CN107679453A - Weather radar electromagnetic interference echo recognition methods based on SVMs - Google Patents
Weather radar electromagnetic interference echo recognition methods based on SVMs Download PDFInfo
- Publication number
- CN107679453A CN107679453A CN201710750362.6A CN201710750362A CN107679453A CN 107679453 A CN107679453 A CN 107679453A CN 201710750362 A CN201710750362 A CN 201710750362A CN 107679453 A CN107679453 A CN 107679453A
- Authority
- CN
- China
- Prior art keywords
- electromagnetic interference
- interference echo
- image
- echo
- sample
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
-
- 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
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Signal Processing (AREA)
- Image Analysis (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
The present invention relates to Image Information Processing technology, makes full use of the different characteristic of three class electromagnetic interference echoes and the superiority of SVM methods to realize, can identify three class electromagnetic interference echoes simultaneously, improves the accuracy rate of electromagnetic interference echo identification.The present invention, the weather radar electromagnetic interference echo recognition methods based on SVMs, step are as follows：Step 1, training sample set step 2 is established, positive negative sample is loaded into and distributes label；Step 3, the positive and negative sample characteristics step 4 of extraction, training SVM classifier, and generate the SVM classifier of corresponding training sample data；Step 5, preserve grader step 6, the accuracy step 7 of detection grader classification, be loaded into test image；Step 8, extract test image characterization step 9, test chart is classified using the SVM classifier trained；The identification of step 10, realization to spiral electromagnetic interference echo and pit shape electromagnetic interference echo.Present invention is mainly applied to Image Information Processing occasion.
Description
Technical field
The present invention relates to technical field of image information processing, specifically, is related to the weather radar electricity based on SVMs
Magnetic disturbance echo recognition methods.
Background technology
Weather Radar is divided into weather echo and non-meteorological echo, rather than weather echo can be made to the identification of weather echo
Into interference.Non-meteorological echo is broadly divided into background return, biological echo, electromagnetic interference echo and radar fault angel.Its
In, electromagnetic interference echo refer to radar return by extraneous with frequency or the interference echo phenomenon faced the interference of frequency equipment and occurred, this
A little interference echos can be divided into three classes, i.e. diametral interference echo, helical form interference echo and pit shape interference echo.Closely
Electromagnetic interference, interference pit all over the screen can be produced on radar screen；And remote electromagnetic interference, then can exist in a certain direction
Fixed interference.If single-frequency point electromagnetic interference, interference is linearly；And for the interference with certain bandwidth, then can be in thunder
Interference pit all over the screen is produced on up to screen.
Electromagnetic interference echo is that its appearance can reduce radar an important factor for influenceing CINRAD Product quality
The product quality of generation, the mainly pollution to reflectance data, this brings very big puzzlement to the application of Radar Data.Though
Right radar controller or forecaster can taking human as observe electromagnetic interference echo and provide mark, but if without one kind
Automatic recognition methods, very big influence can be caused for the precipitation estimation in real time business and radar rotating platform.Therefore,
It is desirable to propose the method for eliminating electromagnetic interference in terms of hardware and software, to improve base data and other afterproducts
Quality.Not yet realized at present by the method for hardware to solve this problem, it is different mainly by designing in software aspects
Algorithm electromagnetic interference echo is identified, current existing recognizer is seldom, mainly there is an isolated echo cancellation algorithms
With diametral interference recognizer.Isolated echo algorithm is filtered processing for the interference echo with isolatism, first, is seeing
A mobile window is established on mapping, if the point of the virtual value in the window around central point is less than a certain threshold value,
The central point is removed, that is, regards electromagnetic interference echo of the point as isolatism.Therefore this method is adapted to identify or remove spiral shell
Revolve shape and pit electromagnetic interference echo.However, due to interference source occur distance is indefinite, orientation is random, pulsewidth and work repetition
Difference, the form of helical form interference echo are not quite similar, i.e., diametrically different with the continuity in orientation, so isolated echo
Elimination algorithm can not filter out the spiral helicine electromagnetic interference echo of different shape completely, equally, for pit shape interference echo
For there is also this problem；Diametral interference recognition methods is established according to electromagnetic interference echo form.First, to it is same away from
Wave band is divided back in tangential direction from upper range bin.Then, each radially echo strength effectively total range bin number is counted
RTWidth with the echo section where range bin is the distance between 0~5 storehouse number RN, calculate RD(RD=RN/RT× 100%), if
RDIt is radially electromagnetic interference echo that >=30%, which judges this,.This method is only suitable for for identifying radial direction electromagnetic interference echo, and
The narrow electromagnetic interference echo less than 5 radial directions can only be identified, radial direction electromagnetic interference echo this method beyond this scope can not be entered
Row identification.Therefore, current electromagnetic interference echo recognition methods is required to be improved.
