CN109712156A - A kind of SAR image edge detection method of low error rate - Google Patents
A kind of SAR image edge detection method of low error rate Download PDFInfo
- Publication number
- CN109712156A CN109712156A CN201811515156.8A CN201811515156A CN109712156A CN 109712156 A CN109712156 A CN 109712156A CN 201811515156 A CN201811515156 A CN 201811515156A CN 109712156 A CN109712156 A CN 109712156A
- Authority
- CN
- China
- Prior art keywords
- edge
- sar image
- pixel
- filter
- rbed
- 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
Landscapes
- Image Analysis (AREA)
Abstract
The invention belongs to SAR image detection technique fields, disclose a kind of SAR image edge detection method of low error rate, it is proposed a kind of effective ratio edge detection operator (Ratio-based edge detector, it is abbreviated as RBED), this method situation high mainly for SAR image edge detection false alarm rate, in conjunction with the anisotropy of image border, continuity and extensibility these three features obtain accurate image border.Firstly, selecting a separable 2D boundary filter;Rotation 2D boundary filter counterclockwise, rotation angle are θ, can get anisotropic filter;Use direction filter calculates the local mean value of certain pixel two sides;Edge strength mapping and edge direction mapping are calculated according to the mean value calculated;Finally, mapping resulting edge strength, jointing edge direction, which is mapping through, can obtain binary edge figure using NSHT algorithm.
Description
Technical field
The invention belongs to SAR image detection technique field more particularly to a kind of SAR image side edge detections of low error rate
Method.
Background technique
Edge-Detection Algorithm is divided into common Optical Image Edge detection algorithm and synthetic aperture radar
(Synthetic Aperture Radar, be abbreviated as SAR) Edge-Detection Algorithm.The imaging band point used by sensor
Class, common optical imagery typically refer to the image data that visible light and part infrared band sensor obtain.And SAR image base
Originally belong to microwave frequency band, wavelength is usually in Centimeter Level.Visible images would generally include the grayscale information of multiple wave bands, in order to
Identify that target and classification are extracted.And SAR image then only has recorded the echo information of a wave band, is recorded with binary complex format
Get off;But the corresponding amplitude of the convertible extraction of complex data and phase information based on each pixel.Amplitude information is usually corresponding
It is closely related with destination media, water content and degree of roughness in ground target to the backscatter intensity of radar wave;The information
The grayscale information obtained with visual light imaging has biggish correlation.And phase information then corresponds to sensor platform and ground appearance
Target roundtrip propagation distance.Common Optical Image Edge detection algorithm mainly have Sobel operator, Roberts operator,
Prewitt operator, LapIacian operator, Canny operator, there are two main classes for SAR image edge detection algorithm, the first kind be with
Average ratio (Ratio of average, be abbreviated as ROA) is representative, directly carries out edge using image space grayscale information
The detective operators of detection.Second class is with Active contour (Active Contour Method, be abbreviated as ACM) for representative,
The detective operators at edge are detected by using image distribution information or using image transform domain information.
It is one most representative in first kind detective operators --- ROA detective operators.ROA detective operators include a 2D
Separable boundary filter and a Boundary extracting algorithm (also referred to as post-processing algorithm) --- non-maximum hysteresis threshold
(Non-Maximum Suppression and Hysteretic Thresholding, be abbreviated as NSHT) algorithm.Second class makes
With the detective operators of image distribution information or image transform domain information, usually by optimize one or more cost functions come into
Row edge detection.Such detective operators include likelihood ratio (Likelihood Ratio, be abbreviated as LR) detective operators, are used
The detective operators of ACM, the detection method etc. based on watershed.
