CN112052899A - Single ship target SAR image generation method based on generation countermeasure network - Google Patents
Single ship target SAR image generation method based on generation countermeasure network Download PDFInfo
- Publication number
- CN112052899A CN112052899A CN202010916944.9A CN202010916944A CN112052899A CN 112052899 A CN112052899 A CN 112052899A CN 202010916944 A CN202010916944 A CN 202010916944A CN 112052899 A CN112052899 A CN 112052899A
- Authority
- CN
- China
- Prior art keywords
- image
- training
- network
- ship target
- generation
- 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 24
- 238000012549 training Methods 0.000 claims abstract description 75
- 238000012706 support-vector machine Methods 0.000 claims abstract description 10
- 238000005070 sampling Methods 0.000 claims abstract description 8
- 238000007781 pre-processing Methods 0.000 claims abstract description 5
- 238000010606 normalization Methods 0.000 claims description 6
- 238000012216 screening Methods 0.000 claims description 4
- 230000004913 activation Effects 0.000 claims description 3
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 150000001875 compounds Chemical class 0.000 claims description 3
- 230000008030 elimination Effects 0.000 claims description 3
- 238000003379 elimination reaction Methods 0.000 claims description 3
- 238000013507 mapping Methods 0.000 claims description 3
- 238000001514 detection method Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 238000000605 extraction Methods 0.000 description 3
- 208000009119 Giant Axonal Neuropathy Diseases 0.000 description 2
- 238000013256 Gubra-Amylin NASH model Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000009412 basement excavation Methods 0.000 description 2
- 238000013434 data augmentation Methods 0.000 description 2
- 201000003382 giant axonal neuropathy 1 Diseases 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000002910 structure generation Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 230000001131 transforming effect Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
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 provides a method for generating a single ship target SAR image based on a generation countermeasure network, which comprises the following steps: s1, preprocessing the acquired single ship target SAR image, generating a normalized image recognition sample, and performing down-sampling of different scales on the image recognition sample to generate a plurality of training images; s2, constructing a multi-scale full convolution pyramid network based on N +1 generated countermeasure networks; s3, establishing a network training model, and training a multi-scale full convolution pyramid network from coarse to fine based on a training image; s4, establishing a high-quality ship SAR image screener based on a support vector machine, and training the image screener; and S5, taking the output result of the network training model in the step S3 as the input of an image filter, and outputting correct samples of a plurality of single ship target SAR images through the image filter. The method can generate high-quality multi-scale new images with the same content according to the single ship target SAR image.
Description
Technical Field
The invention relates to the technical field of image processing, in particular to a single ship target SAR image generation method based on a generation countermeasure network.
Background
Deep neural network based target feature extraction and identification is very dependent on the number of samples in the target dataset. Because the existing ship SAR image data are less, how to obtain a better network training result by using limited data is a problem which needs to be solved urgently. The generation of the countermeasure network (GAN generated adaptive Networks) can realize effective sample expansion by generating the network, so that more samples are provided for subsequent target feature extraction and identification, and the utilization efficiency of the existing data is improved. However, experiments show that the quality of the ship SAR image generated by the GAN depends on the number of training samples to a great extent, and the problems of network collapse, unstable training process or incapability of training and the like easily occur in small sample learning or zero sample learning.
The document 'synthetic aperture radar image target recognition based on DRGAN and a Support Vector Machine (SVM)' proposes an SAR image target recognition algorithm based on DRGAN and SVM. The experimental result shows that under the standard operation condition with the variant, the classification precision of 97.97% can be achieved, the classification precision is better than that based on the CNN model, under the condition that certain error exists in SAR image target azimuth estimation, the trained GAN model is used as an SAR image target rotation estimator, and high SAR image target identification precision can be still achieved on the premise that complex sample preprocessing is not performed.