Image classification can be widely applied in every field at present as a kind of method of information extraction.Figure
As the emphasis of sort research is concentrated mainly on the selection of characteristics of image with extraction and the selection of disaggregated model in terms of the two.HOG
(Histogram of Oriented Gradient) is histograms of oriented gradients, be one kind in computer vision and image procossing
In be used for carrying out the feature descriptor of object detection.HOG features can keep good to image geometry and optical deformation
Consistency, there is good robustness to intensity of illumination, using extensive in detecting out of doors.SVMs (SVM) is to establish
A kind of data digging method on the basis of Statistical Learning Theory, it has been proved to for classifying, returning and pattern-recognition etc. is asked
The treatment effect of topic is ideal.SVM is a kind of feedforward neural network in essence, according to empirical risk minimization,
On the premise of training sample error in classification minimization is made, the extensive Generalization Ability of grader is improved as far as possible.Its general principle is
Find an optimal separating hyper plane for meeting classificating requirement so that the hyperplane makes super flat while nicety of grading is ensured
The class interval of face both sides is maximum.
The content of the invention
For overcome the deficiencies in the prior art, the present invention is directed to propose a kind of weather radar electromagnetic interference echo recognition methods.
The different characteristic of three class electromagnetic interference echoes and the superiority of SVM methods are made full use of, can identify that three class electromagnetic interferences are returned simultaneously
Ripple, improve the accuracy rate of electromagnetic interference echo identification.Therefore, the technical solution adopted by the present invention is, based on SVMs
Weather radar electromagnetic interference echo recognition methods, step are as follows：
Step 1, establish training sample set
Establish positive sample to integrate as radial direction electromagnetic interference echo, negative sample integrates as helical form electromagnetic interference echo and pit
Shape electromagnetic interference echo；
Step 2, it is loaded into positive negative sample and distributes label；
Step 3, the positive and negative sample characteristics of extraction
Positive sample image is read in, coloured image is converted into gray level image, then extracts its HOG feature；Read in negative sample
Image, coloured image is converted into gray level image, then extracts its HOG feature；
Step 4, training SVM classifier, and generate the SVM classifier of corresponding training sample data；
Step 5, preserve grader
Step 6, the accuracy of detection grader classification
HOG features corresponding to positive and negative sample set are mixed, are randomly divided into two groups, one group is new training set, and one group is to test
Card collection, Autonomous test, the accuracy rate of testing classification device classification are carried out using checking set pair grader classification results；
Step 7, it is loaded into test image；
Step 8, extraction test image feature
Test image is read in, coloured image is converted into gray level image, then extracts its HOG feature；
Step 9, using the SVM classifier trained test chart is classified, obtain radial direction electromagnetic interference and spiral shell
The classification results of electromagnetic interference and pit shape electromagnetic interference echo are revolved, realize the identification to radial direction electromagnetic interference echo；Such as test
As a result be radial direction electromagnetic interference echo, then EP (end of program)；If test result is non-radial electromagnetic interference echo, then carry out next
Step；
Step 10, establish positive sample and integrate as helical form electromagnetic interference echo, negative sample integrates as pit shape electromagnetic interference
Echo, repeat step 2 arrive step 9, obtain spiral electromagnetic interference and the classification results of pit shape electromagnetic interference echo, realize
Identification to spiral electromagnetic interference echo and pit shape electromagnetic interference echo.
Further technical scheme is in the step 3 and step 8, and the specific method of HOG feature extractions is：First,
Calculate the gradient magnitude and gradient direction of each pixel in sample image；Then, some UNICOM regions are divided the image into, these
Connected region is called cell factory, then cell factory is divided into several blocks, and several passages are equally divided into by 0 °~180 °；
Secondly, the direction histogram of the gradient of each pixel in cell factory is gathered；Finally, it is the gradient of the pixel of different passages is big
It is small cumulative, one group is obtained by the vector that each passage pixel gradient is cumulative and forms, then in units of block, normalizing is carried out to vector
Change is handled, and the vector after normalized is chained up, and forms HOG sample characteristics descriptors；
It is to use two discrete differential templates (- 1,0,1) and (- 1,0,1) to calculate gradientT, T represents transposition, respectively to figure
As horizontal direction and vertical direction are handled, the gradient of each pixel is obtained.