Common edge detection operator is obtained every in image by the way that image and boundary filter (group) are carried out convolution
The gradient value of a pixel.It can determine whether the point is edge pixel according to the gradient magnitude of pixel each in image.When
When edge detection operator based on gradient is used for SAR image edge detection, a large amount of false edge not only will detect that, it is more important
Be that these detective operators have not had CFAR characteristic.This makes them that can not be suitable for SAR image edge detection well
Task.1985, Bovik and David proposed ROA detective operators, and Touzi demonstrates this detective operators with constant false alarm rate
(Constant False Alarm Rate, be abbreviated as CFAR) characteristic.Due to ROA and do not have good edge resolution, because
This RogerEtc. propose exponential weighting average ratio (Ratioof Exponentially Weighted Averages,
It is abbreviated as ROWEA) edge detection operator;In order to obtain the edge of single pixel width, it is double that Shui etc. has also been proposed Gauss-gamma type
Window (Gaussian-Gamma-Shaped Bi-WindoWs, be abbreviated as GGS) edge detection operator.
It was noted that the vertical filter of index class usually has long streaking phenomenon, and such filtering in frequency domain
Device often also side lobe height with higher.There is experiment to show usually have using the edge detection operator of such vertical filter
There is higher false alarm rate.Since the quality of SAR image is much not as good as optical imagery, so it is considered that only rely on single pixel
The edge pixel that infomation detection obtains is unreliable.
Summary of the invention
In view of the above-mentioned problems, being mentioned the purpose of the present invention is to provide a kind of SAR image edge detection method of low error rate
An effective ratio edge detection operator RBED (Ratio-based edge detector, be abbreviated as RBED) out, this method
Mainly for the high situation of SAR image edge detection false alarm rate, in conjunction with the anisotropy of image border, continuity and extensibility
These three features obtain accurate image border, realize the object of the invention.
In order to achieve the above objectives, the present invention is realised by adopting the following technical scheme.
A kind of SAR image edge detection method of low error rate, described method includes following steps:
Step 1, a separable two-dimentional boundary filter is determined;
Step 2, multiple rotation angles are set, successively rotate the two-dimentional edge filter counterclockwise according to each rotation angle
Device obtains corresponding multiple directions filter;
Step 3, according to multiple directions filter, the local mean value of each pixel two sides in SAR image is calculated;
Step 4, according to the local mean value of pixel each in SAR image two sides, determine that the edge of the SAR image is strong
Degree figure and edge orientation map;
Step 5, according to the edge strength figure and edge orientation map of the SAR image, the edge two-value of SAR image is obtained
Figure, and using the edge binary map of the SAR image as the SAR image edge detection results of low error rate.
The characteristics of technical solution of the present invention and further improvement are as follows:
(1) step 1 specifically:
Determine a separable two-dimentional boundary filter fRBED(x, y):
Wherein,For One-dimensional Vertical Gaussian modulation power function filter,For one-dimensional horizontal flat bell filter,
X, y is respectively corresponding filter input variable,For the left son of two-dimentional boundary filter
Window,For the right sub- window of two-dimentional boundary filter.
(2) step 2 specifically:
(2a) sets k-th of rotation angle, θk=k π/P, k=0,1 ..., P-1, P indicate the total number of rotation angle;
(2b) rotates the two-dimentional boundary filter according to k-th of rotation angle counterclockwise, obtains corresponding k-th of direction
Filter
Wherein,For k-th anisotropic filter
Left sub- window,For the right sub- window of k-th of anisotropic filter;
To obtain P anisotropic filter.
(3) step 3 specifically includes:
According to k-th of anisotropic filterCalculate any one in SAR image
Pixel (x0, y0) left side local mean valueWith right side local mean value
Wherein,Indicate that convolution symbol, m, n indicate the variable in convolution algorithm, two-dimentional boundary filter fRBED(x's, y)
The size of sliding window is (2W1+1)×(2W2+ 1), I (x0, y0) it is pixel (x in SAR image0, y0) corresponding pixel value.
(4) step 4 specifically includes: according to any one pixel (x in the SAR image0, y0) two sides local mean value, really
Any one pixel (x in the fixed SAR image0, y0) edge strength ESRBED(x0, y0) and edge direction EDRBED(x0, y0):
ESRBED(x0, y0)=1-min { a (θk)}
Wherein,
To obtain the corresponding edge strength of each pixel in SAR image and edge direction, and respectively obtain SAR image pair
The edge strength figure and edge orientation map answered.
