CN116844059B - Polarization SAR image target detection method based on double diagonal change - Google Patents
Polarization SAR image target detection method based on double diagonal change Download PDFInfo
- Publication number
- CN116844059B CN116844059B CN202311101949.6A CN202311101949A CN116844059B CN 116844059 B CN116844059 B CN 116844059B CN 202311101949 A CN202311101949 A CN 202311101949A CN 116844059 B CN116844059 B CN 116844059B
- Authority
- CN
- China
- Prior art keywords
- matrix
- polarization
- diagonal
- target detection
- detection method
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 230000010287 polarization Effects 0.000 title claims abstract description 72
- 238000001514 detection method Methods 0.000 title claims abstract description 49
- 230000008859 change Effects 0.000 title claims description 7
- 239000011159 matrix material Substances 0.000 claims abstract description 82
- 239000013598 vector Substances 0.000 claims abstract description 30
- 238000000034 method Methods 0.000 claims abstract description 10
- 238000012549 training Methods 0.000 claims abstract description 9
- 238000001914 filtration Methods 0.000 claims abstract description 5
- 238000012545 processing Methods 0.000 claims abstract description 5
- 230000009977 dual effect Effects 0.000 claims description 10
- 238000004590 computer program Methods 0.000 claims description 7
- 238000013135 deep learning Methods 0.000 abstract description 2
- 238000012797 qualification Methods 0.000 abstract description 2
- 230000009466 transformation Effects 0.000 description 5
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000000342 Monte Carlo simulation Methods 0.000 description 1
- 229920000291 Poly(9,9-dioctylfluorene) Polymers 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 239000003305 oil spill Substances 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
- 230000002087 whitening effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/36—Applying a local operator, i.e. means to operate on image points situated in the vicinity of a given point; Non-linear local filtering operations, e.g. median filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Nonlinear Science (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
The invention provides a polarized SAR image target detection method based on double diagonal variation, which converts a polarized covariance matrix detected by a polarization detector for each pixel point into a polarized diagonal matrix; extracting main diagonal elements of the polarization diagonal matrix; inputting the main diagonal elements as training samples into a pocket perceptron algorithm, and outputting weight vectors; and processing the PolSAR image to be detected by taking the weight vector as a filtering matrix to obtain a target detection result. The method can be widely applied to any PolSAR image with priori information, breaks away from the constraint of a polarization covariance matrix, has no requirement on the forward qualification of a filter matrix, and lays a certain foundation for the application of interpretative deep learning in the aspect of PolSAR image ship target detection.
Description
Technical Field
The invention belongs to the field of microwave remote sensing image processing, and particularly relates to a polarized SAR image target detection method based on double diagonal change.
Background
The polarization synthetic aperture radar PolSAR reserves more complete polarization information on the basis of wide application and various advantages of the traditional synthetic aperture radar SAR, and can detect, classify and identify targets more comprehensively. The PolSAR marine remote sensing has great value in aspects of sea surface oil spill detection, ship target detection, marine floating object identification, iceberg detection, polar region monitoring, early warning detection of sensitive areas and the like, and is deployed on various platforms. There are many polarization detectors that have their own theoretical support and physical connotation.
The optimal polarization detector OPD judges whether the pixel to be detected belongs to a target or clutter through likelihood ratio detection, a polarization whitening filter PWF minimizes clutter fluctuation, a polarization notch filter PNF minimizes clutter energy, a polarization matching filter PMF maximizes signal to noise ratio, a polarization detection optimizing filter PDOF combines PWF and PMF, the signal to noise ratio is maximized, the clutter fluctuation is minimized at the same time, and a minimum clutter signal to noise ratio subspace detector MCSR searches the largest signal to noise ratio in subspace for detecting the target.
The final expression forms of the polarization detectors are all filter matrixes obtained after various priori information on the PolSAR image is optimized, and the detection result graph is a result of tracing after the filter matrixes and the polarization covariance matrix are multiplied. Therefore, the key to detector optimization is the formation of a filter matrix, and the above polarization detectors based on the optimization technique are essentially linear combinations of real and imaginary elements of a polarization covariance matrix, which is a unified framework of the polarization detector. Along with the influence of radar platform parameter diversity and sea state environment complexity, complex sea clutter statistical modeling and parameter estimation, slow small target detection, dense target detection and the like are still difficult problems of current PolSAR image ship target detection, and when the dimension of a polarization covariance matrix is n, the degree of freedom is n 2 Due to the fact that the degree of freedom is increased, the memory occupation is too high, and detection accuracy under the conditions of different sea conditions, different platforms and different observation angles is difficult to ensure.