The document "application of synthetic aperture radar ship data augmentation based on a generative countermeasure network in improving a single multi-box detector" proposes a data augmentation technique for generating a countermeasure network based on pixel-to-pixel (pix2 pix). A data set for pix2pix GAN is made, and a SAR ship detection algorithm based on an improved single multi-box detector (SSD) is provided, and experimental results show that: after the generated sample is added into the original SSD, the detection precision is improved by 4.3 percent compared with the original SSD detection algorithm; when the generated samples were added to the improved SSD, the detection accuracy increased by 1.9% compared to the improved SSD.
The document 'SAR image ship target detection based on generation of countermeasure network and on-line difficult case excavation' proposes an SAR image ship target detection method based on generation of countermeasure network (GAN) and on-line difficult case excavation (OHEM). And transforming on the characteristic diagram by using a spatial transformation network to generate characteristic diagrams of ship samples with different sizes and rotation angles, so that the adaptability of the detector to ship targets with different sizes and rotation angles is improved.
Patent application CN108399625A discloses a method for generating an added azimuth angle discrimination model of a countermeasure network by depth convolution, which is based on target segmentation and edge extraction, and calculates a clockwise included angle between the north end in the vertical direction and the longest edge of a minimum circumscribed rectangle through an edge-circumscribed minimum circumscribed rectangle method, and the included angle is used as the azimuth angle of a generated image target, so as to generate an image in an oriented manner.
Patent application CN109190684A discloses a method for generating SAR image samples based on sketch and structure generation countermeasure network. The method mainly solves the problem of sample imbalance in SAR image semantic segmentation, can generate SAR image samples which are consistent with the ground structure of the original SAR image according to a sketch, and can solve the problem of sample imbalance of SAR image extremely heterogeneous region classification.
Patent application CN109766835A discloses a synthetic aperture radar SAR target recognition method for generating a countermeasure network based on multi-parameter optimization, which mainly solves the problems that in the prior art, the recognition rate is low during classifier training, and the classifier parameters obtained by training cannot be guaranteed to be optimal solutions. The accuracy of SAR target identification can be improved.
On the basis of a large number of image samples, the scholars expand the number of samples by using different types of GANs, and the target detection and identification precision is improved. However, the current ship sample set is unbalanced, the number of targets is small or even only one, and the current GAN model relies on a large amount of data, and for few samples and a single image, the problems of network collapse or poor quality of generated images and the like are easy to occur. .
Disclosure of Invention
The invention aims to provide a method for generating a single ship target SAR image based on a generation countermeasure network. Firstly, establishing a multi-scale full convolution pyramid network based on a plurality of GANs, and downsampling collected single ship target SAR images through downsampling factors of different scales to obtain a plurality of training images; learning the training images through the multi-scale full convolution pyramid network (each GAN is responsible for learning the training images with different scales), and generating high-quality and multi-scale fake samples with the same visual content; and finally, establishing a high-quality ship SAR image screener based on a support vector machine, and screening to obtain a high-quality single ship target SAR image correct sample. The invention can generate high-quality ship target SAR images with different scales according to a single ship target SAR image, and solves the problem that training samples are lacked when the SAR ship target is identified through a neural network.
In order to achieve the purpose, the invention provides a method for generating a single ship target SAR image based on a generation countermeasure network, which comprises the following steps:
s1, preprocessing the acquired single ship target SAR image to generate a normalized image recognition sample; downsampling the image recognition samples in different scales to generate a plurality of training images;
s2, constructing a multi-scale full convolution pyramid network based on N +1 generated countermeasure networks, wherein N is more than or equal to 1;
s3, establishing a network training model, and training the multi-scale full convolution pyramid network from coarse to fine based on the training image;
s4, establishing a high-quality ship SAR image screener based on a support vector machine, and training the image screener through a training image;
and S5, taking the output result of the network training model in the step S3 as the input of an image filter, and outputting correct samples of a plurality of single ship target SAR images through the image filter.