(5) step 5 specifically includes following sub-step:
Any one pixel (x in (5a) edge strength figure corresponding for SAR image1, y1), judge edge strength
Value ESRBED(x1, y1) it whether is along edge direction EDRBED(x1, y1) Local modulus maxima, if it is, retain current pixel
The edge intensity value computing ES of pointRBED(x1, y1), otherwise enable ESRBED(x1, y1)=0;
(5b) sets high threshold ThWith low threshold Tl;And Th> Tl;
(5c) is if ESRBED(x1, y1)≥Th, it is determined that the pixel (x in SAR image1, y1) it is strong edge pixel;
(5d) is if Th> ESRBED(x1, y1) > Tl, it is determined that the pixel (x in SAR image1, y1) it is weak edge pixel
Point;
(5e) is for weak edge pixel point, if it is connected to any one strong side with the relationship of four neighborhoods or eight neighborhood
On edge pixel, then the weak edge pixel point is labeled as strong edge pixel;
(5f) is labeled as 1 in SAR image, by the pixel value of all strong edge pixels, goes out in the SAR image strong
The pixel value of residual pixel point is labeled as 0 after edge pixel point, to obtain the edge binary map of SAR image.
The present invention is relative to the major advantage of existing method: first, the innovation of the invention consists in that SAR image side
In edge detection process, the edge detected is more continuous, and false alarm rate is lower.Second, the present invention uses the separable edge 2D
Filter has preferable filtering characteristic as boundary filter.Third, the present invention also take into account while reducing false alarm rate
Calculation amount, calculation amount are suitable with similar detective operators algorithm.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of invention for those of ordinary skill in the art without creative efforts, can be with
It obtains other drawings based on these drawings.
Fig. 1 is the flow diagram of technical solution of the present invention;
Fig. 2 is image to be detected and its standard picture schematic diagram;
Fig. 3 is to detect 3- view and 6- two width emulating images of view, obtained tested work respectively using RBED, GGS and ROEWA
Person's indicatrix (Receiver-Operating-Characteristiccurves is abbreviated as ROC) curve synoptic diagram;
Fig. 4 is the quality factor of RBED, GGS and ROEWA with the variation schematic diagram of view number;
Fig. 5 is that different detective operators illustrate the comparison of the testing result of true SAR image.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
The embodiment of the present invention provides a kind of SAR image edge detection method of low error rate, as shown in Figure 1, the method
Include the following steps:
Step 1, a separable two-dimentional boundary filter is determined;
Step 1 specifically:
Determine a separable two-dimentional boundary filter fRBED(x, y):
Wherein,For One-dimensional Vertical Gaussian modulation power function filter,For one-dimensional horizontal flat bell filter,
X, y is respectively corresponding filter input variable,For the left son of two-dimentional boundary filter
Window,For the right sub- window of two-dimentional boundary filter.
Specifically, the vertical filter of RBEDFor Gaussian modulation power
Function (power-modulated-Gaussian function, be abbreviated as PMG).Parameter alpha and β codetermine filterPeak position and its rate of decay.Horizontal filterFor flat clock
Shape function (flat bell-shaped function, be abbreviated as FBF).Parameter l is controlledFlat-top length, σ//Then shadow
It ringsThe rate of decay.
Step 2, multiple rotation angles are set, successively rotate the two-dimentional edge filter counterclockwise according to each rotation angle
Device obtains corresponding multiple directions filter;
Step 2 specifically:
(2a) sets k-th of rotation angle, θk=k π/P, k=0,1 ..., P-1, P indicate the total number of rotation angle;
θk[0, π) on uniform sampling, it is contemplated that the calculation amount of edge detection precision and detective operators, usual situation are adopted
Sample interval is set as π/8.Certainly, π/4 or π/16 are also two kinds of common sampling interval values.
