CN105607049B - A kind of data partition method based on homogeneity test - Google Patents

A kind of data partition method based on homogeneity test Download PDF

Info

Publication number
CN105607049B
CN105607049B CN201510990130.9A CN201510990130A CN105607049B CN 105607049 B CN105607049 B CN 105607049B CN 201510990130 A CN201510990130 A CN 201510990130A CN 105607049 B CN105607049 B CN 105607049B
Authority
CN
China
Prior art keywords
data
matrix
amplitude
sectionized
carry
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
Application number
CN201510990130.9A
Other languages
Chinese (zh)
Other versions
CN105607049A (en
Inventor
李杨
郭美玲
张宁
位寅生
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Harbin Institute of Technology
Original Assignee
Harbin Institute of Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Harbin Institute of Technology filed Critical Harbin Institute of Technology
Priority to CN201510990130.9A priority Critical patent/CN105607049B/en
Publication of CN105607049A publication Critical patent/CN105607049A/en
Application granted granted Critical
Publication of CN105607049B publication Critical patent/CN105607049B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
    • G01S7/02—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00
    • G01S7/41—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00 using analysis of echo signal for target characterisation; Target signature; Target cross-section
    • G01S7/414—Discriminating targets with respect to background clutter

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Radar Systems Or Details Thereof (AREA)

Abstract

A kind of data partition method based on homogeneity test, the present invention relates to the data partition method based on homogeneity test.When clutter environment is non-homogeneous in practical situations both the invention aims to solving prior art, classical way robust performance is low, target detection performance is low and only considers the homogeney of local data, for other classical ways and it is inapplicable the problem of.Detailed process is：First, start；2nd, radar return amplitude data is obtained, carry is set；3rd, radar return amplitude data is segmented on demand, and initializes R；4th, the homogeneity calculated between different segmentations is horizontal, and assignment is carried out to R；5th, carry is adjusted, by carry+1；6th, the data element for being more than carry in R is found, if it is possible to find, carry out homogeneous level calculation and carry out five again, if can not find, carry out seven；7th, new sectionized matrix tt is generated according to sectionized matrix R；8th, terminate.The present invention is applied to field of signal processing.