Preferably, step S1 includes:
s11, rotating the single ship target SAR image to a normal view angle;
s12, cutting the single ship target SAR image into a set size, and performing brightness normalization and noise elimination on the cut image to obtain the normalized image identification sample;
s13, carrying out rough-to-fine down-sampling on the image recognition sample to obtain a training image xN～x0(ii) a Training image xnDown-sampling by a factor rnR is more than 1; using LabelImg software for xnMarking is carried out; wherein N is an element of [1, N ∈]。
Preferably, in step S2, the multi-scale full convolution pyramid network includes: the sequentially connected zeroth to Nth generation countermeasure networks; wherein, the image scale obtained by the n-th generation countermeasure network is larger than that obtained by the n-1-th generation countermeasure network;
the nth generation countermeasure network includes: mutually coupled generators GnAnd a discriminator DnThe training image xnFor training the arbiter DnGenerator GnAnd a discriminator DnHas the same receptive field, wherein N is equal to 0, N]；
Generator GnThe system comprises a convolution network, a batch normalization layer and a LeakyReLU activation function; the convolution network comprises a plurality of convolution layers; generator Gn+1Is connected with the generator GnAn input terminal of (1); discriminator DnIs a Markov arbiter where N is e [0, N]。
Preferably, the network training model includes: training image xN～x0White spatial gaussian noise zN～z0A multi-scale full convolution pyramid network;
the multi-scale full convolution pyramid network is trained from coarse to fine by sequentially training the Nth to the zeroth generation countermeasure networks, and when the N-th generation countermeasure network finishes training, the generator GnAnd discriminator DnIs fixed N ∈ [0, N ∈ >](ii) a Wherein,
generator GNIs z isNGenerator GNOutputting a counterfeit sample
When N ≠ N, the generator GnIs z isnAnd
generator GnOutput of (2)
Presentation pair
Performing r times of upsampling;
is composed of Gn+1Outputting the forged sample;
the n-th generation of a loss function against the network is:
wherein L isadv(Gn,Dn) For penalising counterfeit samples against loss
And training image xnThe distance between them;
Lrec(Gn) For reconstruction of losses, for ensuring znCan generate a sum xnA similar sample;
when N is less than N, the reaction solution is,
when N is equal to N, the compound is,
Lrec(Gn)＝||GN(z*)-xN||2；
preferably, the data used for training the image filter in step S4 is the training image xN～x0。
Preferably, step S5 includes:
s51, sample to be forged
Inputting the image filter;
and S52, obtaining the confidence coefficient of the forged samples through an SVM classifier, and screening the forged samples with the confidence coefficient higher than a set threshold value as correct samples of the single ship target SAR image.
Compared with the prior art, the invention has the following beneficial effects:
(1) the method does not depend on the existing data set, and can learn to obtain a plurality of high-quality single ship target SAR images with different scales according to the single ship target SAR image, so that the problem of unbalanced sample of the ship target SAR image data set is solved;
(2) the method adopts a multi-scale full convolution pyramid structure, obtains multi-scale images by down-sampling a single image, and increases the perception of the network to different scales so as to generate a new sample image;
(3) in the invention, each GAN is responsible for learning the distribution information of different scales of the image, so that new samples with any size and aspect ratio can be generated, the samples have obvious change, and the integral structure and fine texture characteristics of the training image can be maintained.