(2b) rotates the two-dimentional boundary filter according to k-th of rotation angle counterclockwise, obtains corresponding k-th of direction
Filter
Wherein,For k-th anisotropic filter
Left sub- window,For the right sub- window of k-th of anisotropic filter;
To obtain P anisotropic filter.
Step 3, according to multiple directions filter, the local mean value of each pixel two sides in SAR image is calculated;
Step 3 specifically includes:
According to k-th of anisotropic filterCalculate any one in SAR image
Pixel (x0, y0) left side local mean valueWith right side local mean value
Wherein,Indicate that convolution symbol, m, n indicate the variable in convolution algorithm, two-dimentional boundary filter fRBED(x's, y)
The size of sliding window is (2W1+1)×(2W2+ 1), I (x0, y0) it is pixel (x in SAR image0, y0) corresponding pixel value.
You need to add is that the local mean value of each pixel two sides in SAR image is calculated, when encountering edge pixel
It waits, zero padding filling is carried out to the part beyond boundary, is then calculated according to non-edge pixels.
Step 4, according to the local mean value of pixel each in SAR image two sides, determine that the edge of the SAR image is strong
Degree figure and edge orientation map;
Step 4 specifically includes: according to any one pixel (x in the SAR image0, y0) two sides local mean value, determine
Any one pixel (x in the SAR image0, y0) edge strength ESRBED(x0, y0) and edge direction EDRBED(x0, y0):
ESRBED(x0, y0)=1-min { a (θk)}
Wherein,
To obtain the corresponding edge strength of each pixel in SAR image and edge direction, and respectively obtain SAR image pair
Edge strength figure (Edge Strength Map, be abbreviated as ESM) and edge orientation map (Edge Direction Map, the contracting answered
It is written as EDM).
Step 5, according to the edge strength figure and edge orientation map of the SAR image, the edge two-value of SAR image is obtained
Figure, and using the edge binary map of the SAR image as the SAR image edge detection results of low error rate.
Step 5 specifically includes following sub-step:
Any one pixel (x in (5a) edge strength figure corresponding for SAR image1, y1), judge edge strength
Value ESRBED(x1, y1) it whether is along edge direction EDRBED(x1, y1) Local modulus maxima, if it is, retain current pixel
The edge intensity value computing ES of pointRBED(x1, y1), otherwise enable ESRBED(x1, y1)=0;
(5b) sets high threshold ThWith low threshold Tl;And Th> Tl;
(5c) is if ESRBED(x1, y1)≥Th, it is determined that the pixel (x in SAR image1, y1) it is strong edge pixel;
(5d) is if Th> ESRBED(x1, y1) > Tl, it is determined that the pixel (x in SAR image1, y1) it is weak edge pixel
Point;
(5e) is for weak edge pixel point, if it is connected to any one strong side with the relationship of four neighborhoods or eight neighborhood
On edge pixel, then the weak edge pixel point is labeled as strong edge pixel;
(5f) is labeled as 1 in SAR image, by the pixel value of all strong edge pixels, goes out in the SAR image strong
The pixel value of residual pixel point is labeled as 0 after edge pixel point, to obtain the edge binary map of SAR image.
Further verifying is done to effect of the present invention below by experiment.
(1) experiment content
Experiment one: ROC curve measures detection accuracy.
Detective operators are shown below AndROC
Curve.These three detective operators have used identical post-processing approach --- NSHT.The adjustable parameter α of fixed RBED, parameter
Space is shown below:
Wherein tlowAnd thighFor parameter required by NSHT.The two parameters are common parameters, and value range is fitted simultaneously
For GGS and ROEWA.The adjustable parameter α of fixed GGS, parameter space are shown below:
As for ROEWA, since only there are two adjustable parameters for it, we are not fixed any one parameter.ROEWA
Parameter setting can arbitrarily be chosen from following formula
According to the 3- of emulation view and 6- view, this two width emulating image is detected respectively using RBED, GGS and ROEWA,
Obtained ROC curve.