In step 4, obtaining SVM classifier detailed process is：The mistake of optimal svm classifier function is asked for according to known sample collection
Journey can solve classification problem, be in some higher dimensional space first nonlinear problem shift conversion for nonlinear problem
Linear problem, optimal classification surface then is solved in space after transformation, if given sample set is s={ (xi,yi) | i=1,
2 ... n }, xi∈RdFor the vector in d dimension spaces, yi={+1, -1 } is xiCorresponding category label, i are sample set numberings, it is assumed that
Nonlinear Mapping isWhereinFor mapping function, support vector cassification face is write as：w
For weight vector, b is constant offset,For class interval, the optimization problem that class interval obtains maximum is represented by askingMinimum problems, constraints isAccording to KKT (Karush-Kuhn-
Tucker) condition and Lagrangian obtain classification function and are：
Wherein, sgn () is sign function, aiFor optimal Lagrange multiplier, x is sample to be tested,Point multiplication operation (the x of higher dimensional spaceiX) by the kernel function k (x of lower dimensional spacei, xj) replace, classification
Function is rewritable to be
Conventional kernel function is as follows：
Polynomial kernel function k (x, xi)=[(xxi)+1]qQ=1,2,3 ...；
Gaussian radial basis functionσ is parameter；
Sigmoid function k (x, xi)=tanh (b (x, xi)+c) b, c be constant；
Fourier spaceWherein N is constant；
B-spline kernel function k (x, xi)=B2p+1(x-xi) wherein B2p+1(x) it is 2p+1 ranks B；
Different svm classifier functions can be produced by choosing different kernel functions, so as to obtain different SVM classifiers, therefore
Different svm classifier methods can be expanded to by the selection of kernel function.
The features of the present invention and beneficial effect are：
Compared with prior art, the inventive method have the advantage that for：The present invention can be to three class weather radar electromagnetism
Interference echo is identified, and algorithm principle is simple, has real-time, while self detection can be carried out to the accuracy rate of classification,
Ensure the accuracy of electromagnetic interference echo identification.
Brief description of the drawings：
Fig. 1 the inventive method basic flow sheets.
Fig. 2 embodiment classification results.
Embodiment
The weather radar electromagnetic interference echo recognition methods of the present invention specifically includes following steps：
Step 1, establish training sample set.
Establish positive sample and integrate (positive sample) as radial direction electromagnetic interference echo, negative sample collection (negative
Sample) it is helical form electromagnetic interference echo and pit shape electromagnetic interference echo.
Step 2, it is loaded into positive negative sample and distributes label.
Step 3, the positive and negative sample characteristics of extraction.
Positive sample image is read in, coloured image is converted into gray level image, then extracts its HOG feature；Read in negative sample
Image, coloured image is converted into gray level image, then extracts its HOG feature.
Step 4, training SVM classifier, and generate the SVM classifier of corresponding training sample data.
Step 5, preserve grader.
Step 6, the accuracy of detection grader classification.
HOG features corresponding to positive and negative sample set are mixed, are randomly divided into two groups, one group is new training set, and one group is to test
Card collection, Autonomous test, the accuracy rate of testing classification device classification are carried out using detection set pair grader classification results.
Step 7, it is loaded into test image.
Step 8, extraction test image feature.
Test image is read in, coloured image is converted into gray level image, then extracts its HOG feature.
Step 9, using the SVM classifier trained test chart is classified, obtain radial direction electromagnetic interference and spiral shell
The classification results of electromagnetic interference and pit shape electromagnetic interference echo are revolved, realize the identification to radial direction electromagnetic interference echo.Such as test
As a result be radial direction electromagnetic interference echo, then EP (end of program).If test result is non-radial electromagnetic interference echo, then carry out next
Step.