Description

A kind of data partition method based on homogeneity test
Technical field
The present invention relates to the data partition method based on homogeneity test.
Background technology
Classical signal processing algorithm and detection technique, all assume that whole clutter background meets same probability distribution, with skill Art develops, it has now been found that the clutter background of the usual actual especially airborne head-down radar of clutter background is simultaneously unsatisfactory for same probability Distribution, this otherness bring the reduction of classical way robust performance.Such as space-time adaptive treatment technology (STAP), to obtain Robustness is obtained, then requiring can accurate estimate covariance matrix.Covariance matrix is obtained using maximal possibility estimation, is estimated used Sample is generally from unit both sides to be detected, to obtain superperformance (reality output believes that miscellaneous noise ratio loss is no more than 3dB), training Sample need to meet following two condition：1st, training sample meets independent same distribution (I.I.D) with sample to be detected；2nd, training sample 2 times of number more than degree of freedom in system.But in practical situations both, clutter environment is heterogeneous, be on the one hand due to landform in itself The characteristics of cause；Another aspect is due to that the array structure of airborne radar causes.In the case, the condition of satisfaction is hardly resulted in Sample, so STAP performances can be greatly reduced.The problem of such is equally existed for traditional target detection technique, is passed The CFAR detection technology (CFAR) of system is applied to statistical property identical region.But in practical situations both, clutter environment right and wrong The decline of target detection performance can uniformly be brought.Jeong in 2013 et al. proposes CI-CFAR detectors, excellent by being fitted Degree test selection homogeneity unit carries out CFAR detections, although improving the detection probability of target, the method as reference unit Only consider the homogeney of local data, for other classical ways such as STAP technologies and do not apply to.
The content of the invention
When clutter environment is non-homogeneous in practical situations both the invention aims to solving prior art, classical way Robust performance is low, target detection performance is low and only considers the homogeney of local data, for other classical ways and does not apply to The problem of, and a kind of data partition method based on homogeneity test proposed.
Above-mentioned goal of the invention is achieved through the following technical solutions：
Step 1: start；
Step 2: obtaining radar return amplitude data, carry is set；
Step 3: being segmented on demand to radar return amplitude data, and initialize sectionized matrix R；
Step 4: the homogeneity between calculating different segmentations is horizontal, assignment is carried out to sectionized matrix R；
Step 5: adjustment carry, by carry+1；
Step 6: find the data element for being more than carry in sectionized matrix R, if it is possible to find and then carry out homogeneity level Calculating carries out step 5 again, if can not find, carries out step 7；
Step 7: new sectionized matrix tt is generated according to sectionized matrix R；
Step 8: terminate.
Invention effect
Method proposes a kind of data partition algorithm based on homogeneity test, realize according to homogeney to real data It is overall carry out subregion, the data positioned at same subregion is met same probability distribution, reach classical signal processing algorithm And the data of detection technique meet the requirement of same probability distribution, it is easy to the signal processing algorithm of classics and the reality of detection technique Apply, realize the raising of robust performance, target detection performance improves and equally applicable for other method.Solves the letter of classics Number Processing Algorithm and detection technique in practical situations both, clutter environment be it is heterogeneous can bring robust performance decline the problem of And K samples Anderson-Darling inspections are not asked what the homogeneity sexual intercourse between data entirety was tested in CI-CFAR Topic.Subregion is carried out to radar return data based on this method, the data progress covariance square for belonging to same subregion is utilized after subregion Battle array estimation carries out STAP processing again, and the signal to noise ratio somewhere obtained carries out covariance matrix again than not using this method is direct Carry out the high 13.68dB of signal to noise ratio that STAP processing is obtained at same one again.
It is as shown in Figure 1 using emulation data, emulation data mode in experiment.Share 1396 range gates, 128 Doppler Unit, wherein logn1 are logarithm normal distribution 1：Subregion meets logarithm normal distribution, logarithmic average 7, and logarithm standard deviation is 0.7.Wbl1 is Weibull distribution 1：Subregion meets Weibull distribution, scale parameter 400, form parameter 2.Logn2 be logarithm just State distribution 2：Subregion meets logarithm normal distribution, logarithmic average 5, logarithm standard deviation 0.4.Wbl2 is Weibull distribution 2：Subregion Meet Weibull distribution, scale parameter 200, form parameter 8.It is as shown in Figure 2 to emulate data dB values.