Disclosure of Invention
The invention provides a polarized SAR image target detection method based on double diagonal changes, which aims to solve the technical problem that the detection accuracy is difficult to guarantee under the conditions of different sea conditions, different platforms and different observation angles in the existing detection method.
In order to solve the technical problems, the invention provides a polarized SAR image target detection method based on double diagonal variation, which comprises the following steps:
step S1: converting a polarization covariance matrix detected by a polarization detector for each pixel point into a polarization diagonal matrix;
step S2: extracting main diagonal elements of the polarization diagonal matrix;
step S3: inputting the main diagonal elements as training samples into a pocket perceptron algorithm, and outputting weight vectors;
step S4: and processing the PolSAR image to be detected by taking the weight vector as a filtering matrix to obtain a target detection result.
Preferably, the expression of the polarization diagonal matrix D is:
；
wherein C represents a polarization covariance matrix and U represents a matrixA feature matrix composed of feature vectors of (a) is provided.
Preferably, the conversion of the polarization covariance matrix into the polarization diagonal matrix comprises the following steps:
step S11: converting the detection result of the polarization detector into a trace of the product of the filter matrix and the polarization covariance matrix;
step S12: expanding the product of the filtering matrix and the polarization covariance matrix, and converting the product into linear combination of the real part and the imaginary part of the polarization covariance matrix;
step S13: and performing hermite property change on the filter matrix to obtain a new filter matrix so as to convert the product of the original filter matrix and the polarization covariance matrix into the product of the new filter matrix and the polarization diagonal matrix.
Preferably, the expression of step S12 is expressed as:
；
where tr () represents the trace, P represents the filter matrix of the polarization detector, C represents the polarization covariance matrix, superscript x represents the conjugate, re () represents the real part of the data, R represents the real part of the corresponding element, and I represents the imaginary part of the corresponding element.
Preferably, the expression of step S13 is:
；
wherein Q represents a new filter matrix subjected to the change of hermite properties by the filter matrix P, D represents a polarization diagonal matrix, and U representsA feature matrix composed of feature vectors of (a) is provided.
Preferably, the pocket perceptron algorithm in step S3 includes the steps of:
step S31: setting training samplesAnd tag value->Wherein->，x i1 、x i2 And x i3 Representing the principal diagonal element values of the polarization diagonal matrix;
step S32: randomly selecting samples (x) 1 ,y 1 ) Obtaining whether the pixel belongs to clutter or a target through the weight vector and the main diagonal element;
step S33: updating the weight vector, and selecting the weight vector with the minimum clutter as an optimal weight vector;
step S34: and repeating the steps S32 to S33 until no clutter exists or the maximum iteration number is reached, and obtaining the optimal weight vector.
Preferably, in step S32, the expression for determining whether the pixel belongs to the clutter or the target is obtained by using the weight vector and the main diagonal element:
；
wherein w represents a weightVector b represents the bias parameter whenIndicating that the target pixel belongs to clutter, otherwise belongs to the target.
Preferably, the expression for updating the weight in step S33 is:
；
wherein y is i Representing the tag value.
The invention also provides an electronic device, comprising: a memory, a processor and a computer program stored in the memory and configured to be executed by the processor to implement one of the dual diagonal variation based polarized SAR image target detection methods described above.
The present invention also provides a computer readable storage medium having stored therein a computer program for execution by a processor to implement a dual diagonal variation-based polarized SAR image target detection method as described above.
The beneficial effects of the invention at least comprise: the method for detecting the PolSAR image target based on the double diagonal transformation is provided, and the transformed matrix main diagonal elements are extracted to be used as a new characterization mode through the double diagonal transformation, so that the degree of freedom of a model can be effectively reduced, and the memory is saved.
The new features can filter and detect pixels in a variety of ways, data driven, model driven. The sparse Lu Bang characteristic after the double diagonal transformation ensures the accuracy of the detection process under the conditions of different sea conditions, different platforms and different observation angles.
The method can be widely applied to any PolSAR image with priori information, breaks away from the constraint of a polarization covariance matrix, has no requirement on the forward qualification of a filter matrix, and lays a certain foundation for the application of interpretative deep learning in the aspect of PolSAR image ship target detection.
Drawings
FIG. 1 is a schematic flow chart of a method according to an embodiment of the invention;
FIG. 2 is a schematic diagram of polarization diagonal matrix transformation according to an embodiment of the present invention;
FIG. 3 is a simulated ROC curve of Monte Carlo according to an embodiment of the invention;
FIG. 4 is a high-resolution image of a region III according to an embodiment of the present invention;
FIG. 5 is a schematic diagram of a visual result of ship detection according to an embodiment of the present invention;
fig. 6 is a schematic view of a ROC curve of an actual measurement image according to an embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is evident that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by a person skilled in the art without any inventive effort, are intended to be within the scope of the present invention, based on the embodiments of the present invention.