Step 10, establish positive sample and integrate as helical form electromagnetic interference echo, negative sample integrates as pit shape electromagnetic interference
Echo, repeat step 2 arrive step 9, obtain spiral electromagnetic interference and the classification results of pit shape electromagnetic interference echo, realize
Identification to spiral electromagnetic interference echo and pit shape electromagnetic interference echo.
Further technical scheme is in the step 3 and step 8, and the specific method of HOG feature extractions is：First,
Calculate the gradient magnitude and gradient direction of each pixel in sample image.Then, some UNICOM regions are divided the image into, these
Connected region is called cell factory.Cell factory is divided into several blocks again, several passages are equally divided into by 0 °~180 °.
Secondly, the direction histogram of the gradient of each pixel in cell factory is gathered.Finally, it is the gradient of the pixel of different passages is big
It is small cumulative, one group is obtained by the vector that each passage pixel gradient is cumulative and forms, then in units of block, normalizing is carried out to vector
Change is handled, and the vector after normalized is chained up, and forms HOG sample characteristics descriptors.
It is to use two discrete differential templates (- 1,0,1) and (- 1,0,1) to calculate gradientT(T represents transposition) is respectively to figure
As horizontal direction and vertical direction are handled, the gradient of each pixel is obtained.
In step 4, obtaining SVM classifier detailed process is：The mistake of optimal svm classifier function is asked for according to known sample collection
Journey can solve classification problem.It is in some higher dimensional space first nonlinear problem shift conversion for nonlinear problem
Linear problem, then solve optimal classification surface in space after transformation.If given sample set is s={ (xi,yi) | i=1,
2n }, xi∈RdFor the vector in d dimension spaces.yi={+1, -1 } is xiCorresponding category label.I is sample set numbering.It is it is assumed that non-
Linear Mapping isWhereinFor mapping function, support vector cassification face can be write as：w
For weight vector, b is constant offset.For class interval, the optimization problem that class interval obtains maximum is represented by askingMinimum problems.Constraints isAccording to Karush-Kuhn-
Tucker (KKT) conditions and Lagrangian can obtain classification function and beWherein,
Sgn () is sign function.aiFor optimal Lagrange multiplier.X is sample to be tested.Higher dimensional space
Point multiplication operation (xiX) can be by the kernel function k (x of lower dimensional spacei, xj) replace, classification function is rewritable to be
Conventional kernel function is as shown in table 1, and different svm classifier functions can be produced by choosing different kernel functions, so as to
To different SVM classifiers, therefore, the inventive method can expand to different svm classifier sides by the selection of kernel function
Method.
Table 1 often uses kernel function
In order to make the purpose , technical scheme and advantage of the present invention be clearer, it is right below in conjunction with drawings and Examples
The present invention is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, and
It is not used in the restriction present invention.The basic procedure of the inventive method is as shown in figure 1, specifically include following steps：
Step 1, establish training sample set.It is 20 radial direction electromagnetic interferences to establish positive sample collection (positive sample)
Echo, establish positive sample listed files pos_list.txt；Negative sample collection (negative sample) is 20 helical forms
Electromagnetic interference echo and pit shape electromagnetic interference echo, establish negative sample listed files neg_list.txt.
Step 2, it is loaded into positive negative sample and distributes label.Positive sample list is loaded into, label is designated as label1, is loaded into negative sample
List, label is designated as label2, and label is collected and is designated as label.
Step 3, the positive and negative sample characteristics of extraction.Positive sample image is read in, coloured image is converted into gray level image, then made
Its HOG feature is extracted with hogcalculator functions；Negative sample image is read in, coloured image is converted into gray level image, then
Its HOG feature is extracted using hogcalculator functions.Here the specific method of hogcalculator functions extraction HOG features
For：First, the gradient magnitude and gradient direction of each pixel in sample image are calculated.Then, some 6*6 pictures are divided the image into
The UNICOM region of element, these connected regions are called cell factory.3*3 cell factory forms a block, average by 0 °~180 °
It is divided into 9 passages.Secondly, the direction histogram of the gradient of each pixel in cell factory is gathered.Finally, 9 differences are led to
The gradient magnitude of the pixel in road adds up, and obtains one group by the vector that each passage pixel gradient is cumulative and forms, then using block to be single
Position, is normalized to vector, the vector after normalized is chained up, and forms HOG sample characteristics descriptors.