This emulation digital simulation be airborne head-down radar certain channel data, to this data carry out subregion purpose be just In the implementation of follow-up space-time adaptive Processing Algorithm.Because STAP algorithms use 3DT-STAP fast algorithms, and the radar simulated Echo data includes 20 passages, to meet that the data volume that the minimum subregion of RMB criterions includes should be greater than being equal to 120.It is so sharp With the data partition method based on homogeneity test, using the reference windows type of 120*1 (range gate * doppler cells). The division result that arrives is as shown in figure 3,8.0852s when whole process shares.Rayleigh distributed then is carried out to each subregion, logarithm is just State is distributed, the Probability Distribution Fitting of Weibull distribution, calculating parameter size, utilizes the KS test of fitness of fots more each subregion one by one With the degree of closeness of three kinds of probability distribution, the immediate probability distribution as the subregion is selected.Obtain result and reality The comparison of situation is as shown in table 1.
The actual conditions of table 1 contrast with testing result
Fig. 3 shows that the data partition method based on homogeneity test can realize the differentiation to different probability distribution.But this In division result and actual conditions it is not fully identical.Belong to probability distribution of the same race apart from gate cell 1-698 in emulation data, But the result that this method obtains is to belong to probability distribution of the same race apart from gate cell 1-720.Because the selection of reference windows is made Into, K-AD homogeneity tests algorithm is assumed by data in reference windows and meets same probability distribution first, i.e., every 120 distances Unit is one and is segmented sample, and data belong to same probability distribution in sample.720 be the segmentation nearest from 698, so can obtain The above situation.Fig. 4 to Fig. 7 is observed, shows the Probability Distribution Fitting situation of different subregions, wherein having what is come in and gone out with actual conditions It is second subregion, the actual conditions subregion meets Weibull distribution, but testing result shows that this distribution meets logarithm normal distribution. Caused by such case Producing reason is also reference windows selection, between the 699 to 720th range gate of each doppler cells Data are subdivided into the subregion different from itself probability distribution, while the probability distribution of the subregion is changed.Such as Fig. 4 and figure Shown in 5, first subregion (Fig. 4) by it is this influence it is smaller Probability Distribution Fitting is not influenced too much, the second subregion by It is this to have a great influence, the probability distribution of script is produced change, turned to by Weibull distribution and be more likely to lognormal subregion.But This species diversity has no effect on the checking of K-AD homogeneity test subregion validity, can equally realize point of K-AD homogeneity tests Area's ability.
Brief description of the drawings
Fig. 1 is emulation data mode figure, and wbl 1 is Weibull distribution 1, and wbl2 is Weibull distribution 2, and logn1 is lognormal 1, logn2 of distribution is logarithm normal distribution 2；
Fig. 2 is emulation data dB value figures；
Fig. 3 is emulation data K-AD division result figures；
Fig. 4 is the Probability Distribution Fitting schematic diagram of subregion 1；
Fig. 5 is the Probability Distribution Fitting schematic diagram of subregion 2；
Fig. 6 is the Probability Distribution Fitting schematic diagram of subregion 3；
Fig. 7 is the Probability Distribution Fitting schematic diagram of subregion 4；
Fig. 8 is flow chart of the present invention.
Embodiment
Embodiment one：With reference to Fig. 8 illustrate present embodiment, present embodiment it is a kind of based on homogeneity test Data partition method is specifically to be prepared according to following steps：
Step 1: start；
Step 2: obtaining radar return amplitude data, carry is set；
Step 3: being segmented on demand to radar return amplitude data, and initialize sectionized matrix R；
Step 4: the homogeneity between calculating different segmentations is horizontal, assignment is carried out to sectionized matrix R；
Step 5: adjustment carry, by carry+1；
Step 6: find the data element for being more than carry in sectionized matrix R, if it is possible to find and then carry out homogeneity level Calculating carries out step 5 again, if can not find, carries out step 7；
Step 7: new sectionized matrix tt is generated according to sectionized matrix R；
Step 8: terminate.
Embodiment two：Present embodiment is unlike embodiment one：Thunder is obtained in the step 2 Up to echo amplitude data, carry is set；Detailed process is：
