation network to expand the number of samples, and then the process returns to step S14; if not, step S15 is executed.
In this embodiment, sample expansion is performed.
When the number M of radar pulse signal samples is small, the feature matrix K is enhancedM×nInputting the data into an enhanced conditional countermeasure generating network (SCGAN) for training, extracting a generating network Model _ G in the SCGAN after the training is finished, and obtaining M through the Model _ G1Generating characteristic samples, mixing the generated characteristic samples with the enhanced characteristic samples to obtain a new enhanced characteristic matrixIf the number of radar pulse signal samples is sufficient, this step is skipped.
The SCGAN changes condition information y on the basis of CGAN, adds an aliasing penalty term in a loss function of a decision network, and combines an error label with a real sample to be input into the decision network. The basic structure of SCGAN is shown in fig. 4. The objective function is:
wherein the content of the first and second substances,indicating an error label that does not match the authentic sample.
In step S15, the enhanced feature matrix is input to a convolutional neural network for training, and the features are input to a classifier for classification and recognition.
In the present embodiment, classification recognition is performed.
And inputting the enhanced feature matrix K into a Convolutional Neural Network (CNN) for training, finally identifying through a SoftMax classifier, and outputting an identification probability.
The method comprehensively considers the difference caused by unintentional modulation between radar individual signals, extracts corresponding characteristics of the radar individual signals according to the difference, enhances the characteristics, and guarantees the effectiveness and the reliability of classification and identification of different radar radiation source pulse signals through convolutional neural network training and identification.
It is to be noted that the foregoing is only illustrative of the preferred embodiments of the present invention and the technical principles employed. It will be understood by those skilled in the art that the present invention is not limited to the particular embodiments described herein, but is capable of various obvious changes, rearrangements and substitutions as will now become apparent to those skilled in the art without departing from the scope of the invention. Therefore, although the present invention has been described in greater detail by the above embodiments, the present invention is not limited to the above embodiments, and may include other equivalent embodiments without departing from the spirit of the present invention, and the scope of the present invention is determined by the scope of the appended claims.
Claims (8)
1. A radiation source individual identification method based on small sample learning and feature enhancement is characterized by comprising the following steps:
s1, carrying out short-time Fourier transform and gray processing on radar pulse signals of different individuals to obtain a gray matrix of the radar pulse signals;
s2, inputting the obtained gray matrix into a sparse self-encoder for feature extraction to obtain a feature matrix;
s3, performing feature enhancement on the extracted feature matrix in an image enhancement mode to obtain an enhanced feature matrix;
s4, judging whether the number of the samples of the radar pulse signals is smaller than a threshold value, if so, inputting the enhanced feature matrix into an enhanced conditional countermeasure generation network to expand the number of the samples, and returning to the step S4; if not, go to step S5;
and S5, inputting the enhanced feature matrix into a convolutional neural network for training, and inputting the features into a classifier for classification and identification.
2. The method for identifying individual radiation source based on small sample learning and feature enhancement as claimed in claim 1, wherein the step S1 is specifically to perform short-time fourier transform on radar signals of different individuals to obtain a time-frequency diagram matrix Z (t, f); and carrying out gray processing on the obtained time-frequency diagram matrix Z (t, f) to obtain a gray matrix G.
3. The method for identifying individuals as radiation sources based on small sample learning and feature enhancement as claimed in claim 2, wherein the step S1 further includes vectorizing the obtained gray matrix G to obtain a gray matrix V of the radar pulse signal.
4. The method as claimed in claim 3, wherein the step S2 is to input the obtained gray matrix V into a sparse self-encoder for feature extraction, and obtain a feature matrix H by adjusting network parameters.
5. The method for individual identification of radiation source based on small sample learning and feature enhancement as claimed in claim 1, wherein the image enhancement mode in step S3 includes one or more of histogram modification, gray scale transformation, local statistics, image filtering, and color enhancement.
6. The method for individual identification of a radiation source based on small sample learning and feature enhancement as claimed in claim 4, wherein the step S3 is specifically to perform feature enhancement on the output feature matrix H by means of image enhancement, and output the enhanced feature matrix for feature enhancement K.