Drawings
In order to more clearly illustrate the technical solution of the present invention, the drawings used in the description will be briefly introduced, and it is obvious that the drawings in the following description are an embodiment of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts according to the drawings:
FIG. 1 is a schematic diagram of a multi-scale full convolution pyramid network trained by a network training model according to the present invention;
FIG. 2 shows a generator G according to the present inventionnSchematic diagram of input and output results;
FIG. 3 is a flow chart of a single ship target SAR image generation method based on generation of a countermeasure network according to the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The invention provides a method for generating a single ship target SAR image based on a generation countermeasure network, which comprises the following steps as shown in figure 3:
s1, preprocessing the acquired single ship target SAR image to generate a normalized image recognition sample; downsampling the image recognition samples in different scales to generate a plurality of training images;
step S1 includes:
s11, rotating the single ship target SAR image to a normal view angle;
s12, cutting the single ship target SAR image into a set size, and performing brightness normalization and noise elimination on the cut image to obtain the normalized image identification sample;
s13, carrying out rough-to-fine down-sampling on the image recognition sample to obtain a training image xN～x0(ii) a Training xnDown-sampling by a factor rnR is more than 1; for xnLabeling (in this example, the training images are labeled by the sprite labeling software or the LabelImg software); wherein N is an element of [1, N ∈]。
S2, constructing a multi-scale full convolution pyramid network based on N +1 generated countermeasure networks, wherein N is more than or equal to 1;
in step S2, the multi-scale full convolution pyramid network includes: the sequentially connected zeroth to Nth generation countermeasure networks; wherein, the image scale obtained by the n-th generation countermeasure network is larger than that obtained by the n-1-th generation countermeasure network;
the nth generation countermeasure network includes: mutually coupled generators GnAnd a discriminator DnGenerator GnAnd a discriminator DnHas the same receptive field, wherein N is equal to 0, N]；
Generator GnThe system comprises a convolution network, a batch normalization layer and a LeakyReLU activation function; the convolution network comprises a plurality of convolution layers; generator Gn+1Is connected with the generator GnAn input terminal of (1); discriminator DnIs a Markov arbiter where N is e [0, N]。
S3, establishing a network training model, and training the multi-scale full convolution pyramid network from coarse to fine based on the training image; as shown in fig. 1, the network training model includes: training image xN～x0White spatial gaussian noise zN～z0A multi-scale full convolution pyramid network;
the multi-scale full convolution pyramid network is trained from coarse to fine by sequentially training the Nth to the zeroth generation countermeasure networks, and when the N-th generation countermeasure network finishes training, the generator GnAnd discriminator DnIs fixed N ∈ [0, N ∈ >](ii) a Wherein,
generator GNIs z isNGenerator GNOutputting a counterfeit sample
When N ≠ N, the generator GnIs z isnAnd
as shown in fig. 2, generator GnOutput of (2)
Presentation pair
Performing r times of upsampling;
is composed of Gn+1Outputting the forged sample;
the n-th generation of a loss function against the network is:
wherein L isadv(Gn,Dn) For penalising counterfeit samples against loss
And training image xnThe distance between them;
Lrec(Gn) For reconstruction of losses, for znCan generate a sum xnA similar sample;
when N is less than N, the reaction solution is,
when N is equal to N, the compound is,
Lrec(Gn)＝||GN(z*)-xN||2；
the data used to train the image filter in step S4 is the training image xN～x0。
S4, establishing a high-quality ship SAR image filter based on a support vector machine, and training an image xN～x0Training the image filter;
and S5, taking the output result of the network training model in the step S3 as the input of an image filter, and outputting correct samples of a plurality of single ship target SAR images through the image filter.
Step S5 includes:
s51, sample to be forged
Inputting the image filter;
and S52, obtaining the confidence coefficient of the forged samples through an SVM classifier, and screening the forged samples with the confidence coefficient higher than a set threshold value as correct samples of the single ship target SAR image.
While the invention has been described with reference to specific embodiments, the invention is not limited thereto, and various equivalent modifications and substitutions can be easily made by those skilled in the art within the technical scope of the invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (6)
1. A single ship target SAR image generation method based on generation of a countermeasure network is characterized by comprising the following steps:
s1, preprocessing the acquired single ship target SAR image to generate a normalized image recognition sample; downsampling the image recognition samples in different scales to generate a plurality of training images;
s2, constructing a multi-scale full convolution pyramid network based on N +1 generated countermeasure networks, wherein N is more than or equal to 1;
s3, establishing a network training model, and training the multi-scale full convolution pyramid network from coarse to fine based on the training image;
s4, establishing a high-quality ship SAR image screener based on a support vector machine, and training the image screener through a training image;
and S5, taking the output result of the network training model in the step S3 as the input of an image filter, and outputting correct samples of a plurality of single ship target SAR images through the image filter.