Radar return data matrix is obtained, echo data element is plural number, takes amplitude to obtain radar return amplitude data square Battle array, it is 0 to set carry.
Other steps and parameter are identical with embodiment one.
Embodiment three：Present embodiment is unlike embodiment one or two：It is right in the step 3 Radar return amplitude data piecemeal, and initialize sectionized matrix R on demand；Detailed process is：
Reference windows matrix is selected by actual demand, reference windows matrix is m'*n', and m' represents reference windows row Number, n' represent reference windows columns；If it is m*n to take amplitude to obtain radar return amplitude data matrix, m is to take amplitude to obtain radar Echo amplitude data matrix line number, n are to take amplitude to obtain radar return amplitude data matrix columns；By reference windows to taking amplitude Obtain radar return amplitude data matrix to be segmented, be divided intoSection, sectionized matrix R, sectionized matrix R line numbers are set ForColumns isIt is 0 to initialize all data elements of sectionized matrix R；M' spans areN' takes Value scope isM spans are 1≤m≤10000, and n spans are 1≤n≤10000；
The * is multiplication sign.
Other steps and parameter are identical with embodiment one or two.
Embodiment four：Unlike one of present embodiment and embodiment one to three：It is described to press reference Window is segmented to taking amplitude to obtain radar return amplitude data matrix, is divided intoSection；Detailed process is：
IfAmplitude will be taken to obtain in radar return amplitude data matrixIs incorporated into m elementDuan Zhong, amplitude will be taken to obtain in radar return amplitude data matrixIs incorporated into n elementDuan Zhong；The * is multiplication sign.
Other steps and parameter are identical with one of embodiment one to three.
Embodiment five：Unlike one of present embodiment and embodiment one to four：The step 4 The middle homogeneity level calculated between different segmentations, assignment is carried out to sectionized matrix R；Detailed process is：
Make sectionized matrix R (1,1)=1；Calculating (1,1) section takes amplitude to obtain the data of radar return amplitude data matrix Duan Yu (i, j) section takes amplitude to obtain the data segment homogeneity level of radar return amplitude data matrix, Calculate gained homogeneity level and two-symbol and as sectionized matrix R (i, j) the value of step.
Other steps and parameter are identical with one of embodiment one to four.
Embodiment six：Unlike one of present embodiment and embodiment one to five：It is described to calculate the (1,1) section takes that amplitude obtains the data segment of radar return amplitude data matrix and (i, j) section takes amplitude to obtain radar return width The data segment homogeneity of Value Data matrix is horizontal,Detailed process is：
Calculate (1,1) section and take that amplitude obtains the data segment of radar return amplitude data matrix and (i, j) section takes amplitude Obtain the test statistics T of the standardization of the data segment of radar return amplitude data matrixkN；Sectionized matrix R the 1st row the 1st row Element value is 1；
According to the test statistics T of standardizationkNAnd K-AD homogeneity tests tables of critical values sets homogeneity horizontal, setting -5 <TkNHorizontal≤0.326 homogeneity is 1,0.326<TkNHorizontal≤1.225 homogeneities are 2,1.225<TkN≤ 1.960 homogeneous water Put down as 3,1.960<TkNHorizontal≤2.719 homogeneities are 4,2.719<TkNHorizontal≤3.752 homogeneities are 5,3.752<TkN≤ Horizontal 100 homogeneities are 6；K-AD is K samples Anderson-Darling.
Other steps and parameter are identical with one of embodiment one to five.
Embodiment seven：Unlike one of present embodiment and embodiment one to six：The step 6 It is middle to find the data element for being more than carry in sectionized matrix R, if it is possible to find, carry out homogeneous level calculation and walked again Rapid five, if can not find, carry out step 7；Detailed process is：
Start from sectionized matrix R (1,1) individual element, find first and be more than the data element of carry in step 5 Position, if can search out and first position for being more than the data element of carry in step 5 is located at sectionized matrix R xth row the Y is arranged, then carries out homogeneous level calculation, x spans areY spans arePerform step Rapid five；
If the data element more than carry in step 5 can not be searched out, step 7 is performed.
Other steps and parameter are identical with one of embodiment one to six.
Embodiment eight：Unlike one of present embodiment and embodiment one to seven：It is described from subregion Matrix R (1,1) individual element starts, and first position for being more than the data element of carry in step 5 is found, if can find Arrive and it is located at xth row y row, then carry out homogeneous level calculation, detailed process is：