Step 4, positive sample collection and negative sample collection training SVM classifier (positive sample collection and negative sample here are used respectively
Collection refers respectively to each self-corresponding HOG features), and generate the SVM classifier of corresponding training sample data.Generate SVM classifier
Detailed process is：Given known sample integrates as m={ (xi,yi) | i=1,2n }, xiFor positive and negative training sample set.yi={+1, -1 }
It is xiCorresponding category label.I is sample set numbering.The process of optimal svm classifier function is asked for according to known sample collection
Solves classification problem.The process of asking for of classification function has been described in detail in content of the invention step 4.Classification function isWherein, sgn () is sign function.aiFor optimal Lagrange multiplier.X is to treat
Test sample sheet.k(xi, xj) it is kernel function, the present embodiment Kernel Function is gaussian radial basis function：σ is kernel functional parameter.
Step 5, preserve grader.
Step 6, detection classification accuracy.HOG features corresponding to positive and negative sample set are mixed, are randomly divided into two groups, one group
For new training set, one group is checking collection, carries out self checking test using checking set pair grader classification results, obtains grader
The accuracy of classification.When concentrating the 15th figure to carry out self checking test using positive sample, classification results are positive sample, self checking
Detected value is 0.8947.When concentrating the 15th figure to carry out self checking test using negative sample, classification results are negative sample, self-correcting
Detected value is tested as 0.9474.As can be seen here, this time checking grader classification results are correct, and confidence level is higher.
Step 7, it is loaded into test image.Test image sample set is opened for radial direction electromagnetic interference echo 10, spiral electromagnetism
Interference figure is opened as 10 and pit electromagnetic interference image 10, and image size is 64 × 64.
Step 8, extraction test image feature.Test image is read in, coloured image is converted into gray level image, then used
Hogcalculator functions extract its HOG feature, and HOG feature extractions are identical with step 3.
Step 9, using the SVM classifier trained test chart is classified, obtain radial direction electromagnetic interference and spiral shell
The classification results of electromagnetic interference and pit shape electromagnetic interference echo are revolved, realize the identification to radial direction electromagnetic interference echo.Such as test
As a result be radial direction electromagnetic interference echo, then EP (end of program).If test result is non-radial electromagnetic interference echo, then carry out next
Step.
Step 10, establish positive sample and integrate (positive sample) as helical form electromagnetic interference echo listed files,
Negative sample integrates (negative sample) and arrives step 9 as pit shape electromagnetic interference echo listed files, repeat step 2, obtain
Spiral electromagnetic interference and the classification results of pit shape electromagnetic interference echo are obtained, is realized to spiral electromagnetic interference echo and pit shape electricity
The identification of magnetic disturbance echo.
The classification results of the present embodiment are as shown in Figure 2.The result that " class " presentation class device is classified in figure, " true " table
Affiliated type, " 1 " represent diametral interference electromagnetic echoes to diagram piece in itself, and " 2 " represent helical form electromagnetic interference echo, and " 3 " represent
Pit electromagnetic interference echo.Wherein, the 1st~2 behavior diametral interference electromagnetic echoes recognition result, the identification of diametral interference echo are accurate
Rate is 90%, and a spiral interference echo is included in 10 diametral interference echo recognition results；3rd~4 behavior spiral electromagnetism is done
Echo recognition result is disturbed, helical form interference echo recognition accuracy is 90%, is wrapped in 10 helical form interference echo recognition results
Containing a pit interference echo；5th~6 behavior pit electromagnetic interference echo recognition result, the identification of pit shape interference echo are accurate
Rate is 90%, and a spiral interference echo is included in 10 pit interference echo recognition results.Total accuracy rate, which can be calculated, is
90%.
Although reference be made herein to invention has been described for embodiments of the invention, however, it is to be understood that art technology
Personnel can be designed that a lot of other modifications and embodiment, and these modifications and embodiment will fall in original disclosed in the present application
Then within scope and spirit.More specifically, can be to theme combination layout in the range of disclosure and claim
Building block and/or layout carry out a variety of variations and modifications.