As shown in fig. 1, the embodiment of the invention provides a polarized SAR image target detection method based on dual diagonal variation, which comprises the following steps:
step S1: the polarization covariance matrix detected by the polarization detector for each pixel is converted into a polarization diagonal matrix.
In particular, for conventional polarization detectors, they can be regarded as a transformation of the three-dimensional polarization vector, and the final detection map is typically a power map. The detection result of the pixel position of any PolSAR image can be expressed as:
；
where z represents the detection result of the pixel point, L represents the multiview number, s represents the polarization scattering vector, P represents the filter matrix of the polarization detector, and C represents the polarization covariance matrix, and it can be seen from the above equation that the detection result of the polarization detector can be represented as a trace of the product of the filter matrix and the polarization covariance matrix, so that the difference in value between different detectors is different from the matrix. The expression can therefore be multiplied and the expression analyzed in depth:
；
where tr () represents the trace, P represents the filter matrix of the polarization detector, C represents the polarization covariance matrix, superscript x represents the conjugate, re () represents the real part of the data, R represents the real part of the corresponding element, and I represents the imaginary part of the corresponding element. The polarization detector can be found to be essentially a linear combination of the real and imaginary values of the polarization covariance matrix. In this case, for a matrix of dimension n, the degree of freedom is n 2 I.e. the input is n 2 Dimension.
In the embodiment of the invention, if the filter matrix is converted into Q through hermite property conversion, z can be expressed as:
；
wherein U represents a matrixA feature matrix composed of feature vectors of (a) is provided.
Thus, the expression for the polarization diagonal matrix can be found as:
；
then for pure clutter, there is:
；
for pure targets there are:
。
step S2: and extracting the main diagonal elements of the polarization diagonal matrix.
Specifically, the principal diagonal element of the polarization diagonal matrix D is extracted as a new polarization feature as shown in FIG. 2, where the dimension of the detection model is defined by the previous n 2 The dimension is reduced to n dimensions.
Step S3: and taking the main diagonal elements as training samples to input a pocket perceptron algorithm, and outputting weight vectors.
Specifically, input training samplesWherein->Corresponding to the main diagonal elements of matrix D. Tag value->The maximum iteration number is n max The method comprises the steps of carrying out a first treatment on the surface of the Output of optimal weight vectorBias parameter b.
(1) Initializing parameters w and b, setting the iteration number to k=0, and setting a "pocket" such as best w ，best b Storing the optimal parameter value, and firstly storing the initialized parameter in a pocket;
(2) Optionally a sample in the training setCalculating a weight vector and a main diagonal element to obtain whether the pixel belongs to clutter or target, wherein the expression is as follows:
；
the time is clutter, otherwise is the goal;
(3) Updating weights，/>Comparing the updated parameter values w and b with the parameter value best in the pocket w ，best b The error points caused by the two parameters are respectively, and the corresponding parameter value with the minimum error point is put into a pocket;
(4) Returning to the step (2) until the training set has no wrong classification point or the maximum iteration number n is reached max Finally, returning the parameter value in the pocket to be the required value.
Step S4: and processing the PolSAR image to be detected by taking the weight vector as a filtering matrix to obtain a target detection result.
Examples are as follows:
simulation is carried out aiming at the method, 10000 clutter covariance matrixes and target covariance matrixes are generated, and the signal to noise ratio is set to be 0.5. Through Monte Carlo simulation verification of a PolSAR ship target detection theory based on a perceptron, an ROC curve is obtained, and is shown in a figure 3, so that the detection potential of the current polarization detector can be maximized by the method provided by the invention.
The PolSAR image from the high-resolution third-order synthetic aperture radar is processed by using a perceptron method, as shown in figure 4. The image is a PolSAR image of a certain area, and the radar works in a C wave band. Scene number 3180124, incidence angle 42 °. Wind speed is 17m/s through radar data inversion. The rectangular box represents a large target, the circle represents a small target which is difficult to detect, the detection result is shown in fig. 5, all ships can be successfully detected, and x in the figure represents false alarms which are automatically filtered.
The ROC curves of the results of the different detector outputs are compared as shown in fig. 6. It can be found that the perceptron-based polarization detector performs best because the pocket perceptron algorithm PPLA combines model driving with data driving to form an optimal detector that is best suited to the present scenario.