2. The method for generating a single ship target SAR image based on generation of an countermeasure network as claimed in claim 1, wherein the step S1 comprises:
s11, rotating the single ship target SAR image to a normal view angle;
s12, cutting the single ship target SAR image into a set size, and performing brightness normalization and noise elimination on the cut image to obtain the normalized image identification sample;
s13, carrying out rough-to-fine down-sampling on the image recognition sample to obtain a training image xN～x0(ii) a Training image xnDown-sampling by a factor rnR is more than 1; using LabelImg software for xnMarking is carried out; wherein N is an element of [1, N ∈]。
3. The method for generating a single ship target SAR image based on generation of a countermeasure network as claimed in claim 2, wherein said multi-scale full convolution pyramid network in step S2 comprises: the sequentially connected zeroth to Nth generation countermeasure networks; wherein, the image scale obtained by the n-th generation countermeasure network is larger than that obtained by the n-1-th generation countermeasure network;
the nth generation countermeasure network includes: mutually coupled generators GnAnd a discriminator DnThe training image xnFor training the arbiter DnGenerator GnAnd a discriminator DnHas the same receptive field, wherein N is equal to 0, N]；
Generator GnThe system comprises a convolution network, a batch normalization layer and a LeakyReLU activation function; the convolution network comprises a plurality of convolution layers; generator Gn+1Is connected with the generator GnAn input terminal of (1); discriminator DnIs a Markov arbiter where N is e [0, N]。
4. The method for generating a single ship target SAR image based on generation of a countermeasure network as claimed in claim 3, wherein the network training model comprises: training image xN～x0White spatial gaussian noise zN～z0A multi-scale full convolution pyramid network;
the multi-scale full convolution pyramid network is trained from coarse to fine by sequentially training the Nth to the zeroth generation countermeasure networks, and when the N-th generation countermeasure network finishes training, the generator GnAnd discriminator DnIs fixed N ∈ [0, N ∈ >](ii) a Wherein,
generator GNIs z isNGenerator GNOutputting a counterfeit sample
When N ≠ N, the generator GnIs z isnAnd
generator GnOutput of (2)
Presentation pair
Performing r times of upsampling;
is composed of Gn+1Outputting the forged sample;
the n-th generation of a loss function against the network is:
wherein L isadv(Gn,Dn) For penalising counterfeit samples against loss
And training image xnThe distance between them;
Lrec(Gn) For reconstruction of losses, for ensuring znCan generate a sum xnA similar sample;
when N is less than N, the reaction solution is,
when N is equal to N, the compound is,
Lrec(Gn)＝||GN(z*)-xN||2；
5. the method for generating a single ship target SAR image based on generation of a countermeasure network as claimed in claim 2, wherein the data used for training the image filter in step S4 is the training image xN～x0。
6. The method for generating a single ship target SAR image based on generation of an countermeasure network as claimed in claim 1, wherein the step S5 comprises:
s51, sample to be forged
Inputting the image filter;
and S52, obtaining the confidence coefficient of the forged samples through an SVM classifier, and screening the forged samples with the confidence coefficient higher than a set threshold value as correct samples of the single ship target SAR image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010916944.9A CN112052899A (en) | 2020-09-03 | 2020-09-03 | Single ship target SAR image generation method based on generation countermeasure network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010916944.9A CN112052899A (en) | 2020-09-03 | 2020-09-03 | Single ship target SAR image generation method based on generation countermeasure network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112052899A true CN112052899A (en) | 2020-12-08 |
Family
ID=73606724