Claims (4)
1. a kind of weather radar electromagnetic interference echo recognition methods based on SVMs, it is characterized in that, step is as follows：
Step 1, establish training sample set
Establish positive sample to integrate as radial direction electromagnetic interference echo, negative sample integrates as helical form electromagnetic interference echo and pit shape electricity
Magnetic disturbance echo；
Step 2, it is loaded into positive negative sample and distributes label；
Step 3, the positive and negative sample characteristics of extraction
Positive sample image is read in, coloured image is converted into gray level image, then extracts its HOG feature；Negative sample image is read in,
Coloured image is converted into gray level image, then extracts its HOG feature；
Step 4, training SVM classifier, and generate the SVM classifier of corresponding training sample data；
Step 5, preserve grader
Step 6, the accuracy of detection grader classification
HOG features corresponding to positive and negative sample set are mixed, are randomly divided into two groups, one group is new training set, and one group is checking collection,
Autonomous test, the accuracy rate of testing classification device classification are carried out using checking set pair grader classification results；
Step 7, it is loaded into test image；
Step 8, extraction test image feature
Test image is read in, coloured image is converted into gray level image, then extracts its HOG feature；
Step 9, using the SVM classifier trained test chart is classified, obtain radial direction electromagnetic interference and spiral electricity
Magnetic disturbance and the classification results of pit shape electromagnetic interference echo, realize the identification to radial direction electromagnetic interference echo；Such as test result
For radial direction electromagnetic interference echo, then EP (end of program)；If test result is non-radial electromagnetic interference echo, then carry out in next step；
Step 10, establish positive sample and integrate as helical form electromagnetic interference echo, negative sample integrates as pit shape electromagnetic interference echo
Image, repeat step 2 arrive step 9, obtain spiral electromagnetic interference and the classification results of pit shape electromagnetic interference echo, realize to spiral shell
Revolve the identification of electromagnetic interference echo and pit shape electromagnetic interference echo.
2. the weather radar electromagnetic interference echo recognition methods based on SVMs as claimed in claim 1, it is characterized in that,
Further technical scheme is in the step 3 and step 8, and the specific method of HOG feature extractions is：First, sample is calculated
The gradient magnitude and gradient direction of each pixel in image；Then, some UNICOM regions, these connected regions are divided the image into
It is called cell factory, then cell factory is divided into several blocks, several passages is equally divided into by 0 °~180 °；Secondly, adopt
Collect the direction histogram of the gradient of each pixel in cell factory；Finally, the gradient magnitude of the pixel of different passages is added up,
One group is obtained by the vector that each passage pixel gradient is cumulative and forms, then in units of block, vector is normalized,
Vector after normalized is chained up, forms HOG sample characteristics descriptors；
It is to use two discrete differential templates (- 1,0,1) and (- 1,0,1) to calculate gradientT, T represents transposition, respectively to image level
Direction and vertical direction are handled, and obtain the gradient of each pixel.
3. the weather radar electromagnetic interference echo recognition methods based on SVMs as claimed in claim 1, it is characterized in that,
In step 4, obtaining SVM classifier detailed process is：The process of optimal svm classifier function is asked for according to known sample collection to be solved
Certainly classification problem, it is the linear problem in some higher dimensional space first nonlinear problem shift conversion for nonlinear problem,
Then optimal classification surface is solved in space after transformation, if given sample set is s={ (xi,yi) | i=1,2n }, xi∈Rd
For the vector in d dimension spaces, yi={+1, -1 } is xiCorresponding category label, i are sample set numberings, it is assumed that Nonlinear Mapping isWhereinFor mapping function, support vector cassification face is write as：W is weight vector, b
For constant offset,For class interval, the optimization problem that class interval obtains maximum is represented by askingMinimum value ask
Topic, constraints areAccording to KKT (Karush-Kuhn-Tucker) conditions and drawing
Ge Lang functions obtain classification function：
Wherein, sgn () is sign function, aiFor optimal Lagrange multiplier, x is sample to be tested,