The invention also provides an electronic device, comprising: a memory, a processor and a computer program stored in the memory and configured to be executed by the processor to implement one of the dual diagonal variation based polarized SAR image target detection methods described above.
The present invention also provides a computer readable storage medium having stored therein a computer program for execution by a processor to implement a dual diagonal variation-based polarized SAR image target detection method as described above.
The foregoing embodiments may be combined in any way, and all possible combinations of the features of the foregoing embodiments are not described for brevity, but only the preferred embodiments of the invention are described in detail, which should not be construed as limiting the scope of the invention. The scope of the present specification should be considered as long as there is no contradiction between the combinations of these technical features.
It should be noted that it will be apparent to those skilled in the art that several variations and modifications can be made without departing from the spirit of the invention, which are all within the scope of the invention. Accordingly, the scope of protection of the present invention is to be determined by the appended claims.
Claims (7)
1. A polarized SAR image target detection method based on double diagonal variation is characterized in that: the method comprises the following steps:
step S1: converting the polarization covariance matrix detected by the polarization detector for each pixel point into a polarization diagonal matrix, comprising the following steps:
step S11: converting the detection result of the polarization detector into a trace of the product of the filter matrix and the polarization covariance matrix;
step S12: performing hermite property change on the filter matrix to obtain a new filter matrix so as to convert the product of the original filter matrix and the polarization covariance matrix into the product of the new filter matrix and the polarization diagonal matrix;
the expression of the polarization diagonal matrix D is as follows:
D＝U * Σ C -1/2 CΣ C -1/2 U；
wherein C represents the polarization covariance matrix and U represents the matrix Σ C -1/2 Σ T Σ C -1/2 A feature matrix composed of feature vectors of (a);
step S2: extracting main diagonal elements of the polarization diagonal matrix;
step S3: inputting the main diagonal elements as training samples into a pocket perceptron algorithm, and outputting weight vectors;
step S4: and processing the PolSAR image to be detected by taking the weight vector as a filtering matrix to obtain a target detection result.
2. The polarized SAR image target detection method based on dual diagonal variation according to claim 1, wherein: the expression of step S12 is:
wherein Q represents a new filter matrix subjected to a change in hermite properties by the filter matrix P, D represents a polarization diagonal matrix, and U represents Σ C -1/2 Σ T Σ C -1/2 A feature matrix composed of feature vectors of (a) is provided.
3. The polarized SAR image target detection method based on dual diagonal variation according to claim 1, wherein: the pocket perceptron algorithm in step S3 comprises the following steps:
step S31: set training sample x= (X) 1 ,x 2 ......x n ) And a tag value (y 1 ,y 2 ,y 3 ......y n ),y i E {1, -1}, where x i ＝(x i1 ,x i2 ,x i3 )，x i1 、x i2 And x i3 Representing the principal diagonal element values of the polarization diagonal matrix;
step S32: randomly selecting samples (x) 1 ,y 1 ) Obtaining whether the pixel belongs to clutter or a target through the weight vector and the main diagonal element;
step S33: updating the weight vector, and selecting the weight vector with the minimum clutter as an optimal weight vector;
step S34: and repeating the steps S32 to S33 until no clutter exists or the maximum iteration number is reached, and obtaining the optimal weight vector.
4. A polarized SAR image target detection method according to claim 3, wherein: in step S32, the expression for determining whether the pixel belongs to clutter or target is obtained by using the weight vector and the main diagonal element:
where w represents the weight vector and b represents the bias parameter, whenIndicating that the target pixel belongs to clutter, otherwise belongs to the target.
5. The polarized SAR image target detection method based on dual diagonal variation as set forth in claim 4, wherein: the expression for updating the weights in step S33 is:
w＝w+y i ·x i ；
wherein y is i Representing the tag value.
6. An electronic device, comprising: memory, processor and computer program characterized by: the computer program is stored in the memory and configured to be executed by the processor to implement a dual diagonal variation based polarized SAR image target detection method according to any one of claims 1 to 5.