Determine to take amplitude to obtain (x, y) place number of radar return amplitude data matrix according to sectionized matrix R x rows y row According to section, regeneration block matrix R xth row y column datas element is carry+1 in step 5, continually looks for sectionized matrix R data Element is more than all positions of carry in step 5, and positional representation is (i1, j1),I1 tables Line number where showing position, columns where j1 represents position, takes amplitude to obtain radar return amplitude data according to i1 row j1 row determinations (i1, j1) place data segment of matrix, the homogeneity level of (a, b) place data segment and (i1, j1) place data segment is calculated, more The column data elements of row jth 1 of new sectionized matrix R the i-th 1 are carry in homogeneous level+step 5 for calculating.
Other steps and parameter are identical with one of embodiment one to seven.
Embodiment nine：Unlike one of present embodiment and embodiment one to eight：The step 7 It is middle that new sectionized matrix tt is generated according to sectionized matrix R；Detailed process is：
New sectionized matrix tt sizes are identical with taking amplitude to obtain radar return amplitude data matrix size, are m rows, n row； If a is not equal toB is not equal toA (m'-1)+1 in tt matrixes is then made to arrive am' rows, b (n'-1)+1 to bn' row Data element value be equal to sectionized matrix R in a row b column data elements value；
If a is equal toB is not equal toA (m'-1)+1 in tt matrixes is then made to arrive m rows, b (n'-1)+1 is arrived The value of the data element of bn' row is equal to the value of a row b column data elements in sectionized matrix R；
If a is not equal toB is equal toA (m'-1)+1 in tt matrixes is then made to arrive am' rows, b (n'-1)+1 Value to the data element of n row is equal to the value of a row b column data elements in sectionized matrix R；
If a is equal toB is equal toA (m'-1)+1 in tt matrixes is then made to arrive m rows, b (n'-1)+1 to n row Data element value be equal to sectionized matrix R in a row b column data elements value.
Other steps and parameter are identical with one of embodiment one to eight.
Beneficial effects of the present invention are verified using following examples：
Embodiment one：
A kind of data partition method based on homogeneity test of the present embodiment is specifically to be prepared according to following steps：
It is as shown in Figure 1 using emulation data, emulation data mode in experiment.Share 1396 range gates, 128 Doppler Unit, wherein logn1 are logarithm normal distribution 1：Subregion meets logarithm normal distribution, logarithmic average 7, and logarithm standard deviation is 0.7.Wbl1 is Weibull distribution 1：Subregion meets Weibull distribution, scale parameter 400, form parameter 2.Logn2 be logarithm just State distribution 2：Subregion meets logarithm normal distribution, logarithmic average 5, logarithm standard deviation 0.4.Wbl2 is Weibull distribution 2：Subregion Meet Weibull distribution, scale parameter 200, form parameter 8.It is as shown in Figure 2 to emulate data dB values.
This emulation digital simulation be airborne head-down radar certain channel data, to this data carry out subregion purpose be just In the implementation of follow-up space-time adaptive Processing Algorithm.Because follow-up STAP algorithms use 3DT-STAP fast algorithms, and simulate Radar return packet contains 20 passages, to meet that the RMB criterions data volume that then minimum subregion includes should be greater than being equal to 120. So using the data partition method based on homogeneity test, using the reference window of 120*1 (apart from gate cell * doppler cells) Mouth type.Obtained division result is as shown in figure 3,8.0852s when whole process shares.Rayleigh point then is carried out to each subregion Cloth, logarithm normal distribution, the Probability Distribution Fitting of Weibull distribution, calculating parameter size, compared one by one using the KS test of fitness of fots The degree of closeness of more each subregion and three kinds of probability distribution, select the immediate probability distribution as the subregion.Obtain As a result it is as shown in table 1 with the comparison of actual conditions.
The actual conditions of table 1 contrast with testing result
Fig. 3 shows that the data partition method based on homogeneity test can realize the differentiation to different probability distribution.But this In division result and actual conditions it is not fully identical.Belong to probability distribution of the same race apart from gate cell 1-698 in emulation data, But the result that this method obtains is to belong to probability distribution of the same race apart from gate cell 1-720.Because the selection of reference windows is made Into, K-AD homogeneity tests algorithm is assumed by data in reference windows and meets same probability distribution first, i.e., every 120 distances Unit is one and is segmented sample, and data belong to same probability distribution in sample.720 be the segmentation nearest from 698, so can obtain The above situation.Fig. 4 to Fig. 7 is observed, shows the Probability Distribution Fitting situation of different subregions, wherein having what is come in and gone out with actual conditions It is second subregion, the actual conditions subregion meets Weibull distribution, but testing result shows that this distribution meets logarithm normal distribution. Caused by such case Producing reason is also reference windows selection, between the 699 to 720th range gate of each doppler cells Data are subdivided into the subregion different from itself probability distribution, while the probability distribution of the subregion is changed.Such as Fig. 4 and figure Shown in 5, first subregion (Fig. 4) by it is this influence it is smaller Probability Distribution Fitting is not influenced too much, the second subregion by It is this to have a great influence, the probability distribution of script is produced change, turned to by Weibull distribution and be more likely to lognormal subregion.But This species diversity has no effect on the checking of K-AD homogeneity test subregion validity, can equally realize point of K-AD homogeneity tests Area's ability.