Point multiplication operation (the x of higher dimensional spaceiX) by the kernel function k (x of lower dimensional spacei,xj) replace, classification function is rewritable to be
4. the weather radar electromagnetic interference echo recognition methods based on SVMs as claimed in claim 1, conventional core
Function is as follows：
Polynomial kernel function k (x, xi)=[(xxi)+1]qQ=1,2,3 ...；
Gaussian radial basis functionσ is parameter；
Sigmoid function k (x, xi)=tanh (b (x, xi)+c) b, c be constant；
Fourier spaceWherein N is constant；
B-spline kernel function k (x, xi)=B2p+1(x-xi) wherein B2p+1(x) it is 2p+1 ranks B；
Different svm classifier functions can be produced by choosing different kernel functions, so as to obtain different SVM classifiers, therefore can be with
Different svm classifier methods is expanded to by the selection of kernel function.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710750362.6A CN107679453B (en) | 2017-08-28 | 2017-08-28 | Weather radar electromagnetic interference echo identification method based on support vector machine |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710750362.6A CN107679453B (en) | 2017-08-28 | 2017-08-28 | Weather radar electromagnetic interference echo identification method based on support vector machine |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107679453A true CN107679453A (en) | 2018-02-09 |
| CN107679453B CN107679453B (en) | 2020-04-17 |
Family
ID=61134974
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710750362.6A Expired - Fee Related CN107679453B (en) | 2017-08-28 | 2017-08-28 | Weather radar electromagnetic interference echo identification method based on support vector machine |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107679453B (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108491618A (en) * | 2018-03-19 | 2018-09-04 | 武汉三相电力科技有限公司 | Fault current recognition methods, device, equipment and medium |
| CN109190664A (en) * | 2018-07-28 | 2019-01-11 | 天津大学 | The classification method of radar electromagnetic interference echo based on GASVM algorithm |
| CN109343062A (en) * | 2018-12-13 | 2019-02-15 | 中国气象局气象探测中心 | A kind of recognition methods and system of diametral interference echo and precipitation echo |
| CN109766939A (en) * | 2018-12-29 | 2019-05-17 | 佛山科学技术学院 | A kind of galvanized steel and mild steel classification method and device based on photo |
| CN110221266A (en) * | 2019-06-11 | 2019-09-10 | 哈尔滨工程大学 | A kind of marine radar target rapid detection method based on support vector machines |
| CN110533069A (en) * | 2019-07-25 | 2019-12-03 | 西安电子科技大学 | A kind of two-dimentional chaff distribution character recognition methods based on algorithm of support vector machine |
| CN112418309A (en) * | 2020-11-20 | 2021-02-26 | 中国舰船研究设计中心 | Electromagnetic compatibility management and control method prediction method based on machine learning |
| CN114280571A (en) * | 2022-03-04 | 2022-04-05 | 北京海兰信数据科技股份有限公司 | Processing method, device and equipment of rain clutter signals |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1595195A (en) * | 2004-06-17 | 2005-03-16 | 上海交通大学 | Super broad band land radar automatic target identification method based on information fusion |
| CN104063722A (en) * | 2014-07-15 | 2014-09-24 | 国家电网公司 | Safety helmet identification method integrating HOG human body target detection and SVM classifier |
| CN105260524A (en) * | 2015-09-28 | 2016-01-20 | 大连理工大学 | Method for simulating two-dimensional echo sequence image of ship navigation radar in scanning state |
| CN106127725A (en) * | 2016-05-16 | 2016-11-16 | 北京工业大学 | A kind of millimetre-wave radar cloud atlas dividing method based on multiresolution CNN |
| KR101708275B1 (en) * | 2015-03-05 | 2017-02-20 | 국방과학연구소 | Method and Apparatus of recognizing Radar Scan Pattern based on Support Vector Machine |
-
2017
- 2017-08-28 CN CN201710750362.6A patent/CN107679453B/en not_active Expired - Fee Related
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1595195A (en) * | 2004-06-17 | 2005-03-16 | 上海交通大学 | Super broad band land radar automatic target identification method based on information fusion |
| CN104063722A (en) * | 2014-07-15 | 2014-09-24 | 国家电网公司 | Safety helmet identification method integrating HOG human body target detection and SVM classifier |
| KR101708275B1 (en) * | 2015-03-05 | 2017-02-20 | 국방과학연구소 | Method and Apparatus of recognizing Radar Scan Pattern based on Support Vector Machine |
| CN105260524A (en) * | 2015-09-28 | 2016-01-20 | 大连理工大学 | Method for simulating two-dimensional echo sequence image of ship navigation radar in scanning state |
| CN106127725A (en) * | 2016-05-16 | 2016-11-16 | 北京工业大学 | A kind of millimetre-wave radar cloud atlas dividing method based on multiresolution CNN |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108491618A (en) * | 2018-03-19 | 2018-09-04 | 武汉三相电力科技有限公司 | Fault current recognition methods, device, equipment and medium |