7. A computer-readable storage medium, characterized by: the computer-readable storage medium has stored therein a computer program that is executed by a processor to realize a polarization SAR image target detection method based on a dual diagonal variation according to any one of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311101949.6A CN116844059B (en) | 2023-08-30 | 2023-08-30 | Polarization SAR image target detection method based on double diagonal change |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311101949.6A CN116844059B (en) | 2023-08-30 | 2023-08-30 | Polarization SAR image target detection method based on double diagonal change |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116844059A CN116844059A (en) | 2023-10-03 |
| CN116844059B true CN116844059B (en) | 2023-11-21 |
Family
ID=88167444
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311101949.6A Active CN116844059B (en) | 2023-08-30 | 2023-08-30 | Polarization SAR image target detection method based on double diagonal change |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116844059B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2011070538A2 (en) * | 2009-12-11 | 2011-06-16 | Eads Singapore Pte. Ltd | Method for despeckling of single-look dual-polarization synthetic aperture radar (sar) data |
| CN113030954A (en) * | 2021-04-20 | 2021-06-25 | 吉林大学 | Ground penetrating radar data SVD distributed algorithm based on Flink |
| CN116433530A (en) * | 2023-04-19 | 2023-07-14 | 中国人民解放军国防科技大学 | Average value movement-based polarized SAR (synthetic aperture radar) speckle filtering method |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10088555B2 (en) * | 2014-12-15 | 2018-10-02 | Airbus Singapore Private Limited | Automated method for selecting training areas of sea clutter and detecting ship targets in polarimetric synthetic aperture radar imagery |
-
2023
- 2023-08-30 CN CN202311101949.6A patent/CN116844059B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2011070538A2 (en) * | 2009-12-11 | 2011-06-16 | Eads Singapore Pte. Ltd | Method for despeckling of single-look dual-polarization synthetic aperture radar (sar) data |
| CN113030954A (en) * | 2021-04-20 | 2021-06-25 | 吉林大学 | Ground penetrating radar data SVD distributed algorithm based on Flink |
| CN116433530A (en) * | 2023-04-19 | 2023-07-14 | 中国人民解放军国防科技大学 | Average value movement-based polarized SAR (synthetic aperture radar) speckle filtering method |
Non-Patent Citations (3)
| Title |
|---|
| A General Framework of Polarimetric Detectors Based on Quadratic Optimization;Tao Liu 等;IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING;第60卷;5237418 * |
| Joint Polarimetric Subspace Detector Based on Modified Linear Discriminant Analysis;Tao Liu 等;IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING;5223519 * |
| 雷达极化对角加载检测器的最优权重算法;曹运运 等;雷达科学与技术;222-230 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116844059A (en) | 2023-10-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112395987B (en) | SAR image target detection method based on unsupervised domain adaptive CNN | |
| Mukherjee et al. | Symbolic analysis of sonar data for underwater target detection | |
| CN113534120B (en) | Multi-target constant false alarm rate detection method based on deep neural network | |
| CN111046800B (en) | Hyperspectral image abnormal target detection method based on low rank and sparse decomposition | |
| CN103810704A (en) | SAR (synthetic aperture radar) image change detection method based on support vector machine and discriminative random field | |
| CN113378988B (en) | Particle swarm algorithm-based robustness enhancement method and device for deep learning system | |
| Song et al. | Ship detection in SAR imagery via variational Bayesian inference | |
| Zhang et al. | Infrared small target detection via interpatch correlation enhancement and joint local visual saliency prior | |
| Lv et al. | Novel automatic approach for land cover change detection by using VHR remote sensing images | |
| CN111695461A (en) | Radar clutter intelligent classification method based on image frequency characteristics | |
| Smitha et al. | Target detection using supervised machine learning algorithms for GPR data | |
| CN113255603B (en) | Enhancement matrix constant false alarm rate detection method based on Riemann manifold supervision dimension reduction | |
| Yuksel et al. | Hierarchical methods for landmine detection with wideband electro-magnetic induction and ground penetrating radar multi-sensor systems | |
| Huang et al. | Change detection method based on fractal model and wavelet transform for multitemporal SAR images | |
| CN116844059B (en) | Polarization SAR image target detection method based on double diagonal change | |
| Du et al. | An improved prototype network and data augmentation algorithm for few-shot structural health monitoring using guided waves | |
| Wen et al. | Modeling of correlated complex sea clutter using unsupervised phase retrieval | |
| Picard et al. | Seafloor description in sonar images using the monogenic signal and the intrinsic dimensionality | |
| Pelliza et al. | Optimal Canny’s parameters regressions for coastal line detection in satellite-based SAR images | |
| Lin et al. | Optimization of a multi-stage ATR system for small target identification | |
| Li et al. | A small leakage detection approach for gas pipelines based on CNN | |
| Hu et al. | Defects identification and location of underground space for ground penetrating radar based on deep learning | |
| Fan et al. | A deceptive jamming template synthesis method for SAR using generative adversarial nets | |
| Tantum et al. | Frequency domain electromagnetic induction sensor data feature extraction and processing for improved landmine detection | |
| Yu et al. | A lightweight ship detection method in optical remote sensing image under cloud interference |
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 |