Experiment two: quality factor measure positioning accuracy.
As shown in Fig. 2, for image to be detected (emulation 3- regards amplitude SAR image) and its standard picture schematic diagram, according to
Image to be detected simulates some amplitude format SAR images with different view numbers.These images regard number variation range as
1- view is regarded to 5-.In this experiment, the parameter of fixed detector ROEWA isThe parameter of GGS be [α, β,
σ]=[3,2,3.5], the parameter of RBED is [α, β, l, σ//]=[3,5,2,3].Its post-process required high-low threshold value then from
It is chosen in following formula
Using Monte Carlo method, 100 independent experiments are repeated to each detector.
Three: RBED detective operators are tested to the Analysis of test results of true SAR image.
The SAR image polarization mode of this experiment is horizontal polarization.The size of image is 512 × 512.For this experiment
The parameter of SAR image, ROEWA is set asGGS is set as [α, β, σ]=[3, Isosorbide-5-Nitrae], and RBED is [α, β, σ//, l]
=[2,2.5,2.4,2].
(2) interpretation of result
Referring to Fig. 3,3- view is detected respectively for RBED, GGS and ROEWA and 6- regards two width emulating images, obtained ROC is bent
Line.In experiment one, work as nTP/nEWhen < 0.95, RBED has minimum false alarm rate.For 6- view, RBED still have compared with
Good detection performance.
It is the quality factor of RBED, GGS and ROEWA with the variation of view number referring to Fig. 4.In experiment two, in all inspections
In measuring and calculating, RBED has highest quality factor always.
It is comparison of the different detective operators to the testing result of true SAR image referring to Fig. 5.In experiment three, RBED institute
The edge detected has preferable continuity and flatness.Also, RBED can detecte out those and be in low contrast regions
Between edge (please see Figure middle oval marks go out region).In addition, RBED also has those edges being located at small structure
Preferable response.
Experiment one illustrate RBED for SAR image 3- depending on or 6- view all there is preferable detection performance, certain
Under the conditions of RBED can have minimum false alarm rate.Experiment two illustrates that RBED has highest quality factor, that is to say, that bright
The positioning performance of RBED detective operators is best.Experiment three illustrates that the edge of RBED detection has preferable continuity and smooth
Property, and also there is preferable effect in the lower region of contrast.
In conclusion emulation experiment demonstrates correctness of the invention, validity and reliability.
Those of ordinary skill in the art will appreciate that: realize that all or part of the steps of above method embodiment can pass through
The relevant hardware of program instruction is completed, and program above-mentioned can store in computer-readable storage medium, which exists
When execution, step including the steps of the foregoing method embodiments is executed;And storage medium above-mentioned includes: ROM, RAM, magnetic or disk
Etc. the various media that can store program code.
The above description is merely a specific embodiment, but scope of protection of the present invention is not limited thereto, any
Those familiar with the art in the technical scope disclosed by the present invention, can easily think of the change or the replacement, and should all contain
Lid is within protection scope of the present invention.Therefore, protection scope of the present invention should be based on the protection scope of the described claims.
Claims (6)
1. a kind of SAR image edge detection method of low error rate, which is characterized in that described method includes following steps:
Step 1, a separable two-dimentional boundary filter is determined;
Step 2, multiple rotation angles are set, successively rotate the two-dimentional boundary filter counterclockwise according to each rotation angle,
Obtain corresponding multiple directions filter;
Step 3, according to multiple directions filter, the local mean value of each pixel two sides in SAR image is calculated;
Step 4, according to the local mean value of pixel each in SAR image two sides, the edge strength figure of the SAR image is determined
And edge orientation map;
Step 5, according to the edge strength figure and edge orientation map of the SAR image, the edge binary map of SAR image is obtained, and
Using the edge binary map of the SAR image as the SAR image edge detection results of low error rate.