| CN109190664A (en) * | 2018-07-28 | 2019-01-11 | 天津大学 | The classification method of radar electromagnetic interference echo based on GASVM algorithm |
| CN109343062A (en) * | 2018-12-13 | 2019-02-15 | 中国气象局气象探测中心 | A kind of recognition methods and system of diametral interference echo and precipitation echo |
| CN109343062B (en) * | 2018-12-13 | 2021-07-20 | 中国气象局气象探测中心 | Method and system for identifying radial interference echo and precipitation echo |
| CN109766939A (en) * | 2018-12-29 | 2019-05-17 | 佛山科学技术学院 | A kind of galvanized steel and mild steel classification method and device based on photo |
| CN109766939B (en) * | 2018-12-29 | 2023-09-26 | 佛山科学技术学院 | Photo-based galvanized steel and low-carbon steel classification method and device |
| CN110221266B (en) * | 2019-06-11 | 2022-12-13 | 哈尔滨工程大学 | Marine radar target rapid detection method based on support vector machine |
| CN110221266A (en) * | 2019-06-11 | 2019-09-10 | 哈尔滨工程大学 | A kind of marine radar target rapid detection method based on support vector machines |
| CN110533069A (en) * | 2019-07-25 | 2019-12-03 | 西安电子科技大学 | A kind of two-dimentional chaff distribution character recognition methods based on algorithm of support vector machine |
| CN112418309A (en) * | 2020-11-20 | 2021-02-26 | 中国舰船研究设计中心 | Electromagnetic compatibility management and control method prediction method based on machine learning |
| CN112418309B (en) * | 2020-11-20 | 2022-06-03 | 中国舰船研究设计中心 | Electromagnetic compatibility management and control method prediction method based on machine learning |
| CN114280571B (en) * | 2022-03-04 | 2022-07-19 | 北京海兰信数据科技股份有限公司 | Method, device and equipment for processing rain clutter signals |
| CN114280571A (en) * | 2022-03-04 | 2022-04-05 | 北京海兰信数据科技股份有限公司 | Processing method, device and equipment of rain clutter signals |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107679453B (en) | 2020-04-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107679453A (en) | Weather radar electromagnetic interference echo recognition methods based on SVMs | |
| Hu et al. | Detection of PCB surface defects with improved faster-RCNN and feature pyramid network | |
| CN109118479B (en) | Capsule network-based insulator defect identification and positioning device and method | |
| CN106682569A (en) | Fast traffic signboard recognition method based on convolution neural network | |
| CN107945153A (en) | A kind of road surface crack detection method based on deep learning | |
| CN109583324A (en) | A kind of pointer meters reading automatic identifying method based on the more box detectors of single-point | |
| Zhang et al. | Road recognition from remote sensing imagery using incremental learning | |
| CN102945378B (en) | Method for detecting potential target regions of remote sensing image on basis of monitoring method | |
| CN113205063A (en) | Visual identification and positioning method for defects of power transmission conductor | |
| CN106022375A (en) | HU invariant moment and support vector machine-based garment style identification method | |
| CN112990282B (en) | Classification method and device for fine-granularity small sample images | |
| CN107545571A (en) | A kind of image detecting method and device | |
| Zhai et al. | A generative adversarial network based framework for unsupervised visual surface inspection | |
| CN103839078A (en) | Hyperspectral image classifying method based on active learning | |
| CN109858570A (en) | Image classification method and system, computer equipment and medium | |
| CN109145846A (en) | Material microstructure intelligent recognition analysis system and analysis method | |
| CN103177266A (en) | Intelligent stock pest identification system | |
| CN109614866A (en) | Method for detecting human face based on cascade deep convolutional neural networks | |
| CN110659637A (en) | Electric energy meter number and label automatic identification method combining deep neural network and SIFT features | |
| CN110175614A (en) | A kind of detection method of printed circuit board via hole inner wall quality | |
| Zaafouri et al. | A vision approach for expiry date recognition using stretched gabor features. | |
| CN108073940A (en) | A kind of method of 3D object instance object detections in unstructured moving grids | |
| CN108509861A (en) | A kind of method for tracking target and device combined based on sample learning and target detection | |
| CN114529906A (en) | Method and system for detecting abnormity of digital instrument of power transmission equipment based on character recognition | |
| CN106372618A (en) | Road extraction method and system based on SVM and genetic algorithm |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20200417 Termination date: 20200828 |