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010916944.9A Pending CN112052899A (en) | 2020-09-03 | 2020-09-03 | Single ship target SAR image generation method based on generation countermeasure network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112052899A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112862748A (en) * | 2020-12-25 | 2021-05-28 | 重庆大学 | Multidimensional domain feature combined SAR (synthetic aperture radar) ship intelligent detection method |
| CN113012251A (en) * | 2021-03-17 | 2021-06-22 | 厦门大学 | SAR image automatic colorization method based on generation countermeasure network |
| CN113435509A (en) * | 2021-06-28 | 2021-09-24 | 山东力聚机器人科技股份有限公司 | Small sample scene classification and identification method and system based on meta-learning |
| CN113469913A (en) * | 2021-07-06 | 2021-10-01 | 中南大学 | Hot-rolled strip steel surface water drop removing method based on gradual cycle generation countermeasure network |
| CN113610097A (en) * | 2021-08-09 | 2021-11-05 | 西安电子科技大学 | SAR ship target segmentation method based on multi-scale similarity guide network |
| CN113962360A (en) * | 2021-10-09 | 2022-01-21 | 西安交通大学 | Method and system for enhancing sample data based on GAN network |
| CN113962883A (en) * | 2021-09-30 | 2022-01-21 | 西南电子技术研究所（中国电子科技集团公司第十研究所） | Method for enhancing ship target detection SAR image data |
| CN114882236A (en) * | 2021-05-19 | 2022-08-09 | 哈尔滨工业大学 | Method for automatically identifying underground cavity target of ground penetrating radar based on SinGAN algorithm |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108664874A (en) * | 2018-02-14 | 2018-10-16 | 北京广天夏科技有限公司 | Underground coal flow rate testing methods based on image recognition |
| CN109146784A (en) * | 2018-07-27 | 2019-01-04 | 徐州工程学院 | A kind of image super-resolution rebuilding method based on multiple dimensioned generation confrontation network |
| CN110610207A (en) * | 2019-09-10 | 2019-12-24 | 重庆邮电大学 | Small sample SAR image ship classification method based on transfer learning |
| CN111310791A (en) * | 2020-01-17 | 2020-06-19 | 电子科技大学 | Dynamic progressive automatic target identification method based on small sample number set |
-
2020
- 2020-09-03 CN CN202010916944.9A patent/CN112052899A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108664874A (en) * | 2018-02-14 | 2018-10-16 | 北京广天夏科技有限公司 | Underground coal flow rate testing methods based on image recognition |
| CN109146784A (en) * | 2018-07-27 | 2019-01-04 | 徐州工程学院 | A kind of image super-resolution rebuilding method based on multiple dimensioned generation confrontation network |
| CN110610207A (en) * | 2019-09-10 | 2019-12-24 | 重庆邮电大学 | Small sample SAR image ship classification method based on transfer learning |
| CN111310791A (en) * | 2020-01-17 | 2020-06-19 | 电子科技大学 | Dynamic progressive automatic target identification method based on small sample number set |
Non-Patent Citations (3)
| Title |
|---|
| 张明蕊: ""SAR图像数据分集与扩容方法研究"" * |
| 李诗怡等: ""基于多尺度生成对抗网络的SAR图像样本增广"" * |
| 李诗怡等: ""基于多尺度生成对抗网络的SAR图像样本增广"", 《激光与光电子学进展》 * |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112862748A (en) * | 2020-12-25 | 2021-05-28 | 重庆大学 | Multidimensional domain feature combined SAR (synthetic aperture radar) ship intelligent detection method |
| CN112862748B (en) * | 2020-12-25 | 2023-05-30 | 重庆大学 | Multi-dimensional domain feature combined SAR ship intelligent detection method |
| CN113012251A (en) * | 2021-03-17 | 2021-06-22 | 厦门大学 | SAR image automatic colorization method based on generation countermeasure network |
| CN113012251B (en) * | 2021-03-17 | 2022-05-03 | 厦门大学 | SAR image automatic colorization method based on generation countermeasure network |
| CN114882236A (en) * | 2021-05-19 | 2022-08-09 | 哈尔滨工业大学 | Method for automatically identifying underground cavity target of ground penetrating radar based on SinGAN algorithm |
| CN114882236B (en) * | 2021-05-19 | 2024-09-13 | 哈尔滨工业大学 | Automatic identification method for underground cavity target of ground penetrating radar based on SinGAN algorithm |
| CN113435509A (en) * | 2021-06-28 | 2021-09-24 | 山东力聚机器人科技股份有限公司 | Small sample scene classification and identification method and system based on meta-learning |