K-AD homogeneity tests are also known as K samples Anderson-Darling inspections and can be understood as broad sense AD inspections, its Each sample can be examined in the case where sample distribution function expression formula is unknown whether with distribution, H0:F1=F2=F3…FN, H0For Null hypothesis, wherein F1…FNFor each sample distribution function (CDF).For sample vector X=[B1,B2,…BK], K represents data segment Number (K >=2), this method K take 2, use Bi={ x1,x2,…xnRepresent data segment, niWithRepresent data segment BiInterior data sample This number and its empirical distribution function (EDF), N=∑s niRepresent all data sample numbers, H in vectorial XN(x) it is all data samples EDF.K-AD homogeneity tests statistic can be expressed as：
Wherein, DN={ x ∈ R:HN(x)<1 }, DNFor sample, all samples in vectorial X are deployed, arranged from small to large To Z1<…<ZNMake MijFor data segment BiIt is not more than Z in (1≤i≤K)jThe data sample number of (1≤j≤N).The K- of discrete form AD homogeneity test statisticsIt can be expressed as：
K-AD homogeneity test statisticsMathematic expectaionAnd varianceIt is as follows respectively：
Wherein abcd is calculated by following formula：
In formula, H is the sample size of all data segments sum reciprocal；
Measurement can be passed throughNull hypothesis H0 is examined with the similarity degree of Gaussian distribution.By normalized After obtain TKN, use σNRepresentStandard deviation.
TkNTo standardize test statistics；
Similar with the identification of AD family of distributions, K-AD homogeneity tests are by searching tables of critical values, and whether judgement sample is the same as distribution. If TKNLess than critical value t of the Gaussian distribution under level of confidence αK-1(α), then receive null hypothesis H0, otherwise refusal is former false If.
The K-AD homogeneity test tables of critical values t of table 2K-1(α)
For the homogeneity inspection critical values of K samples AD corresponding to other K values, can be obtained by regression equation interpolation calculation Arrive
tK-1(α) is that corresponding K-AD examines critical critical value under level of confidence α, and α is level of confidence, simulation sequence a For Weibull distribution, parameter 3,2；Sequence b is respectively Weibull distribution (3,2) Gaussian Profile (3,2), Gaussian Profile (3,1), Wei Uncle's distribution (2,3).The length for changing two sequences is respectively 30,60,90,120,150,1200.Carry out 10000 Monte Carlo realities Test, calculate respective accuracy.Simulation result is as shown in table 3.
The simulation result of table 3
Simulation result shows：K-AD homogeneity tests can be realized in most cases distinguishes different probability distribution； It is but weaker for similar probability resolution capability.Wherein weber (3,2) and the average of Gauss (3,2) approaches, and variance is close, K-AD Homogeneity test is relatively low to the correctness of the two homogeneity test.The performance of K-AD homogeneity tests is with the increase of sequence length And increase, when using the algorithm of K-AD homogeneity test subregions, the selection of reference windows size is also critically important, it is ensured that reference The length for the sequence that window is converted into has certain confidence level.On the whole, reference windows size selects 120 K-AD homogeneys to examine Test with the ability for distinguishing different probability distribution.
Several steps once are broadly divided into based on K-AD homogeneity test partitioning algorithms：
(1) multidomain treat-ment is carried out to real data by reference windows,
(2) homogeneous horizontal check is carried out to different segmentations, homogeneous level, regeneration block matrix R squares is determined according to table 2 Battle array.
(3) according to carry information and flag judge whether that homogeneous horizontal check need to be continued, it is necessary to then return to (2) step, it is not necessary to then carry out (4) step.
(4) sectionized matrix R is adjusted, is easy to other follow-up steps to use.
The present invention can also have other various embodiments, in the case of without departing substantially from spirit of the invention and its essence, this area Technical staff works as can make various corresponding changes and deformation according to the present invention, but these corresponding changes and deformation should all belong to The protection domain of appended claims of the invention.