2. a kind of SAR image edge detection method of low error rate according to claim 1, which is characterized in that step 1 tool
Body are as follows:
Determine a separable two-dimentional boundary filter fRBED(x, y):
Wherein,For One-dimensional Vertical Gaussian modulation power function filter,For one-dimensional horizontal flat bell filter, x, y
Respectively corresponding filter input variable,For the left sub- window of two-dimentional boundary filter,For the right sub- window of two-dimentional boundary filter.
3. a kind of SAR image edge detection method of low error rate according to claim 1, which is characterized in that step 2 tool
Body are as follows:
(2a) sets k-th of rotation angle, θk=k π/P, k=0,1 ..., P-1, P indicate the total number of rotation angle;
(2b) rotates the two-dimentional boundary filter according to k-th of rotation angle counterclockwise, obtains corresponding k-th of trend pass filtering
Device
Wherein,For the left son of k-th of anisotropic filter
Window,For the right sub- window of k-th of anisotropic filter;
To obtain P anisotropic filter.
4. a kind of SAR image edge detection method of low error rate according to claim 1, which is characterized in that step 3 tool
Body includes:
According to k-th of anisotropic filterCalculate any one picture in SAR image
Element (x0, y0) left side local mean valueWith right side local mean value
Wherein,Indicate that convolution symbol, m, n indicate the variable in convolution algorithm, two-dimentional boundary filter fRBEDThe sliding of (x, y)
The size of window is (2W1+1)×(2W2+ 1), I (x0, y0) it is pixel (x in SAR image0, y0) corresponding pixel value.
5. a kind of SAR image edge detection method of low error rate according to claim 1, which is characterized in that step 4 tool
Body includes: according to any one pixel (x in the SAR image0, y0) two sides local mean value, determine in the SAR image appoint
Anticipate a pixel (x0, y0) edge strength ESRBED(x0, y0) and edge direction EDRBED(x0, y0):
ESRBED(x0, y0)=1-min { a (θk)}
Wherein,
K=0,1 ..., P-1
To obtain the corresponding edge strength of each pixel in SAR image and edge direction, and it is corresponding to respectively obtain SAR image
Edge strength figure and edge orientation map.
6. a kind of SAR image edge detection method of low error rate according to claim 1, which is characterized in that step 5 tool
Body includes following sub-step:
Any one pixel (x in (5a) edge strength figure corresponding for SAR image1, y1), judge edge intensity value computing
ESRBED(x1, y1) it whether is along edge direction EDRBED(x1, y1) Local modulus maxima, if it is, retain current pixel point
Edge intensity value computing ESRBED(x1, y1), otherwise enable ESRBED(x1, y1)=0;
(5b) sets high threshold ThWith low threshold Tl;And Th> Tl;
(5c) is if ESRBED(x1, y1)≥Th, it is determined that the pixel (x in SAR image1, y1) it is strong edge pixel;
(5d) is if Th> ESRBED(x1, y1) > Tl, it is determined that the pixel (x in SAR image1, y1) it is weak edge pixel point;
(5e) is for weak edge pixel point, if it is connected to any one strong edge picture with the relationship of four neighborhoods or eight neighborhood
On vegetarian refreshments, then the weak edge pixel point is labeled as strong edge pixel;
(5f) is labeled as 1 in SAR image, by the pixel value of all strong edge pixels, strong edge of going out in the SAR image
The pixel value of residual pixel point is labeled as 0 after pixel, to obtain the edge binary map of SAR image.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811515156.8A CN109712156A (en) | 2018-12-12 | 2018-12-12 | A kind of SAR image edge detection method of low error rate |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811515156.8A CN109712156A (en) | 2018-12-12 | 2018-12-12 | A kind of SAR image edge detection method of low error rate |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109712156A true CN109712156A (en) | 2019-05-03 |
Family
ID=66255665