| CN113435509B (en) * | 2021-06-28 | 2022-03-25 | 山东力聚机器人科技股份有限公司 | Small sample scene classification and identification method and system based on meta-learning |
| CN113469913A (en) * | 2021-07-06 | 2021-10-01 | 中南大学 | Hot-rolled strip steel surface water drop removing method based on gradual cycle generation countermeasure network |
| CN113469913B (en) * | 2021-07-06 | 2022-06-03 | 中南大学 | Hot-rolled strip steel surface water drop removing method based on gradual cycle generation countermeasure network |
| CN113610097A (en) * | 2021-08-09 | 2021-11-05 | 西安电子科技大学 | SAR ship target segmentation method based on multi-scale similarity guide network |
| CN113610097B (en) * | 2021-08-09 | 2023-05-05 | 西安电子科技大学 | SAR ship target segmentation method based on multi-scale similarity guidance network |
| CN113962883B (en) * | 2021-09-30 | 2023-02-03 | 西南电子技术研究所（中国电子科技集团公司第十研究所） | Method for enhancing ship target detection SAR image data |
| CN113962883A (en) * | 2021-09-30 | 2022-01-21 | 西南电子技术研究所（中国电子科技集团公司第十研究所） | Method for enhancing ship target detection SAR image data |
| CN113962360B (en) * | 2021-10-09 | 2024-04-05 | 西安交通大学 | Sample data enhancement method and system based on GAN network |
| CN113962360A (en) * | 2021-10-09 | 2022-01-21 | 西安交通大学 | Method and system for enhancing sample data based on GAN network |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112052899A (en) | Single ship target SAR image generation method based on generation countermeasure network | |
| CN111415316B (en) | Defect data synthesis method based on generation countermeasure network | |
| Zou et al. | Ship detection in spaceborne optical image with SVD networks | |
| CN107358262B (en) | High-resolution image classification method and classification device | |
| CN103824088B (en) | SAR target variant recognition method based on multi-information joint dynamic sparse representation | |
| CN112287784B (en) | Radar signal classification method based on deep convolutional neural network and feature fusion | |
| CN110633640A (en) | Method for identifying complex scene by optimizing PointNet | |
| CN109359661B (en) | Sentinel-1 radar image classification method based on convolutional neural network | |
| WO2018000252A1 (en) | Oceanic background modelling and restraining method and system for high-resolution remote sensing oceanic image | |
| CN113255837A (en) | Improved CenterNet network-based target detection method in industrial environment | |
| CN111783819A (en) | Improved target detection method based on region-of-interest training on small-scale data set | |
| CN112784757B (en) | Marine SAR ship target significance detection and identification method | |
| Wu et al. | T-SCNN: A two-stage convolutional neural network for space target recognition | |
| CN114283431B (en) | Text detection method based on differentiable binarization | |
| CN112149526A (en) | Lane line detection method and system based on long-distance information fusion | |
| CN114926691A (en) | Insect pest intelligent identification method and system based on convolutional neural network | |
| CN112784836A (en) | Text and graphic offset angle prediction and correction method thereof | |
| CN113283371A (en) | Local discharge feature extraction and classification method based on BRISK features | |
| CN112329798A (en) | Image scene classification method based on optimized visual bag-of-words model | |
| CN116863293A (en) | Marine target detection method under visible light based on improved YOLOv7 algorithm | |
| CN116563636A (en) | Synthetic aperture radar image generation method and system | |
| CN115331008A (en) | End-to-end target detection method based on target probability density graph | |
| Rusli et al. | Contrast enhanced object recognition (CLAHE_YOLOv3) technique for clear and medium level turbidity lake underwater images | |
| CN113706469A (en) | Iris automatic segmentation method and system based on multi-model voting mechanism | |
| Wang et al. | MSSD-Net: Multi-Scale SAR Ship Detection Network |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20201208 |
|
| RJ01 | Rejection of invention patent application after publication |