Claims (7)

  1. A kind of a kind of 1. data partition method based on homogeneity test, it is characterised in that data partition based on homogeneity test Method is specifically what is followed the steps below：
    Step 1: start；
    Step 2: obtaining radar return amplitude data, carry is set；Process is：
    Radar return data matrix is obtained, echo data element is plural number, takes amplitude to obtain radar return amplitude data matrix, if Carry is put as 0；
    Step 3: being segmented on demand to radar return amplitude data, and initialize sectionized matrix R；Process is：
    Reference windows matrix is selected by actual demand, reference windows matrix is m'*n', and m' represents reference windows line number, N' represents reference windows columns；If it is m*n to take amplitude to obtain radar return amplitude data matrix, m returns to take amplitude to obtain radar Wave amplitude data matrix line number, n are to take amplitude to obtain radar return amplitude data matrix columns；By reference windows to taking amplitude to obtain It is segmented, is divided into radar return amplitude data matrixSection, sets sectionized matrix R, and sectionized matrix R line numbers areColumns isIt is 0 to initialize all data elements of sectionized matrix R；M' spans areN' value models Enclose forM spans are 1≤m≤10000, and n spans are 1≤n≤10000；
    Step 4: the homogeneity between calculating different segmentations is horizontal, assignment is carried out to sectionized matrix R；
    Step 5: adjustment carry, by carry+1；
    Step 6: find the data element for being more than carry in sectionized matrix R, if it is possible to find, carry out homogeneous level calculation Step 5 is carried out again, if can not find, carries out step 7；
    Step 7: new sectionized matrix tt is generated according to sectionized matrix R；
    Step 8: terminate.
  2. A kind of 2. data partition method based on homogeneity test according to claim 1, it is characterised in that：It is described to press reference Window is segmented to taking amplitude to obtain radar return amplitude data matrix, is divided intoSection；Detailed process is：
    IfAmplitude will be taken to obtain in radar return amplitude data matrix Is incorporated into m elementDuan Zhong, amplitude will be taken to obtain in radar return amplitude data matrixTo n Individual element is incorporated intoDuan Zhong.
  3. A kind of 3. data partition method based on homogeneity test according to claim 2, it is characterised in that：The step 4 The middle homogeneity level calculated between different segmentations, assignment is carried out to sectionized matrix R；Detailed process is：
    Make sectionized matrix R (1,1)=1；Calculate (1,1) section take amplitude obtain the data segment of radar return amplitude data matrix with (i, j) section takes amplitude to obtain the data segment homogeneity level of radar return amplitude data matrix, Calculate gained homogeneity level and two-symbol and as sectionized matrix R (i, j) the value of step.
  4. A kind of 4. data partition method based on homogeneity test according to claim 3, it is characterised in that：It is described to calculate the (1,1) section takes that amplitude obtains the data segment of radar return amplitude data matrix and (i, j) section takes amplitude to obtain radar return width The data segment homogeneity of Value Data matrix is horizontal,Detailed process is：
    Calculate (1,1) section and take that amplitude obtains the data segment of radar return amplitude data matrix and (i, j) section takes amplitude to obtain The test statistics T of the standardization of the data segment of radar return amplitude data matrixkN；The sectionized matrix R column element of the 1st row the 1st It is worth for 1；
    According to the test statistics T of standardizationkNAnd K-AD homogeneity tests tables of critical values sets homogeneity horizontal, sets -5 ＜ TkN ≤ 0.326 homogeneous level is 1,0.326 ＜ TkN≤ 1.225 homogeneous levels are 2,1.225 ＜ TkN≤ 1.960 homogeneous water Put down as 3,1.960 ＜ TkN≤ 2.719 homogeneous levels are 4,2.719 ＜ TkN≤ 3.752 homogeneous levels are 5,3.752 ＜ TkN Horizontal≤100 homogeneities are 6.
  5. A kind of 5. data partition method based on homogeneity test according to claim 4, it is characterised in that：The step 6 It is middle to find the data element for being more than carry in sectionized matrix R, if it is possible to find, carry out homogeneous level calculation and walked again Rapid five, if can not find, carry out step 7；Detailed process is：Start from sectionized matrix R (1,1) individual element, seek Look for first be more than step 5 in carry data element position, if can search out and first be more than step 5 in carry Data element position be located at sectionized matrix R xth rows y row, then carry out homogeneous level calculation, x spans areY spans arePerform step 5；
    If the data element more than carry in step 5 can not be searched out, step 7 is performed.
  6. A kind of 6. data partition method based on homogeneity test according to claim 5, it is characterised in that：It is described from subregion Matrix R (1,1) individual element starts, and first position for being more than the data element of carry in step 5 is found, if can find Arrive and it is located at xth row y row, then carry out homogeneous level calculation, detailed process is：
    Determine to take amplitude to obtain (x, y) place data segment of radar return amplitude data matrix according to sectionized matrix R x rows y row, Regeneration block matrix R xth row y column datas element is carry+1 in step 5, and the data element for continually looking for sectionized matrix R is big All positions of carry in step 5, positional representation are (i1, j1),I1 represents position Place line number, j1 represent columns where position, determine to take amplitude to obtain radar return amplitude data matrix according to i1 rows j1 row (i1, j1) place data segment, calculate the homogeneity level of (a, b) place data segment and (i1, j1) place data segment, regeneration block The column data elements of row jth 1 of matrix R the i-th 1 for the homogeneity that calculates it is horizontal with step 5 carry and.
  7. A kind of 7. data partition method based on homogeneity test according to claim 6, it is characterised in that：The step 7 It is middle that new sectionized matrix tt is generated according to sectionized matrix R；Detailed process is：
    New sectionized matrix tt sizes are identical with taking amplitude to obtain radar return amplitude data matrix size, are m rows, n row；If A is not equal toB is not equal toA (m'-1)+1 in tt matrixes is then made to arrive am' rows, b (n'-1)+1 arrives the number of bn' row It is equal to the value of a row b column data elements in sectionized matrix R according to the value of element；
    If a is equal toB is not equal toA (m'-1)+1 in tt matrixes is then made to arrive m rows, b (n'-1)+1 to bn' row Data element value be equal to sectionized matrix R in a row b column data elements value；
    If a is not equal toB is equal toA (m'-1)+1 in tt matrixes is then made to arrive am' rows, b (n'-1)+1 to n row Data element value be equal to sectionized matrix R in a row b column data elements value；
    If a is equal toB is equal toA (m'-1)+1 in tt matrixes is then made to arrive m rows, b (n'-1)+1 arrives the number of n row It is equal to the value of a row b column data elements in sectionized matrix R according to the value of element.
CN201510990130.9A 2015-12-24 2015-12-24 A kind of data partition method based on homogeneity test Active CN105607049B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201510990130.9A CN105607049B (en) 2015-12-24 2015-12-24 A kind of data partition method based on homogeneity test