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811515156.8A Pending CN109712156A (en) | 2018-12-12 | 2018-12-12 | A kind of SAR image edge detection method of low error rate |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109712156A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110796674A (en) * | 2019-10-15 | 2020-02-14 | 中国人民解放军国防科技大学 | SAR image edge detection method and device combining wide line extraction |
CN112308872A (en) * | 2020-11-09 | 2021-02-02 | 西安工程大学 | Image edge detection method based on multi-scale Gabor first-order derivative |
CN113205540A (en) * | 2021-05-28 | 2021-08-03 | 西安工程大学 | Multi-scale automatic anisotropic morphological direction derivative edge detection method |
-
2018
- 2018-12-12 CN CN201811515156.8A patent/CN109712156A/en active Pending
Non-Patent Citations (1)
Title |
---|
魏倩茹: ""合成孔径雷达图像特征提取的方法研究"", 《中国博士学位论文全文数据库 信息科技辑》 * |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110796674A (en) * | 2019-10-15 | 2020-02-14 | 中国人民解放军国防科技大学 | SAR image edge detection method and device combining wide line extraction |
CN112308872A (en) * | 2020-11-09 | 2021-02-02 | 西安工程大学 | Image edge detection method based on multi-scale Gabor first-order derivative |
CN112308872B (en) * | 2020-11-09 | 2023-06-23 | 西安工程大学 | Image edge detection method based on multi-scale Gabor first derivative |
CN113205540A (en) * | 2021-05-28 | 2021-08-03 | 西安工程大学 | Multi-scale automatic anisotropic morphological direction derivative edge detection method |
CN113205540B (en) * | 2021-05-28 | 2024-05-24 | 西安工程大学 | Multi-scale automatic anisotropic morphological direction derivative edge detection method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Guo et al. | Edge guided reconstruction for compressive imaging | |
CN102509290B (en) | Saliency-based synthetic aperture radar (SAR) image airfield runway edge detection method | |
Hermosilla et al. | Non-linear fourth-order image interpolation for subpixel edge detection and localization | |
US20120328161A1 (en) | Method and multi-scale attention system for spatiotemporal change determination and object detection | |
CN109712156A (en) | A kind of SAR image edge detection method of low error rate | |
CN109492543A (en) | The small target detecting method and system of infrared image | |
CN104820987B (en) | A kind of method based on optical imagery and microwave imagery detection target scattering performance deficiency | |
Cokelaer et al. | Efficient robust d-dimensional path operators | |
CN108765476A (en) | Polarized image registration method | |
Lee et al. | Edge detection analysis | |
Nair et al. | Fuzzy logic-based automatic contrast enhancement of satellite images of ocean | |
CN106022354B (en) | Image MTF measurement methods based on SVM | |
CN110223312A (en) | A kind of SAR image edge detection method based on shearing wave | |
CN110148149A (en) | Underwater vehicle thermal trail segmentation method based on local contrast accumulation | |
Li et al. | Bionic vision-based synthetic aperture radar image edge detection method in non-subsampled contourlet transform domain | |
CN107369163A (en) | A kind of quick SAR image object detection method based on best entropy Double Thresholding Segmentation | |
CN106910178A (en) | A kind of multi-angle SAR image fusion method based on hue statistical property sort | |
Zhu et al. | A novel change detection method based on high-resolution SAR images for river course | |
CN109142530A (en) | Defects in Pressure Container guided wave localization method based on information fusion and elliptic imaging | |
CN115100446A (en) | Similarity measurement method for matching SAR and visible light remote sensing image | |
CN106897985A (en) | A kind of multi-angle SAR image fusion method based on visibility classification | |
Law et al. | Segmentation of vessels using weighted local variances and an active contour model | |
Colin et al. | Segmentation of rainfall regimes by machine learning on a colocalized Nexrad/Sentinel-1 dataset | |
Chen et al. | Level sets with self-guided filtering for marine oil spill segmentation | |
CN109712155A (en) | A kind of unbiased SAR edge detection method of difference-proportional-type |
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 | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190503 |
|
WD01 | Invention patent application deemed withdrawn after publication |