7. The method for individual identification of a radiation source based on small sample learning and feature enhancement as claimed in claim 6, wherein in step S4, if the number of samples of the radar pulse signal is less than a threshold, the enhanced feature matrix K is input into the enhanced conditional robust generation network to expand the number of samples, after training is completed, a generation network model in the enhanced conditional robust generation network is extracted, a plurality of generated feature samples are obtained through the generation network model, and the generated feature samples and the enhanced feature samples are mixed to obtain a new enhanced feature matrix.
8. The individual identification method for the radiation source based on the small sample learning and the feature enhancement as claimed in claim 1, wherein the step S5 of inputting the features into the classifier for classification and identification is performed by a SoftMax classifier.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911302131.4A CN111126226B (en) | 2019-12-17 | 2019-12-17 | Radiation source individual identification method based on small sample learning and feature enhancement |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911302131.4A CN111126226B (en) | 2019-12-17 | 2019-12-17 | Radiation source individual identification method based on small sample learning and feature enhancement |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111126226A true CN111126226A (en) | 2020-05-08 |
CN111126226B CN111126226B (en) | 2023-07-28 |
Family
ID=70498236
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201911302131.4A Active CN111126226B (en) | 2019-12-17 | 2019-12-17 | Radiation source individual identification method based on small sample learning and feature enhancement |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111126226B (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111741018A (en) * | 2020-07-24 | 2020-10-02 | 中国航空油料集团有限公司 | Industrial control data attack sample generation method and system, electronic device and storage medium |
CN111832404A (en) * | 2020-06-04 | 2020-10-27 | 中国科学院空天信息创新研究院 | Small sample remote sensing ground feature classification method and system based on feature generation network |
CN112034434A (en) * | 2020-09-04 | 2020-12-04 | 中国船舶重工集团公司第七二四研究所 | Radar radiation source identification method based on sparse time-frequency detection convolutional neural network |
CN112084906A (en) * | 2020-08-27 | 2020-12-15 | 上海朱光亚战略科技研究院 | Radar pulse signal classification method and device, computer equipment and storage medium |
CN114609493A (en) * | 2022-05-09 | 2022-06-10 | 杭州兆华电子股份有限公司 | Partial discharge signal identification method with enhanced signal data |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100253540A1 (en) * | 2009-04-02 | 2010-10-07 | Gm Global Technology Operations, Inc. | Enhanced road vision on full windshield head-up display |
CN101882304A (en) * | 2010-06-24 | 2010-11-10 | 西北工业大学 | Self-adaptive de-noising and characteristic enhancing method of SAR (Synthetic Aperture Radar) image |
US20170200092A1 (en) * | 2016-01-11 | 2017-07-13 | International Business Machines Corporation | Creating deep learning models using feature augmentation |
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 |
CN109409413A (en) * | 2018-09-28 | 2019-03-01 | 贵州大学 | X-ray breast lump image automatic classification method |
US20190197358A1 (en) * | 2017-12-21 | 2019-06-27 | International Business Machines Corporation | Generative Adversarial Network Medical Image Generation for Training of a Classifier |
CN110472483A (en) * | 2019-07-02 | 2019-11-19 | 五邑大学 | A kind of method and device of the small sample semantic feature enhancing towards SAR image |
-
2019
- 2019-12-17 CN CN201911302131.4A patent/CN111126226B/en active Active
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100253540A1 (en) * | 2009-04-02 | 2010-10-07 | Gm Global Technology Operations, Inc. | Enhanced road vision on full windshield head-up display |
CN101882304A (en) * | 2010-06-24 | 2010-11-10 | 西北工业大学 | Self-adaptive de-noising and characteristic enhancing method of SAR (Synthetic Aperture Radar) image |
US20170200092A1 (en) * | 2016-01-11 | 2017-07-13 | International Business Machines Corporation | Creating deep learning models using feature augmentation |
US20190197358A1 (en) * | 2017-12-21 | 2019-06-27 | International Business Machines Corporation | Generative Adversarial Network Medical Image Generation for Training of a Classifier |
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 |
CN109409413A (en) * | 2018-09-28 | 2019-03-01 | 贵州大学 | X-ray breast lump image automatic classification method |
CN110472483A (en) * | 2019-07-02 | 2019-11-19 | 五邑大学 | A kind of method and device of the small sample semantic feature enhancing towards SAR image |
Non-Patent Citations (4)
Title |
---|
"GAN-based synthetic medical image augmentation for increased CNN performance in liver lesion classification", 《NEUROCOMPUTING》, vol. 321, pages 321 - 331 * |
"高频地波雷达海杂波抑制与目标检测技术研究", 《中国博士学位论文全文数据库基础科学辑》, pages 1 - 114 * |
RICCARDO VOLPI ET AL.: "Adversarial Feature Augmentation for Unsupervised Domain Adaptation", pages 5495 - 5504 * |
朱克凡: "小样本条件下SCGAN+CNN 低分辨雷达目标 一步识别算法", pages 1 - 13 * |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111832404A (en) * | 2020-06-04 | 2020-10-27 | 中国科学院空天信息创新研究院 | Small sample remote sensing ground feature classification method and system based on feature generation network |
CN111832404B (en) * | 2020-06-04 | 2021-05-18 | 中国科学院空天信息创新研究院 | Small sample remote sensing ground feature classification method and system based on feature generation network |
CN111741018A (en) * | 2020-07-24 | 2020-10-02 | 中国航空油料集团有限公司 | Industrial control data attack sample generation method and system, electronic device and storage medium |
CN111741018B (en) * | 2020-07-24 | 2020-12-01 | 中国航空油料集团有限公司 | Industrial control data attack sample generation method and system, electronic device and storage medium |
CN112084906A (en) * | 2020-08-27 | 2020-12-15 | 上海朱光亚战略科技研究院 | Radar pulse signal classification method and device, computer equipment and storage medium |
CN112034434A (en) * | 2020-09-04 | 2020-12-04 | 中国船舶重工集团公司第七二四研究所 | Radar radiation source identification method based on sparse time-frequency detection convolutional neural network |
CN112034434B (en) * | 2020-09-04 | 2022-05-20 | 中国船舶重工集团公司第七二四研究所 | Radar radiation source identification method based on sparse time-frequency detection convolutional neural network |
CN114609493A (en) * | 2022-05-09 | 2022-06-10 | 杭州兆华电子股份有限公司 | Partial discharge signal identification method with enhanced signal data |
CN114609493B (en) * | 2022-05-09 | 2022-08-12 | 杭州兆华电子股份有限公司 | Partial discharge signal identification method with enhanced signal data |
Also Published As
Publication number | Publication date |
---|---|
CN111126226B (en) | 2023-07-28 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111126226A (en) | Radiation source individual identification method based on small sample learning and feature enhancement | |
CN110084281B (en) | Image generation method, neural network compression method, related device and equipment | |
CN109254274B (en) | Radar radiation source identification method based on feature fusion | |
CN109583501B (en) | Method, device, equipment and medium for generating image classification and classification recognition model | |
CN111126386B (en) | Sequence domain adaptation method based on countermeasure learning in scene text recognition | |
CN109766835B (en) | SAR target recognition method for generating countermeasure network based on multi-parameter optimization | |
CN111325152A (en) | Deep learning-based traffic sign identification method | |
CN111507993A (en) | Image segmentation method and device based on generation countermeasure network and storage medium | |
US20230334632A1 (en) | Image recognition method and device, and computer-readable storage medium | |
CN110109060A (en) | A kind of radar emitter signal method for separating and system based on deep learning network | |
CN108021908B (en) | Face age group identification method and device, computer device and readable storage medium | |
US11067786B2 (en) | Artifact regulation methods in deep model training for image transformation | |
CN113537031B (en) | Radar image target identification method for generating countermeasure network based on condition of multiple discriminators | |
CN111126332B (en) | Frequency hopping signal classification method based on contour features | |
US11551080B2 (en) | Learning dataset generation method, new learning dataset generation device and learning method using generated learning dataset | |
CN112232395B (en) | Semi-supervised image classification method for generating countermeasure network based on joint training | |
CN113159045A (en) | Verification code identification method combining image preprocessing and convolutional neural network | |
CN111986126A (en) | Multi-target detection method based on improved VGG16 network | |
Guo et al. | ELAA: An efficient local adversarial attack using model interpreters | |
CN114882278A (en) | Tire pattern classification method and device based on attention mechanism and transfer learning | |
CN114140645A (en) | Photographic image aesthetic style classification method based on improved self-supervision feature learning | |
CN112818774A (en) | Living body detection method and device | |
Olisah et al. | Understanding unconventional preprocessors in deep convolutional neural networks for face identification | |
US11373066B2 (en) | Deep model matching methods for image transformation | |
CN116704208A (en) | Local interpretable method based on characteristic relation |
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 |