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201510990130.9A CN105607049B (en) 2015-12-24 2015-12-24 A kind of data partition method based on homogeneity test

Publications (2)

Publication Number Publication Date
CN105607049A CN105607049A (en) 2016-05-25
CN105607049B true CN105607049B (en) 2018-03-13

Family

ID=55987127

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201510990130.9A Active CN105607049B (en) 2015-12-24 2015-12-24 A kind of data partition method based on homogeneity test

Country Status (1)

Country Link
CN (1) CN105607049B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111073499B (en) * 2019-12-27 2021-06-15 哈尔滨工业大学 Preparation method of red flag mark for deep space detector

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103018727A (en) * 2011-09-27 2013-04-03 中国科学院电子学研究所 Sample-training-based non-stationary clutter suppression method of vehicle-mounted radar
CN104714215A (en) * 2015-03-19 2015-06-17 电子科技大学 Clutter intensity zoning based constant-false-alarm detection method

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103018727A (en) * 2011-09-27 2013-04-03 中国科学院电子学研究所 Sample-training-based non-stationary clutter suppression method of vehicle-mounted radar
CN104714215A (en) * 2015-03-19 2015-06-17 电子科技大学 Clutter intensity zoning based constant-false-alarm detection method

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
"Cognitive Detector for Complex Multi-source Environment: Preliminary Results Based on Background Perception for HFSWR";LI Yang等;《Radar Reference》;20120511;第0547-0552页 *
"Modified Anderson-Darling Test-Based Target Detector in Non-Homogenous Environments";Yang Li 等;《Sensors 2014》;20141231(第14期);正文第2.2节及图3 *
"基于区域生长的杂波分区方法";韩莹 等;《指挥信息系统与技术》;20150630;第6卷(第3期);第80-83页 *
"近程杂波的非均匀分段处理法";李萍;《雷达科学与技术》;20101231;第8卷(第1期);第60-64页 *

Also Published As

Publication number Publication date
CN105607049A (en) 2016-05-25

Similar Documents

Publication Publication Date Title
CN108062764A (en) A kind of object tracking methods of view-based access control model
CN104931934B (en) A kind of radar plot condensing method based on PAM cluster analyses
CN111722199B (en) Radar signal detection method based on convolutional neural network
CN108765470A (en) One kind being directed to the improved KCF track algorithms of target occlusion
CN106872958A (en) Radar target self-adapting detecting method based on linear fusion
CN105259541B (en) A kind of method of multistation radar anti-active cheating formula interference
CN107656265A (en) Particle filter fusion method for tracking short flight path before multi frame detection
CN108896973A (en) Scaling method, point the mark method for evaluating quality, storage medium of radar data
CN106226751A (en) Maneu-vering target detection based on DP TBD and tracking
CN109143195A (en) Radar target detection method based on full KL divergence
CN110208789B (en) Multi-frame combined target track identification method
CN108919223A (en) Broadband radar target adaptive structuring matched filtering method
CN113673312B (en) Deep learning-based radar signal intra-pulse modulation identification method
CN103777189A (en) Radar weak target detecting method based on information geometry multiple autoregressive model
CN111880158A (en) Radar target detection method and system based on convolutional neural network sequence classification
CN104537384B (en) A kind of SAR target discrimination methods of combination likelihood ratio judgement
CN108872961B (en) Radar weak target detection method based on low threshold
CN105259540B (en) A kind of optimization method of multistation radar anti-active cheating formula interference
CN105259546A (en) Dim sea surface radar target detection method based on AR spectrum fractal
Zhang et al. Weighted likelihood CFAR detection for Weibull background
CN105607049B (en) A kind of data partition method based on homogeneity test
CN105954739A (en) Knowledge-aided nonparametric constant false alarm detection method
CN105353371A (en) AR spectrum extended fractal-based sea surface radar target detection method
CN105184829A (en) Closely spatial object detection and high-precision centroid location method
CN104198998A (en) Clustering treatment based CFAR (Constant False Alarm Rate) detection method under non-uniform background

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant